Public Comments

Status: Original and derived artifacts are available for this document.

Download: pdf

Page 001

Who we are
• David Boundy
o Executive committee of PTAAARMIGAN, a nonprofit representing
Patent and Trademark Attorneys, Agents, and Applicants
o Former software engineer—invited to lecture at MIT, HewlettPackard’s alternate representative on an ANSI standards committee.
Software used by Boeing and other high-reliability users.
o Expertise in administrative law—in 2018, United States Court of
Appeals for Federal Circuit (Jay Plager’s court) flew me down to D.C.
for a day to coach on Chevron deference
o https://www.potomaclaw.com/professionals-David-Boundy
• Brad Forrest
o General Counsel and Chairman of the Board of Schwegman
Lundberg in Minneapolis, one of the ten largest filers of patent
applications
o Former Counsel and Manager of Intellectual Property Law for a
division of IBM (1984-1993)
o Currently Vice Chair of the AIPLA Legislation Committee (AIPLA is
the American Intellectual Property Law Association, a nonprofit, and
the largest organization of patent attorneys)
o https://www.slwip.com/people/bradley-forrest/
• Rich Baker
o President of NAPP, the National Association of Patent Practitioners, a
nonprofit, the third-largest organization of patent attorneys and
agents, generally representing small entities
o President of New England Intellectual Property in Massachusetts
o Former Director of Intellectual Property for Bottomline Technology,
3Com Corporate, and Schneider Automation
o Former Director of Patent Sales for HP
o Former Trustee for the Licensing Executives Society, a nonprofit
organization of technology transfer professionals
o Computer scientist with 20 years experience in developing and
leading software teams
o https://www.newenglandip.com/#h.5twh6ycmz7z
• All three appearing today pro bono

Issue 1: DOCX filing
The Patent Office proposes to displace a voluntary consensus standard for patent
filing–PDF–with a single company software product – Microsoft Word – that does not
observe any standard, let alone a voluntary consensus standard. This violates OMB
Circular A-119.
The Patent Office adopted a DOCX filing system in which (a) the Patent Office changes
the document that the applicant uploads, and (b) the Patent Office’s computers render
the file differently than the applicant’s computer. Both of these factors introduce errors.
The Patent Office’s ICR does not recognize the costs of proofreading or error
correction. Comment letters estimate burden at $200 million to $600 million (assuming
sufficient proofreading to elevate the reliability of DOCX filing to the same reliability as
PDF).
The process followed by the Patent Office was flawed

Page 003

Patent applications require
extraordinary precision
Except in extraordinarily narrow circumstances, a patent
application can’t be modified or corrected to add new material
after it is filed without sacrificing the original filing date. If the
Patent Office’s software makes an error, the applicant is stuck.
Almost every other legal document can be revised or amended
after initial filing—but for patent applications, that door is very
narrow.
About half of U.S. filings serve as a basis for foreign filings. Any
correction to a U.S. application also has to meet the requirements
of foreign law (and in many countries, the rules are even stricter).
Thus, the Patent Office’s upload software must be uniquely
reliable. The form in which a patent application is filed must be
reliable, and must be uniform so that a given file will appear
identical on every computer that displays it. The Patent Office’s
software must reflect high-reliability design principles.
An entire patent’s life (26 years) depend on correctness on day 1.
Other parts of a patent’s life are more error-tolerant, but not initial
filing.
Part 1 of this presentation: text-based PDF meets those
requirements. DOCX violates them.

Page 004

Upshot:
When a patent application is filed in DOCX, the Patent Office
cannot guarantee that what the Patent Office’s computers
see will be the same as what the patent applicant saw on
his/her computer before filing.
We have observed a number of differences.
Some are innocuous, formatting only—
• Different pagination
• Different indents, which can lead to very different layouts for
paragraphs and pages
• Different fonts
• A pain in litigation when cites to individual pages are required, and
different readers will see different displays, but not substantively
catastrophic

Some are catastrophic, potentially rendering a patent
application valueless:
• Multiple examples in the 60-day letter in ICR 202309-0651-002,
https://www.reginfo.gov/public/do/DownloadDocument?objectID=135
640900, at PDF pages 68-94

Page 005
Example 1: During 2019 NPRM comment period, commenters pointed to errors introduced by
the PTO's DOCX software:
DOCX is stable and governed by two international standards (ECMA-376 and ISO/IEC
29500).
This statement is sim ly false. There is no single DOCX standard to which Microsoft Word
and the other word rocessors are all com liant.
To give a sim le exam le, consider this math equation in a atent a
filed as a PDF-based PCT a lication using Libre Office:

lication that I recently

As an ex eriment I u loaded the DOCX file of this PCT a lication to EFS-Web as if I were
filing a domestic US atent a lication. The way the USPTO has designed EFS-Web, what
ha ens next is that the ractitioner sees this message in red letters:
The PD (s) have been generated from the docx file(s). Please review the
PD (s) for accuracy. By clicking the continue button, you agree to accept
any changes made by the conversion and that it will become the final
submission.
It is easy to see that this filing rocedure, as contem lated by USPTO, im oses an enormous
rofessional liability risk on the ractitioner. The ractitioner is obligated to roofread the
entire atent a lication, from to to bottom, for any corru tion introduced by the USPTO’s
rendering system.
Here is how the USPTO rendered this math equation:

The alert reader will notice that the USPTO inserted a s urious digit “1” into the math
equation. Had I overlooked this corru tion of the document by the USPTO, I might then have
clicked “continue”, at which oint it would have been USPTO’s osition that I had agreed to
acce t USPTO’s change of “0.2” to “10.2”. TYFNIL the accused infringer would be able to
ten years from now in litigation
seize u on this.
There are a dozen other laces in this atent a lication where USPTO corru ted math
equations; Equation 14 is merely the most striking so that is the one that I quoted here.
As a beta-tester of USPTO’s DOCX systems, I have used a retty sim le way of choosing
which of my atent a lications I am willing to subject to the risks of filing in DOCX. Basically
if there is any math equation or chemical formula, or anything other than very sim le
al hanumerical characters, I don’t take the risk. Every now and then, on a whim, I will
ex eriment with something like this “Equation 14” document, but I don’t risk any actual
substantive rights of a client by actually clicking “submit” in such a case.

In the Final Rule, the Patent Office denied having received notification of errors.
46956

Page 006

Federal Register / Vol. 85, No. 149 / Monday, August 3, 2020 / Rules and Regulations

providing ample time for applicants to
transition to this new process.
Comment 43: One commenter
questioned the statement made by the
USPTO Acting Deputy Director in the
September 2018 PPAC Fee Setting
Hearing that ‘‘fees for search and
examination are set below cost,’’ and

Therefore, they will continue to be
automatically removed, and a warning
will be provided.
To date, the Office has not received
notifications of any issues resulting
from the filing of applications in DOCX
format. If there is an instance in which
an error occurs, the Electronic Business

non-DOCX surcharge. Furthermore, the
fee is reduced by 50 percent for small
entities and 75 percent for applicants
that qualify as micro entities.
The Office recognizes the need for
freedom of choice to file in different
formats. Therefore, image-based PDFs
will continue to be accepted for

Perhaps this is literally true: as Carl's letter explains, after Carl detected the error to be introduced
by the PTO's DOCX software, Carl abandoned the attempt, and filed that patent application as a
reliable PDF. Thus, taken hyper-literally, there was no "issue resulting from the filing of an
application in DOCX format."
This is a symptom of a larger disease: in the Final Rule and Supporting Statements, the Patent
Office made dozens of representations to OMB that are only true if read hyper-literally.
For a Libre Office user (like Carl), (a) the error cannot be detected except by sharp-eyed characterby-character proofreading, and (b) this error cannot be overcome except by resorting to PDF. The
very same bits mean different things to Carl's Libre Office and to the Patent Office's DOCX engine.
The Patent Office has never responded to the public's concern that this is not a problem of
individual bugs. The problem is a design that can never be reliable, no matter how much
debugging. The Patent Office's foundation is built on sand that is designed to shift.
Response: The use of image-based
publication automatically route into the PDFs incurs many costs over the
lifetime of an application. There are
U.S. national stage application in PDF
large costs associated with the USPTO’s
form. However, the commenter stated
systems and personnel, from prethat this highlights the unwarranted
examination, examination, and
nature of the non-DOCX surcharge. If a
relatively moderate increase in price for publication, due to the need to apply
PCT national stage entry applications is OCR to convert image-based PDFs into
structured text that can be leveraged by
believed to be fiscally sustainable
downstream systems. The surcharge is
within the proposed fee structure, then
applied not only to account for these
the same should be true of regular noninefficiencies, but also to address rising
provisional application filings, and the
expenses.
imposition of a new non-DOCX filing
surcharge fee for one and not the other This answer is evasive. No one
is thus inconsistent.
suggested "image-based PDFs."
Response: Processing DOCX in
national stage applications presents
The suggestion was text-based
additional challenges and burdens on
PDFs, specifically PDF/A.
the Office and applicants that are not
encountered with a standard utility
Nowhere in 2019-2020 NPRM
application. Further investigation is
phase nor 2020 triennial
needed as to the possibility of
alternative means for obtaining text data ICR for 0651-0032 nor 2023
(i.e., via the International Bureau) that DOCX 0651-0089 does the
would not burden applicants. This is
PTO respond to the suggestion.
being considered for the future.
As patent applications have become
Comment 45: One commenter stated
increasingly complicated, the nonthat filing in DOCX is a wonderful idea
DOCX surcharge is an effective measure
in theory, but says that bugs have not
to recover the cost of converting PDFs
been worked out of the process. The
commenter writes that EFS-Web should to text. The text is essential for efficient
stop removing ‘‘text ornaments.’’ Until it examination and maintaining the
quality of patents issued. According to
stops doing this, and until a DOCX
surveys conducted by the USPTO, the
filing will reliably result in an identical
majority of applicants use wordPDF document, there should be no
processing software, such as Microsoft
penalty for filing PDF specifications or
Office and LibreOffice, to author
other application elements.
applications in DOCX format. These
Response: ‘‘Text ornaments,’’ or text
applicants will now be able to submit
decorations, may not be presented in a
applications in this same format to the
form that allows direct reproduction of
USPTO, therefore avoiding the new
readily legible copies. See CFR 1.52.

to provide a PDF version of the patent
application being filed, along with the
DOCX file. The PDF version would
serve as the controlling version in the
event of any discrepancy in the
USPTO’s rendering of the DOCX file.
Response: Many applications are
originally created in DOCX and
subsequently converted to PDF by
applicants prior to submission. An
advantage of submitting in DOCX format
directly is that submitted files from all
applicants are validated and converted
to PDF by USPTO systems in a
consistent manner. This eliminates the
unnecessary step for applicants to
generate and attach their own PDF
documents. The generated PDF is
available pre-submission to provide the
applicant an opportunity to review the
document before selecting the submit
button.
As a part of the DOCX intake process,
preliminary validation is performed on
DOCX documents at the time of upload.
The system immediately detects and
supplies the applicant with useful error
and warning messages, allowing for
adjustments to patent applications early
in the process. This saves time, reduces
potential costs, and prevents delays in
processing by minimizing notices of
missing parts or incomplete
applications from the Office.
Furthermore, the USPTO continuously
performs rigorous testing to ensure that
document integrity is preserved.
Comment 49: One commenter asked
whether the surcharge would be waived
if an applicant filed on paper because
the electronic filing system was not

Page 007

Example 2
•

Uploaded DOCX:

•

[0128]

•

Patent Office’s “validated” DOCX:

•

[0128]
i=1:h,

[h,w]=size(Iin);
Iout=zeros(size(Iin));
for i=1:h,
for j=1:w,
x=j;
y=i;
S=dZ/(f*H);
x1=x(:)-x0;
y1=y(:)-y0;
y2=y1./(1+y1*S);
x2=x1./(1+y1*S);
x2=x2+x0;
y2=y2+y0;
Iout(i,j)=bilinearInterpolate(Iin,x2,y2);
end;
end;

[h,w]=size(Iin);
for j=1:w,

Iout=zeros(size(Iin));
for
x=j;
y=i;
S=dZ/(f*H);
x1=x(:)-x0;
y1=y(:)-y0;
y2=y1./(1+y1*S);
x2=x1./(1+y1*S);
x2=x2+x0;
y2=y2+y0;
Iout(i,j)=bilinearInterpolate(Iin,x2,y2);
end;
end;

The top pseudocode listing (uploaded DOCX) is what was displayed on the applicant’s
computer. The bottom listing (Patent Office’s “validated” DOCX) is what showed up
when displayed on the Patent Office’s system. All the formatting showing how the code
worked is missing from the Patent Office’s DOCX.

Page 008

Example 3
Example 3 shows images of an equation copied directly from a display of the applicant.
The Patent Office’s displayed version shows squares for many of the parameters.

Example 4
Example 4 shows equation parameters converted to squares by the Patent Office’s
DOCX software:

Page 009

Example 5

We worked with a programmer at the Patent Office to try and correct some errors during
testing. Many hidden errors, like a hyperlink that may not even be visible in the display
of the DOCX file can result in an error message that prevents filing of the application.
Many hidden errors have been found and will continue to be found. After spending tens
of hours drafting an application that needs to be filed quickly, these errors, can both
prevent filing, or lie undetected potentially threatening the enforceability of the patent.

Page 010

Example 6
Both Adobe Sign and the Patent Office’s DOCX software changed this equation

to this:

In Example 6, as displayed on the applicant’s computer, the equation was displayed
perfectly. In filing the application, Adobe sign was used to generate an auxiliarly PDF to
use to correct the Patent Office’s “validated” version. Both Adobe Sign and the Patent
Office’s DOCX software generated the incorrect formula, replacing good symbols V and
D with boxed question marks. Similar boxes with question marks were sprinkled
through out the 50 plus pages of the application.
If the developer of the PDF standard can’t reliably convert a DOCX file, what chance
does the Patent Office have of getting it correct?

Page 011

Example 7 (April 2024)
Prevented filing until fixed

In Example 7, an error prevented filing of the application using a DOCX file. The cause
of the error was unviewable when performing normal editing. The error code is shown
above. We could not figure out how to fix it.
A Patent Office programmer we contacted recommended “press ALT &F9 to reveal the
automation field code. Highlight/select the field code then hold down CTRL+SHIFT+F9
to remove the field code.”
That did not work. We finally figured out how to fix it a couple days later by using a
feature in MS Word: click: FILE, INFO and then Check for Issues
The solution was found by a specialist in our sophisticated word processing
group who had never used that function before. What about small entities, like
individual inventors who use open source word processing that may not have a
super-specialist that can do that kind of diagnosis?
Can you imagine trying to file your taxes using DOCX files and having the IRS
version show that you under reported income by 10 fold and then not let you
correct it? Next stop, big fine and criminal prosecution. Some of the patents we
work on are worth millions of dollars, and the mistakes introduced by the DOCX
rendering can render them worthless.

Example 8
The next few pages illustrate the manglement that the Patent Office's DOCX
software introduces, and the burden the Patent Office creates when a person
seeks correction of Office error.
First, this is the way it's supposed to be. Notice how each claim begins with
a number, and a single claim can have multiple paragraphs, and there's a total
of 15 claims.

Page 012

Page 013

Page 014

Page 015

Page 016

Example 8 (continued) This is the manglement created by the Patent Office's DOCX software.
Note that many of the claims are separated into separate paragraphs, and then a number
given to each separate paragraph.
Then the Patent Office charges per-claim fees for each claim -- in this case, over $8000 in
additional fees.

Page 017

Page 018

Page 019

The Patent Office charges fees per claim. Because the Patent Office's software changed
15 claims to 31, the fees increased by over $8000.

Page 020
Example 8 (continued): The applicant requested a refund of the fee overcharge.
(Also notice how pixelated the text is.)

Page
021
Example 8 (continued): The Patent Office's first decison was to deny the refund. (This
is one
example of a general principle -- whenever fees are at issue, the Patent Office is extremely
reluctant to admit or fix its own errors.)

Page 022

Page 023
Example 8 (continued): the applicant filed a second, more-detailed request for refund.
I phoned the attorney; this request for refund took about 5 hours.

Page 024

Page 025

Page 026

Page 027

Example 8 (continued): Finally, on May 3 (last Thursday, over a year
after the original error), the Patent Office refunded the error

Lessons from these examples:
1. the DOCX software introduces errors. This is inevitable for three reasons. (a) the Patent
Office's software quality and testing functions are far below private-sector standards. (b) the
DOCX standard is very complex (3X to 9X more pages than the PDF standard, depending on how you
count) -- complexity invites error. (c) DOCX is not designed to be a reliable vehicle for archival
text or inter-computer interchange, and can never be. The standard makes no such guarantees.
Further, Microsoft makes undocumented changes to DOCX/Word all the time (one of the central
design goals of Microsoft Office 365 is to make it easier for Microsoft to distribute updates).
2. Catching these errors is very difficult for applicants, and the Patent Office compounds the
problem by making error correction very difficult. This error came to light because of the
$8000 charge--the error is very hard to catch at the moment a patent application is filed.
For most of these errors, the first time they will be visible to the applicant is when
the patent application publishes (at 18 months) or issues (typically 3-5 years). In the past,
proofreading of issued patents has been common, but no one proofreads the 18 month
publications. The Patent Office's Federal Register notices state that correction requests
will only be entertained within the first year.
3. Error correction requests will be necessary. The PTO books the extra petitions at zero.
4. Procedure matters. The ICR regulations require an agency to gather information and analyze
it fairly. When an agency skips procedure, bad decisions that should be caught early, while they're
easy to fix, propagate for years and can only be backed out at far greater costs. Those costs
should be borne by the agency that skipped procedure, not the public that exercised every
possible procedure to encourage the Patent Office to not make the error.
The solution is simple:
(a) The "of record" copy must be the file uploaded by the applicant, bit-for-bit, no changes. No
tampering in the name of "validation" or "uniform format." This is mandated by Federal Records
Act and Federal Rule of Evidence 1002.
(b) The "of record" copy must be in a format that is uniform, stable, reproducible across all
software and hardware implementations. PDF makes exactly that promise. The DOCX
standard, and Microsoft's web pages, promise the opposite.
That's what all courts and other agencies do. The Patent Office should not be an outlier.

Page 028

In 2015, the Patent Office had a
legitimate problem to solve
The Patent Office was an early adopter of electronic document systems. Their first
systems worked by scanning paper documents to TIFF (a bitmap photograph format.)
When the Patent Office began to accept electronic uploads of patent applications (2004
or so), the Patent Office gave instructions that resulted in a reliable text-based PDF
generated on the users computer, for upload.
https://www.uspto.gov/sites/default/files/ebc/portal/efs/pdf-creation.pdf
20 years ago, the Patent Office made a fateful choice: upload full-text PDFs, but flatten
them to a bitmap, and then try to OCR. In other words, applicants upload PDFs that
have the text in perfect form. Instead of maintaining that PDF in perfect full-text form,
the Patent Office programmed its computers to flatten the text-rich PDF into bitmaps
(photographs). Then the Patent Office has to OCR to reconstruct the characters that
flattening had discarded.
Here’s a patent application I filed a year ago, at 1600% magnification. On the left is the
text-based PDF as I uploaded it. On the right is the same patent application after filing,
as downloaded from the Patent Office’s web site.

As you can see in the left, the characters in a text-based PDF are perfect. But in the
right, they’ve been pixilated by the Patent Office’s computers.
Good software design would say “accept text-based PDF as the input, and don’t change
it.” That’s what every court and agency does (including the Eastern District of Virginia,
the Court of Appeals for the Federal Circuit, the Supreme Court, and the Patent Office’s
own Patent Trial and Appeal Board).
•

We’re as sure as we can be (as outsiders) that scraping text from a text-based
PDF is no more complex than from a DOCX, and probably easier.

•

We believe there’s no advantage in flattening to bitmap: a text-based PDF is
typically 1/6 to 1/10 the size of the bitmap PDF.

•

By law, patent applications must be one column, text only (with tables, equations,
and chemical formulae). As far as we know, patent applications present no
difficulties in text scraping (and despite several queries, the Patent Office has
provided no documentary evidence to the contrary).

Page 029

To formulate a solution to the OCR problem, the PTO commissioned a study. The study concluded:
-- In PDF, "The appearance of everything that each page contains is completely specified."
-- PDF-to-text "takes the right approach towards achieving the objectives"
-- a commercially-available PDF-to-text product largely solves the problem

Text2PTO Proof of Concept White Paper
Version 1.0

24 March 2015

American Environmental and Engineering Consultants (AEEC, LLC)
1925 Ballenger Ave, Suite 450,
Alexandria, VA 22314
http://www.americanconsultants.com
Phone: (703) 317-0800

Page 030
Text2PTO Proof of Concept White Paper
Version 1.0
March 24, 2015

1 Introduction
This white paper presents details about a Text2PTO prototype for extracting text, layout, and
formatting information from PDF files that have text behind them. A proof-of-concept (POC)
was conducted to design a solution that could accept incoming PDF files and extract text content
along with formatting and layout information. This document is presented by AEEC’s
Application Architecture Software Engineering Team (AASET) to the United States Patent and
Trademark Office (USPTO).

1.1 BACKGROUND
USPTO currently accepts patent applications through its Electronic Filing System (EFS). EFS
accepts PDF documents during application submission. Applicants can use various tools to
create PDFs for EFS submission. More than 45% of these submitted PDFs have text behind
them. Due to the differences in the COTS or open source tools used by the applicant to generate
the PDFs, the format and structure of the PDFs differ. Currently, USPTO relies on OCR to
extract text from TIFF representations of these submitted PDFs. The OCR-extracted text and
layout information is used to generate XML4IP documents.

1

Page 031
Text2PTO Proof of Concept White Paper
Version 1.0
March 24, 2015

3 Solution Analysis
3.1 PDF BRIEF INTRODUCTION
A PDF file encapsulates a complete description of a fixed-layout flat document, including the
text, fonts, graphics, and other information needed to display it. The appearance of everything
that each page contains is completely specified. The structure of a PDF file does not match the
structure of the PDF document it describes. PDF documents are display-oriented and the
specification is not built to address text extraction concerns. The underlying structure of a PDF is
multi-layered and very complex. Elaborating on the intricacies of extracting text from a PDF is
outside the scope of this document. However, additional details can be found by referring to the
following articles:
http://partners.adobe.com/public/developer/tips/topic_tip31.html
http://www.planetpdf.com/developer/article.asp?
ContentID=navigating_the_internal_struct&page=0

3.2 SELECTING A TOOL
As discussed in the previous section, text / format extraction from a PDF is a complex task.
There are several tools in the market that have sought to solve this problem with varying degrees
of success. Most of these tools do a good job extracting text from PDFs, but they lack reliability
and consistency in extracting the document’s format, layout, and structures. Developing a tool
from scratch to extract text, format, and layout information from PDF documents would be a
very complex task, and may not produce an acceptable output. Therefore, this prototype effort
focuses instead on identifying a tool that could get as close as possible to our objective.

3.3 SOLUTION OVERVIEW
We followed a 3-step process to identify and develop a solution:
1. Identify format, layout, and structural elements of PDF documents to be extracted.
2. Conduct market research on tools that can extract the identified elements from PDFs.
3. Develop a Java-based solution around selected tools in order to:
a. Integrate the tool in the USPTO environment
b. Support batch processing
c. Output an XML document
d. Address shortcomings of the tool

3

Page 032
Text2PTO Proof of Concept White Paper
Version 1.0
March 24, 2015

http://en.wikipedia.org/wiki/Private_Use_Areas
A custom character set and does not provide mapping between the character codes and
Unicodes
http://blogs.adobe.com/insidepdf/2008/07/text_content_in_pdf_files.html
At this point, it is not clear if there is a resolution for the two scenarios above. The PDFxStream
team has been notified about this and they are looking into it.


Math and Chemical Formulae
Though the prototype attempted to detect and extract math and chemical formulae, the solution is
not complete. Please refer to section 5.3.3 for details.
List Processing
List detection is not built into PDFxStream and there are no indications it will be supported in
the future. This needs to be custom built into the application.

5.9 CONCLUSION
Although the presented solution has some limitations, it takes the right approach towards
achieving the objectives. Limitations may be addressed in the long term by extending the Java
application and using future releases of PDFxStream to customize / enhance the product to suit
USPTO’s needs.
Recommendations:

Obtain support from PDFxStream to fix issues with Vector Images and Special
Characters

Implement additional logic to handle Chemical / Math formula

Implement custom logic to extract List and Superscript / Subscript information

13

Page 033

The PDF standard guarantees the properties that matter:
• PDF is a voluntary consensus standard, designed for
“exchange [of] electronic documents … reliably, independent
of the environment in which they were created or the
environment in which they are viewed or printed”
• A text-based PDF has perfect characters
• A text-based PDF file, regardless of the software that
generated it or that views it, is guaranteed to appear
identically on all computers
• PDF is standardized as ISO 32000, a voluntary consensus
standard
• Many vendors, some of whose products are free
• If you want to test that PDF documents are perfect, open
almost any ICR letter in Adobe Acrobat.
o Do View > Zoom > 1600%. You will see no pixels. Every
character exists in a perfect form that captures every
characteristic.
o Go back to Zoom 100%, then do View > Zoom > Reflow.
You will see that the document has characters and most
of the paragraphs perfectly encoded. (Page breaks are
sometimes handled correctly, sometimes not. There’s a
switch setting that the Patent Office could recommend to
take care of that.)
• Because PDF is so universally used, there are commercial
products for “scraping” text out of text-based PDF
documents.

The very first paragraph of the ISO standard (ISO 32000-2) guarantees
Page 034
reliability and "independence" from hardware and software variability
ISO 32000-2:2020(E)

Introduction
0.1

PDF

The standard guarantees perfect character-by-character fidelity—the file
encodes the difference between the Latin letter X, the Greek letter Х, the
Cyrillic letter Χ, and the multiply symbol ×
PDF is a voluntary consensus standard (19 participating members in
standard committee). Dozens of implementations, many free

PDF enables users to exchange and view electronic documents easily and reliably, independent of the
environment in which they were created or the environment in which they are viewed or printed.
At the core of PDF is an advanced imaging model derived from the PostScript®1 page description
language. This PDF Imaging Model enables the description of text and graphics in a device-independent
and resolution-independent manner at a complete, precise and professional level. Unlike PostScript,
which is a programming language, PDF is based on a structured binary file format that is optimised for
high performance in interactive viewing.
PDF includes objects such as annotations and hypertext links that are not part of the page content itself
but are useful for interactive viewing and document interchange. PDF also includes data structures
such as tagged PDF, XMP and an associated files mechanism, that are useful for document management
and content reuse.
PDF files can be created natively in PDF form, converted from other electronic formats. Since PDF
supports a wide range of image and compression technologies, it is a suitable format for documents
digitised from paper, microform, or other hard copy formats. Businesses, governments, libraries,
archives and other institutions and individuals around the world use PDF to represent considerable
bodies of important information. Since its introduction in 1993, aided by the explosive growth of the
Internet, PDF has become widely used for the electronic exchange of documents.
There are several specific applications of PDF that have evolved in which limiting the use of some
features of PDF while requiring the use of others, enhances the usefulness of PDF. The following
International Standards address specialised uses of PDF:
•

PDF/X (ISO 15930) is the industry standard for the intermediate representation of printed
material in electronic prepress systems for conventional printing applications.

•

PDF/A (ISO 19005) is the industry standard for the archiving of digital documents.

•

PDF/UA (ISO 14289) is the industry standard for accessible PDF documents and processors.

•

PDF/E (ISO 24517) provides a mechanism for representing engineering documents and
exchanging engineering data.

PDF works great for interchange of text documents.
Every court and agency (U.S. and international) that we know of accepts its papers in PDF.
Every court and agency that we talked to issues its decisions in PDF, and gives those PDFs
to legal publishers (Lexis, Westlaw, Bloomberg, etc.) in PDF form. The publishers scrape
the text out of the PDFs. The entire legal system relies on the reliability of PDF.
Patent applications by law must be text only (no pictures). The only complexity allowed is
tables. Court decisions are more complex and less uniform. PDF has every capability
any patent office could ask for patent applications.
viii

© ISO 2020 – All rights reserved

Sold by the PDF Association to David Boudy Potomac Law Group 10513 | May 5, 2024 |
Single user only, copying and networking prohibited.

PDF/A Family, PDF for Long-term Preservation

1 of 17

https://www.loc.gov/preservation/digital/formats/fdd/fdd000318.shtml

Page 035

Digital Preservation Home | Digital Formats Home

Sustainability of Digital Formats:
Planning for Library of Congress
Collections

Search this site
Go

Introduction | Sustainability Factors | Content Categories | Format Descriptions | Contact
Format Description Categories >> Browse Alphabetical List

PDF/A Family, PDF for Long-term Preservation
>> Back
Table of Contents
Identification and description
Local use
Sustainability factors
Quality and functionality factors
File type signifiers
Notes
Format specifications
Useful references
Format Description Properties
ID: fdd000318
Short name: PDF/A_family
Content categories: text
Format Category: family, file-format, encoding
Other facets: unitary, binary, structured, symbolic
Last significant FDD update: 2020-12-31
Draft status: Full

The Library of Congress has a discussion of multiple
documents and their utility for archival document
storage.
The LoC's discussion of PDF (specifically the PDF/A
variant states the properties we would all want:
• long-term preservation
• voluntary consensus standard with wide range of
participants in standard committee

Identification and description
Full name
Description

ISO 19005. Document management - Electronic document file
format for long-term preservation
PDF/A is a family of ISO standards for constrained forms of PDF
(see PDF_family) intended to be suitable for long-term preservation
of page-oriented documents for which PDF is already being used in
practice. The PDF/A standards are developed and maintained by a
working group with representatives from government, industry, and
academia and active support from Adobe Systems Incorporated.
The working group is WG 5 of Technical Committee ISO/TC 171,

5/5/2024, 6:42 AM

Page 036

The DOCX standard does not make the relevant promises,
and therefore does not guarantee identical document display
on different systems:
• DOCX has many “implementation defined” or “unspecified”
behaviors that can change the way a document displays by
merely moving it from one computer to another:
o Different versions of Word—for Latin characters vs.
Hebrew characters vs. Asian characters, Word for
Windows vs. Word for Mac, Word 2023 vs. Word 2013—
render/display the very same PDF file differently
o The standard allows extensions and plugins—no
guarantees possible
o Office 365 exacerbates the problem: updates without
notice, on Microsoft’s schedule
• Zero vendors conform to the DOCX standard: even
Microsoft concedes that standard conformance is not
considered, and it allows Word to “vary from or extend the
specification”
• Patent Office ignores OMB Circular A-119, which directs
agencies to adopt voluntary consensus standards such as
PDF.

This is from the DOCX standard, ISO/IEC 29500-1 (2012), the version in effect at the time of
rulemaking.
Page
037
lesson: the DOCX standard is designed for editability, not uniformity, reproducibility, or archival use
ISO/IEC 29500-1:2012(E)

The DOCX standard expressly does NOT guarantee uniformity or reproducibility across all
implementations (MS Word, Libre Office, Google Docs, etc.) only "within any given implementation."

4.

Terms and Definitions

DOCX is not a voluntary consensus standard—it’s a Microsoft captive standard. As far as we know,
no non-Microsoft products use the DOCX standard as their native language or implement DOCX reliably.
For the purposes of this document, the following terms and definitions apply. Other terms are defined where
they appear in italic typeface, on the left side of a syntax rule, or within subclauses of language-specific
grammars (§17.16 and §18.17). Terms explicitly defined in this Part of ISO/IEC 29500 are not to be presumed to
refer implicitly to similar terms defined elsewhere. [Note: This Part uses OPC-related terms, which are defined in
ISO/IEC 29500-2. end note]

application — A consumer or producer.
behavior — External appearance or action.
behavior, implementation-defined — Unspecified behavior where each implementation is expected to
document that behavior, which would thereby promote predictability and reproducibility within any given
implementation. (This term is sometimes called “application-defined behavior”.)
behavior, locale-specific — Behavior that depends on local conventions of nationality, culture, and language.
behavior, unspecified —Behavior where ISO/IEC 29500 makes no recommendations. (This term is sometimes
called “application-dependent behavior”.) [Note: To add an extension, an implementer must use the
extensibility mechanisms described by ISO/IEC 29500 rather than trying to do so by giving meaning to otherwise
unspecified behavior. end note]
byte — A sequence of 8 bits treated as a unit.
comment — A note that an author or reviewer attaches to content in a document. Although a consumer might
chose to display comments, they are not considered part of the body of the document. A comment might
include the text of the note, the comment author's name and initials, and date of creation, among other things.

In the body of the standard, the words "implementation-defined" or "unspecified" or synonyms
are used literally hundreds of times.

©ISO/IEC 2012 – All rights reserved

11

Page 038

The Library of Congress page relating to DOCX explains:
•

Interoperability with non-Microsoft word processors is not reliable—if one opens
a DOCX file in Libre Office, and stores the result as a DOCX, “a round-trip to an
identical document should never be expected.”

[MS-OI29500]: Office Implementation Information for ISO/IEC 29500...

1 of 6

https://learn.microsoft.com/en-us/openspecs/office_standards/ms-oi295...

039
No one implements the DOCX standard—not even Microsoft. Microsoft explains that WordPage
departs
from
the standard. This page is updated every two or three months--Microsoft does not even promise to be
consistent in the way Word deviates from the ISO standard.

[MS-OI29500]: Office Implementation
Information for ISO/IEC 29500
Standards Support
Article • 03/08/2023

Specifies information about how Microsoft Office implements ISO/IEC-29500 Office
Open XML File Formats [ISO/IEC-29500:2012], including areas where additional detail
may be of benefit to other implementers and areas where the product is known to vary
from or extend the specification.
This page and associated content may be updated frequently. We recommend you
subscribe to the RSS feed

to receive update notifications.

Published Version

The list of departures from the standard is
936 pages.

Expand table
Date

Protocol Revision

Revision Class

Downloads

4/16/2024

22.0

Major

PDF

| DOCX

Click here to download a zip file of all PDF files for Word, Excel, and PowerPoint
Standards Support.

Previous Versions
Expand table
Date

Protocol Revision

Revision Class

Downloads

2/20/2024

21.3

None

PDF

| DOCX

11/13/2023

21.3

Minor

PDF

| DOCX

8/15/2023

21.2

Minor

PDF

| DOCX

3/9/2023

21.1

Minor

PDF

| DOCX

5/5/2024, 7:53 AM

Page 040
Seventy Three P
Pa
Pr
atent P
ractitioners
United States Patent and Trademark Office
re Setting and Adjusting Patent Fees During Fiscal Year 2020

Page 15 of 15
September 27, 2019

One of the 2019 NPRM comment letters https://www.uspto.gov/sites/default/files/documents/
Comment_Seventy_Three_Patent_Practitioners_092719.pdf gave the PTO a tutorial on the problem,
and listed several important examples of "implementation defined" or "unspecified" behaviors.
The same DOCX file can be rendered differently depending on the word
processor, fonts installed, which font vendor supplied the font, whether the word processor
chooses a vector form or bitmap form for the font, and add-ins for the word processor (especially
for equations, pictures and drawings, and chemical formulae). Because a single word
processor’s rendering engine is used to display on screen, print on paper, and print-as-PDF, the
applicant has a trustworthy what-you-see-is-what-you-get. But if that same DOCX is transmitted
to the PTO, for the PTO to render using unidentified software and unidentified environment, the
results will be different.
2.

The factual representations in the NPRM relating to two standards
and portability of DOCX are incorrect

The PTO does not tell us what rendering engine will be used within the PTO. Will it be
MS Word or some other rendering engine? The “viewer” software in Firefox, Internet Explorer,
or Chrome, or the viewer in Google gmail, Word 2003, 2013, or 2016? For Mac or Windows?
All behave differently. With DOCX, no amount of care by a practitioner can possibly ensure
how the document will be interpreted by the PTO’s rendering or conversion software. It is
unreasonable to expect the filer to undertake to proofread, carefully, word-by-word, any
specimen of the conversion result the PTO may provide just before the filing is finally submitted.
Indeed, the very requirement to proofread the rendering (noted below in red text) is an admission
by the PTO that it recognizes that DOCX is a shaky foundation for a legal document filing
system (there’s no such warning in today’s system). For lengthy, complex specifications, the 60minute timeout in EFS would preclude effective review. In the case of a timeout, the subsequent
re-submission would still require the filer to review the entire conversion result from the
beginning.
Standards ECMA-376 and ISO/IEC 29500 themselves disclaim the kind of
interoperability that the PTO assumes. Some example sentences:
• “a software application should be accompanied by documentation that describes what
subset of ECMA-376 it supports” ECMA-376 expressly states that there is no common
set of features that are required to be implemented; all the standard guarantees is that if
certain features are implemented, they will behave in a certain manner. A standard useful
for an electronic filing system cannot rely on features that are optional in some
implementations and unimplemented in others.
• “The application need not implement operations on all XML elements defined in ECMA376.” Some implementations of DOCX are permitted to have features that will cause
errors in others.
• “A batch tool that reads a word-processing document and reverses the order of text
characters in every paragraph with ‘Title’ style before saving it can be conforming even
though ECMA-376 does not recommend this behavior. [A conforming word processor
may] transform the title ‘Office Open XML’ into ‘LMX nepO eciffO’. Its documentation

Page 041
Seventy Three P
Pa
Pr
atent P
ractitioners
United States Patent and Trademark Office
re Setting and Adjusting Patent Fees During Fiscal Year 2020

•

•

•

•

Page 16 of 16
September 27, 2019

should declare its effect on such paragraphs.” The ECMA standard expressly allows for
entirely different renderings, so long as it’s documented.
“These application descriptions should not be taken as limiting the ability of an
application provider to create innovative applications. They are intended as a mechanism
for labelling applications rather than for restricting their capabilities.” A standard useful
for an electronic filing system can’t rely on features that are optional in some
implementations and unimplemented in others.
“[Note: A possible application description would be a ‘standard’ application description
for a wordprocessing application. This could be created by taking the intersection of the
features available in common wordprocessing applications such as Word 2000,
OpenOffice 2, WordPerfect, and iWork Pages. … end note]” ECMA-376 expressly
states that there is no common set of features that are required to be implemented; all
ECMA-376 guarantees is that if an implementer wants to implement a given feature,
there is a format in which to implement it. There are very few behavioral guarantees.
ECMA-376 leaves a number of features “implementation defined,” including whether
and how to save any element that is under the control of a plug-in, how dates are
rendered, how embedded pictures are rendered, whether numerical values are rendered
with a “.” or a “,” as a decimal point, how fonts are chosen in rendering, line number
spacing, and other characteristics. Documents copied from one DOCX program to
another have no guarantee of being rendered consistently.
A Microsoft blog17 writes “One of the great things about ISO/IEC 29500 is its
extensibility mechanisms - implementers can extend the file format while remaining
100% compliant with the standard.” That statement is the admission—there is no
uniform interoperability standard. ISO/IEC 29500 is a baseline, minimum functionality
standard, not an interoperability standard that guarantees bilateral consistency between
any two implementations. That may be a good feature for software developers, but it’s
catastrophic for the use that the PTO contemplates. That bilateral interoperability is the
whole point of the PDF standard.

As technically-trained lawyers, we don’t understand how any person could read ECMA-376 and
not have immediately noticed the glaring deficiencies as a “standard” for legal documents.
One of the signatories of this letter was among the very first of the beta-testers of PTO’s
system for DOCX filings. As implemented by the PTO, the practitioner would upload a DOCX
file, and PTO would render the DOCX file in a human-readable PDF image format. As part of
the e-filing process, the practitioner was expected to proofread the rendered image as provided
by the PTO’s e-filing system. The notion was that the practitioner would be obliged to catch any
instances of PTO’s system rendering the DOCX file differently from the way the practitioner’s
word processor had rendered that same DOCX file. If, for example, some math equation or
chemical formula had gotten corrupted in PTO’s system, the practitioner would expected to catch
this prior to clicking “submit.”

17

https://blogs.msdn.microsoft.com/chrisrae/2010/10/06/where-is-the-documentation-for-officesdocxxlsxpptx-formats-part-2-office-2010/

Page 042

The Patent Office’s response to comments
in the August 2020 Final Rule:

[not a single word, not even a paraphrase of the comment]

Page 043

DOCX screws up law firms’ workflow
Remember Example 8, the $8000 fee guy.
This is the feature of DOCX that irks him most.

•

For decades, the way it was done was:
o Often, the inventor writes the first draft of the application. Attorneys can’t control
the inventor’s use of Word features. In the conversion to PDF, the word
processor takes care of it, and it just works.
o Attorney writes and finishes the patent application in the attorney’s choice of
word processor, prints the last version as text-based PDF—because this is all
done on one computer, it is reliable.
o Hands off the text-based PDF to a legal assistant
o Assistant uploads the text-based PDF to the Patent Office, hits “submit”
o Because all word processing is on one computer, and all motion from one
computer to another is in PDF and PDF is guaranteed to be uniform, it all works

•

Under DOCX filing:
o Many advanced features used by sophisticated Word users (inventors or
attorney) are rejected by the Patent Office’s software. Attorney must review the
inventor’s draft for advanced Word features, and remove them (without
introducing error).
o If the attorney is using a word processor other than Microsoft Word, export from
that word processor’s native form to DOCX. But no software vendor makes it
easy to assist a competitor—the “export to DOCX” features of Word Perfect,
Libre Office, Google Docs, Apple Pages, etc. are not robust. These nonMicrosoft processors generate a DOCX, but for even simple patent applications,
the result is often substantially changed. So this requires a significant proofread.
(The Patent Office’s claim that these other processors “support” DOCX is true in
a hyper-literal sense—but in any practical sense, it’s not true.)
o Hands off the DOCX to a legal assistant
o Assistant uploads the DOCX to the Patent Office
o Assistant downloads the tampered and “check” documents from the Patent Office
o Hands them back to the attorney for review
o Attorney has to do a detailed review—there is no predicting where the errors may
be introduced.
• This review is not necessary in the existing PDF “reporting and
recordkeeping” workflow. It is very time consuming, and despite the
issue raised in three prior ICRs, the PTO continues to book it at zero.
• This review compares a DOCX on the applicant’s computer, so it tells
nothing about how the document will appear to the PTO’s computer.
o Then the attorney hands back the review to the assistant, who completes the
upload
o Three handoffs instead of one. No possibility of reliable review, because the
DOCX document can appear differently on different computers.

Page 044

Sources of burden
•

Proofreading
o After the Patent Office’s DOCX system tampers with the applicant’s uploaded
DOCX (potentially introducing error, as in example 8), the DOCX system gives
the applicant a “review copy” for review, with this instruction box to accept it as
the authoritative version:

o

o
o
o

o

The Patent Office has never acknowledged any burden for this additional review,
let alone requested clearance. The Office neither acknowledges nor requests it
now.
The Patent Office’s “review” happens on the applicant’s computer. That
means that the “review” cannot reliably evaluate how the Patent Office’s
computers will receive/render the very same file.
The Patent Office’s system has a timeout of about 20 minutes. For applications
of any length or complexity, the timeout precludes effective review.
By requesting this additional review, the Patent Office’s acknowledges that its
DOCX requirement imposes substantial burden without any practical utility.
“With DOCX, no amount of care by a practitioner can possibly ensure how the
document will be interpreted by the Patent Office’s rendering or conversion
software.” Seventy-Three letter, page 15.
Microsoft changes Word with no advance notice, in ways that will break the
Patent Office’s software, and that will make review ineffective (it will be on the
applicant’s computer, which will have been updated, while the Patent Office’s
software is still stale). The Patent Office has never addressed those comments.

•

Papers for error correction
o As you see in Example 8, that can be substantial.
o The Patent Office never stores the application as uploaded by the applicant. The
best copy the Patent Office stores has already been document tampered (in the
name of “validation” and “uniform format.”) Proving that an error was introduced
by the Patent Office will be extremely difficult.

•

Malpractice risk: How does OMB value the burden of malpractice risk?
o Defending suit, increased malpractice premiums.
o Raised in public comment letters; no response from Patent Office.

•

Comment letters give build-ups of burden ranging $200 million per year to $600
million per year. There’s a list of those letters toward the end of this document. The
Patent Office has never responded to those comments or disputed their showings,
or offered any alternative estimate with any rationale or supporting record.

Page 045

Procedural shortcutting
•

During NPRM phase (NPRM in July 2019, final rule August 2020), the Patent Office
claimed to have a clearance, but had made no filing.
o During NPRM phase, Patent Office never made the “yearlong study” public—not
even mentioned in the NPRM, cited for the first time in final rule. Not published.
Kept secret, until we got it by FOIA in 2023
o The Patent Office entirely ignored many comments, misparaphrased others to
avoid fair response, or otherwise answered with only a non sequitur.
o Not a word of mention in the Regulatory Impact Analysis.

•

March 2021: original effective date. (Delayed nearly three years—proving that
DOCX is not “ necessary for the proper performance of the functions of the agency”)

•

September 2020 to May 2021, triennial renewal of 0651-0032, ICR 202011-0651006: public comment letters noted that the Patent Office had no clearance for the
DOCX rule. In its May 2021 revised Supporting Statement, the PTO acknowledged
that burden would arise, and admitted that it had no current clearance:

•

In September to December 2023, 0651-0089, ICR 202309-0651-002 estimated only
the short-term burden of “adjusting existing ways to comply” during a transition
period. The PTO evades the major burden in this information collection, “completing
and reviewing” burden for proofreading and filing for correction of software errors.
o The PTO estimated hourly burden at $50 million per year. The Patent Office
provided no rationale or record. $50 million is credible estimate but only for the
interim “adjusting” burden, without considering proofreading and error correction.
o In our 30-day comment letter (Oct 27, 2023) we go though the Supporting
Statement line by line, pointing out cases in which the Supporting Statement:
• evades fair response via misparaphrase of the public comment
• makes a statement that reflects sampling error: for example, the PTO’s claim
to have seen no errors at 18 month publication is meaningless if the rule
hasn’ tbeen in effect for 18 months.
• The Patent Office discards the applicant’s filing—the Patent Office cannot
possibly test for its own errors.
• relies on non sequitur, or “facts” that have no supporting evidence
• statements that are simply false, and must have been known to be false

•

January 17, 2024: the mandatory surcharge went into effect

•

In this triennial renewal, ICR 202304-0651-001, the PTO again omits any discussion
of the DOCX issue. It still isn’t cleared.

Page 046

In the NPRM, the PTO claimed to have a "previously approved" clearance

37431

Federal Register / Vol. 84, No. 147 / Wednesday, July 31, 2019 / Proposed Rules

jbell on DSK3GLQ082PROD with PROPOSALS2

M. Unfunded Mandates Reform Act of
1995: The proposed changes set forth in
this rulemaking do not involve a Federal
intergovernmental mandate that will
result in the expenditure by State, local,
and tribal governments, in the aggregate,
of 100 million dollars (as adjusted) or
more in any one year, or a Federal
private sector mandate that will result
in the expenditure by the private sector
of 100 million dollars (as adjusted) or
more in any one year, and will not
significantly or uniquely affect small
governments. Therefore, no actions are
necessary under the provisions of the
Unfunded Mandates Reform Act of
1995. See 2 U.S.C. 1501 et seq.
N. National Environmental Policy
Act: This rulemaking will not have any
effect on the quality of the environment
and is thus categorically excluded from
review under the National
Environmental Policy Act of 1969. See
42 U.S.C. 4321 et seq.
O. National Technology Transfer and
Advancement Act: The requirements of
section 12(d) of the National
Technology Transfer and Advancement
Act of 1995 (15 U.S.C. 272 note) are not
applicable because this rulemaking does
not contain provisions which involve
the use of technical standards.
P. Paperwork Reduction Act: The
Paperwork Reduction Act of 1995 (44
U.S.C. 3501 et seq.) requires that the
Office consider the impact of paperwork
and other information collection
burdens imposed on the public. This
proposed rule involves information
collection requirements which are
subject to review by the OMB under the
Paperwork Reduction Act of 1995 (44
U.S.C. 3501–3549). The collection of
information involved in this proposed
rule has been reviewed and previously
approved by OMB under control
numbers 0651–0012, 0651–0016, 0651–
0020, 0651–0021, 0651–0031, 0651–
0032, 0651–0033, 0651–0059, 0651–
0063, 0651–0064, 0651–0069, and 0651–
0075.
Notwithstanding any other provision
of law, no person is required to respond
to nor shall any person be subject to a
penalty for failure to comply with a
collection of information subject to the
requirements of the Paperwork
Reduction Act unless that collection of
information displays a currently valid
OMB control number.
List of Subjects
37 CFR Part 1
Administrative practice and
procedure, Biologics, Courts, Freedom
of information, Inventions and patents,
Reporting and recordkeeping
requirements, Small businesses.

VerDate Sep<11>2014

20:55 Jul 30, 2019

Jkt 247001

37 CFR Part 11
Administrative practice and
procedure, Inventions and patents,
Lawyers, Reporting and recordkeeping
requirements.
37 CFR Part 41
Administrative practice and
procedure, Inventions and patents,
Lawyers, Reporting and recordkeeping
requirements.
37 CFR Part 42
Administrative practice and
procedure, Inventions and patents,
Lawyers.
For the reasons set forth in the
preamble, 37 CFR parts 1, 11, 41, and
42 are proposed to be amended as
follows:
PART 1—RULES OF PRACTICE IN
PATENT CASES
■ 1. The authority citation for 37 CFR

part 1 continues to read as follows:

(c) Basic fee for filing each application
for an original plant patent:

TABLE 3 TO PARAGRAPH (c)
By a micro entity (§ 1.29) ....................
By a small entity (§ 1.27(a)) ................
By other than a small or micro entity

$55.00
$110.00
$220.00

(d) Basic fee for filing each
provisional application:

TABLE 4 TO PARAGRAPH (d)
By a micro entity (§ 1.29) ....................
By a small entity (§ 1.27(a)) ................
By other than a small or micro entity

$75.00
$150.00
$300.00

(e) Basic fee for filing each application
for the reissue of a patent:

TABLE 5 TO PARAGRAPH (e)
By a micro entity (§ 1.29) ....................
By a small entity (§ 1.27(a)) ................
By other than a small or micro entity

$80.00
$160.00
$320.00

(f) * * *

Authority: 35 U.S.C. 2(b)(2), unless
otherwise noted.

TABLE 6 TO PARAGRAPH (f)

■ 2. Section 1.16 is amended by:
■ a. Revising paragraphs (a) through (e);
■ b. Adding table headings in

paragraphs (f) and (g);

*

*

*

*

*

(g) * * *

■ c. Revising paragraph (h);
■ d. Adding a table heading in

TABLE 7 TO PARAGRAPH (g)

paragraph (i);
■ e. Revising paragraphs (j) and (k);
■ f. Adding a table heading in paragraph

(l);
■ g. Revising paragraphs (m) through (s);
■ h. Adding a table heading in
paragraph (t); and
■ i. Adding paragraph (u).
The revisions and additions read as
follows:

*

*

*

*

*

(h) In addition to the basic filing fee
in an application, other than a
provisional application, for filing or
later presentation at any other time of
each claim in independent form in
excess of 3:

§ 1.16 National application filing, search,
and examination fees.

(a) Basic fee for filing each application
under 35 U.S.C. 111 for an original
patent, except design, plant, or
provisional applications:

TABLE 1 TO PARAGRAPH (a)
By a micro entity (§ 1.29) ....................
By a small entity (§ 1.27(a)) ................
By a small entity (§ 1.27(a)) if the application is submitted in compliance
with the Office electronic filing system (§ 1.27(b)(2)) .............................
By other than a small or micro entity

$80.00
$160.00

$80.00
$320.00

(b) Basic fee for filing each
application under 35 U.S.C. 111 for an
original design patent:

The public comment letters
(esp. the "Seventy Three
Practitioners" letter) pointed
out that no clearance in the
relevant time window had
sought clearance for the
DOCX rule (or for the CLE
rule that was proposed in
this same NPRM)

TABLE 2 TO PARAGRAPH (b)
By a micro entity (§ 1.29) ....................
By a small entity (§ 1.27(a)) ................
By other than a small or micro entity

PO 00000

Frm 00035

Fmt 4701

Sfmt 4702

$55.00
$110.00
$220.00

E:\FR\FM\31JYP2.SGM

31JYP2

Page 047

In the Final Rule, the PTO falsely claimed to have a clearance.
Federal Register / Vol. 85, No. 149 / Monday, August 3, 2020 / Rules and Regulations
F. Executive Order 13132 (Federalism)
This rulemaking does not contain
policies with federalism implications
sufficient to warrant preparation of a
Federalism Assessment under Executive
Order 13132 (Aug. 4, 1999).
G. Executive Order 13175 (Tribal
Consultation)
This rulemaking will not: (1) Have
substantial direct effects on one or more
Indian tribes, (2) impose substantial
direct compliance costs on Indian tribal
governments, or (3) preempt tribal law.
Therefore, a tribal summary impact
statement is not required under
Executive Order 13175 (Nov. 6, 2000).
H. Executive Order 13211 (Energy
Effects)
This rulemaking is not a significant
energy action under Executive Order
13211 (May 18, 2001) because this final
rulemaking is not likely to have a
significant adverse effect on the supply,
distribution, or use of energy. Therefore,
a Statement of Energy Effects is not
required under Executive Order 13211.
I. Executive Order 12988 (Civil Justice
Reform)
This rulemaking meets applicable
standards to minimize litigation,
eliminate ambiguity, and reduce burden
as set forth in sections 3(a) and 3(b)(2)
of Executive Order 12988 (Feb. 5, 1996).
J. Executive Order 13045 (Protection of
Children)
This rulemaking does not concern an
environmental risk to health or safety
that may disproportionately affect
children under Executive Order 13045
(Apr. 21, 1997).

khammond on DSKJM1Z7X2PROD with RULES3

K. Executive Order 12630 (Taking of
Private Property)
This rulemaking will not affect a
taking of private property or otherwise
have taking implications under
Executive Order 12630 (Mar. 15, 1988).
L. Congressional Review Act
Under the Congressional Review Act
provisions of the Small Business
Regulatory Enforcement Fairness Act of
1996 (5 U.S.C. 801 et seq.), prior to
issuing any final rule, the United States
Patent and Trademark Office will
submit a report containing the rule and
other required information to the United
States Senate, the United States House
of Representatives, and the Comptroller
General of the Government
Accountability Office. The changes in
this Final Rule are expected to result in
an annual effect on the economy of $100
million or more, a major increase in
costs or prices, or significant adverse

VerDate Sep<11>2014

21:09 Jul 31, 2020

Jkt 250001

effects on competition, employment,
investment, productivity, innovation, or
the ability of United States-based
enterprises to compete with foreignbased enterprises in domestic and
export markets. Therefore, this Final
Rule is a ‘‘major rule’’ as defined in 5
U.S.C. 804(2).
M. Unfunded Mandates Reform Act of
1995
The changes set forth in this
rulemaking do not involve a federal
intergovernmental mandate that will
result in the expenditure by state, local,
and tribal governments, in the aggregate,
of $100 million (as adjusted) or more in
any one year, or a federal private sector
mandate that will result in the
expenditure by the private sector of
$100 million (as adjusted) or more in
any one year, and will not significantly
or uniquely affect small governments.
Therefore, no actions are necessary
under the provisions of the Unfunded
Mandates Reform Act of 1995. See 2
U.S.C. 1501 et seq.
N. National Environmental Policy Act
This rulemaking will not have any
effect on the quality of the environment
and is thus categorically excluded from
review under the National
Environmental Policy Act of 1969. See
42 U.S.C. 4321 et seq.
O. National Technology Transfer and
Advancement Act
The requirements of section 12(d) of
the National Technology Transfer and
Advancement Act of 1995 (15 U.S.C.
272 note) are not applicable because this
rulemaking does not contain provisions
that involve the use of technical
standards.
P. Paperwork Reduction Act
The Paperwork Reduction Act of 1995
(44 U.S.C. 3501 et seq.) requires that the
Office consider the impact of paperwork
and other information collection
burdens imposed on the public. This
Final Rule involves information
collection requirements that are subject
to review by the OMB under the
Paperwork Reduction Act of 1995 (44
U.S.C. 3501–3549). The collection of
information involved in this Final Rule
have been reviewed and previously
approved by the OMB under control
numbers 0651–0012, 0651–0016, 0651–
0020, 0651–0021, 0651–0031, 0651–
0032, 0651–0033, 0651–0059, 0651–
0063, 0651–0064, 0651–0069, and 0651–
0075. In addition, updates to the
aforementioned information collections
as a result of this Final Rule have been
submitted to the OMB as nonsubstantive change requests.

PO 00000

Frm 00055

Fmt 4701

Sfmt 4700

46985

Notwithstanding any other provision
of law, no person is required to respond
to, nor shall any person be subject to a
penalty for failure to comply with, a
collection of information subject to the
requirements of the Paperwork
Reduction Act unless that collection of
information has a currently valid OMB
control number.
List of Subjects
37 CFR Part 1
Administrative practice and
procedure, Biologics, Courts, Freedom
of information, Inventions and patents,
Reporting and recordkeeping
requirements, Small businesses.
37 CFR Part 11
Administrative practice and
procedure, Inventions and patents,
Lawyers, Reporting and recordkeeping
requirements.
37 CFR Part 41
Administrative practice and
procedure, Inventions and patents,
Lawyers, Reporting and recordkeeping
requirements.
37 CFR Part 42
Administrative practice and
procedure, Inventions and patents,
Lawyers.
For the reasons set forth in the
preamble, 37 CFR parts 1, 11, 41, and
42 are amended as follows:
PART 1—RULES OF PRACTICE IN
PATENT CASES
■ 1. The authority citation for 37 CFR

part 1 continues to read as follows:
Authority: 35 U.S.C. 2(b)(2), unless
otherwise noted.
■ 2. Section 1.16 is amended by:
■ a. Revising paragraphs (a) through (e);
■ b. Adding table headings to the tables

in paragraphs (f) and (g);
■ c. Revising paragraph (h);
■ d. Adding a heading to the table in

paragraph (i);
■ e. Revising paragraphs (j) and (k);
■ f. Adding a heading to the table in

paragraph (l);
■ g. Revising paragraphs (m) through (s);
■ h. Adding a heading to the table in

paragraph (t); and
■ i. Adding paragraph (u).

The revisions and additions read as
follows:
§ 1.16 National application filing, search,
and examination fees.

(a) Basic fee for filing each application
under 35 U.S.C. 111 for an original
patent, except design, plant, or
provisional applications:

E:\FR\FM\03AUR3.SGM

03AUR3

Page 048
46984

Federal Register / Vol. 85, No. 149 / Monday, August 3, 2020 / Rules and Regulations

board adjustment fees and current fees,
the ‘‘current fees’’ column displays the
fees that are in effect as of January 2018.

khammond on DSKJM1Z7X2PROD with RULES3

Alternative 4: Baseline (Current Fee
Schedule)
The Office considered a no-action
alternative. This alternative would
retain the status quo, meaning that the
Office would continue the small and
micro entity discounts that Congress
provided in section 10 of the Act and
maintain fees as of January 2018.
This approach would not provide
sufficient aggregate revenue, based on
the assumptions found in the FY 2021
Budget, to accomplish the Office’s
rulemaking goals as set forth in Part III
of this Final Rule or the Strategic Plan.
IT improvement, progress on backlog
and pendency, and other improvement
activities would continue, but at a
significantly slower rate, as increases in
core patent examination costs that are
necessary to implement the strategic
objective to issue highly reliable
patents—such as increasing the time
examiners are provided to work on each
application—crowd out funding for
other improvements. Likewise, without
a fee increase, the USPTO would
deplete its operating reserves, leaving
the Office vulnerable to fiscal and
economic events. This would expose
core operations to unacceptable levels of
financial risk and would position the
Office to have to return to making
inefficient, short-term funding
decisions.
The fee schedule for Alternative 4:
Baseline (Current Fee Schedule) is
available at https://www.uspto.gov/
FeeSettingAndAdjusting in the
document entitled ‘‘Final Regulatory
Flexibility Analysis Tables.’’
Alternatives Specified by the RFA
The RFA provides that an agency
should also consider four specified
‘‘alternatives’’ or approaches, namely:
(1) Establishing different compliance or
reporting requirements or timetables
that take into account the resources
available to small entities; (2) clarifying,
consolidating, or simplifying
compliance and reporting requirements
under the rule for small entities; (3)
using performance rather than design
standards; and (4) exempting small
entities from coverage of the rule, or any
part thereof (5 U.S.C. 604(c)). The
USPTO discusses each of these
specified alternatives or approaches
below and describes how this Final Rule
is adopting these approaches.
Differing Requirements
As discussed above, the changes in
this Final Rule would continue existing

fee discounts for small and micro
entities that take into account the
reduced resources available to them as
well as offer new discounts when
applicable under AIA authority.
Specifically, micro entities would
continue to receive a 75 percent
reduction in patent fees under this
proposal and non-micro, small entities
would continue to pay 50 percent of the
fee.
This Final Rule sets fee levels but
does not set or alter procedural
requirements for asserting small or
micro entity status. To pay reduced
patent fees, small entities must merely
assert small entity status to pay reduced
patent fees. The small entity may make
this assertion by either checking a box
on the transmittal form, ‘‘Applicant
claims small entity status,’’ or by paying
the basic filing or basic national small
entity fee exactly. The process to claim
micro entity status is similar in that
eligible entities need only submit a
written certification of their status prior
to or at the time a reduced fee is paid.
This Final Rule does not change any
reporting requirements for any small or
micro entity. For both small and micro
entities, the burden to establish their
status is nominal (making an assertion
or submitting a certification), and the
benefit of the fee reductions (50 percent
for small entities and 75 percent for
micro entities) is significant.
This Final Rule makes the best use of
differing requirements for small and
micro entities. It also makes the best use
of the redesigned fee structure, as
discussed further below.
Clarification, Consolidation, or
Simplification of Requirements
This Final Rule pertains to setting or
adjusting patent fees. Any compliance
or reporting requirements in this rule
are de minimis and necessary to
implement lower fees. Therefore, any
clarifications, consolidations, or
simplifications to compliance and
reporting requirements for small entities
are not applicable or would not achieve
the objectives of this rulemaking.
Performance Standards
Performance standards do not apply
to the Final Rule.
Exemption for Small and Micro Entities
The final patent fee schedule
maintains a 50 percent reduction in fees
for small entities and a 75 percent
reduction in fees for micro entities. The
Office considered exempting small and
micro entities from paying increased
patent fees but determined that the
USPTO would lack statutory authority
for this approach. Section 10(b) of the

Act provides that ‘‘fees set or adjusted
under subsection (a) for filing,
searching, examining, issuing,
appealing, and maintaining patent
applications and patents shall be
reduced by 50 percent [for small
entities] and shall be reduced by 75
percent [for micro entities]’’ (emphasis
added). Neither the AIA nor any other
statute authorizes the USPTO simply to
exempt small or micro entities, as a
class of applicants, from paying
increased patent fees.
C. Executive Order 12866 (Regulatory
Planning and Review)
This Final Rule has been determined
to be economically significant for
purposes of Executive Order 12866
(Sept. 30, 1993). The Office has
developed an RIA as required for
rulemakings deemed to be economically
significant. The complete RIA is
available at https://www.uspto.gov/
FeeSettingAndAdjusting.
D. Executive Order 13563 (Improving
Regulation and Regulatory Review)
The Office has complied with
Executive Order 13563 (Jan. 18, 2011).
Specifically, the Office has, to the extent
feasible and applicable: (1) Made a
reasoned determination that the benefits
justify the costs of this Final Rule; (2)
tailored this Final Rule to impose the
least burden on society consistent with
obtaining the regulatory objectives; (3)
selected a regulatory approach that
maximizes net benefits; (4) specified
performance objectives; (5) identified
and assessed available alternatives; (6)
involved the public in an open
exchange of information and
perspectives among experts in relevant
disciplines, affected stakeholders in the
private sector, and the public as a
whole, and provided online access to
the rulemaking docket; (7) attempted to
promote coordination, simplification,
and harmonization across government
agencies and identified goals designed
to promote innovation; (8) considered
approaches that reduce burdens and
maintain flexibility and freedom of
choice for the public; and (9) ensured
the objectivity of scientific and
technological information and
processes.
E. Executive Order 13771 (Reducing
Regulation and Controlling Regulatory
Costs)
This Final Rule is not subject to the
requirements of Executive Order 13771
(Jan. 30, 2017) because this Final Rule
involves a transfer payment.

Though some components of the 2020 rule legitimately "involve a transfer payment," several
21:09 Jul 31, 2020 Jkt 250001 PO 00000 Frm 00054 Fmt 4701 Sfmt 4700 E:\FR\FM\03AUR3.SGM 03AUR3
components (including the DOCX rule, and the CLE rule curbed in 2023) did not.

VerDate Sep<11>2014

The 2015 "yearlong study" was not mentioned in the NPRM, and has never been made available
to the public (we obtained it by FOIA). Yet the PTO relies on it extensively (four mentions inPage
Final 049
Rule, and several
Supporting
Statements.)
Federal in
Register
/ Vol. 85,
No. 149 / Monday, August 3, 2020 / Rules and Regulations
46957
functioning and a DOCX version was
later filed within a certain time period.
Response: The current policy
regarding significant unplanned
electronic business system outages is
available at: https://
www.federalregister.gov/documents/
2018/08/30/2018-18897/filing-patentapplications-electronically-duringdesignated-significant-outages-of-theunited-states. The USPTO will post a
notice on its website in the event of a
designated significant unplanned
electronic business system outage and
indicate the dates during which the
alternative electronic filing means are
available due to such an outage. An
application filed via the alternative
electronic means during a designated
significant unplanned electronic
business system outage will be
considered to have been filed by the
USPTO’s electronic filing system and
thus will not incur the non-DOCX
surcharge or the fee required by section
10(h) of the AIA for a patent application
not filed by the USPTO’s electronic
filing system.
Comment 50: One commenter stated
that PDF format is the best and safest
format for ensuring that no text becomes
garbled or otherwise corrupted by the
USPTO system.
Response: There have been cases
where an applicant submitted PDF
documents that have been corrupted or
garbled that were traced back to specific
PDF creation software. By submitting in
text format, the extra step to convert to

Response: DOCX is a word-processing
file format that is part of Office Open
XML (OOXML), an XML-based open
standard approved by the Ecma
International® consortium and
subsequently by the ISO/IEC joint
technical committee.
For more information about the
OOXML standard, please see:
• ECMA–376 at http://www.ecmainternational.org/publications/
standards/Ecma-376.htm
• ISO/IEC 29500 at https://www.iso.org/
committee/45374/x/catalogue/
• NIST votes for US. Approval of
OOXML at https://www.nist.gov/
news-events/news/2008/03/nist-votesus-approval-modified-office-openxml-standard
The USPTO conducted a yearlong
study of the feasibility of processing text
in PDF documents. The results showed
that searchable text data is available in
some PDFs, but the order and accuracy
of the content could not be preserved.
With DOCX, the Office is able to use the
text directly and pass it on to USPTO
downstream systems, which results in
increased data accuracy and a more
streamlined patent process.
PDFs are not immune to viruses or
hidden malware. However, the USPTO
filing system is equipped with malware
and virus detection.
DOCX is supported by many popular
word-processing applications, such as
Microsoft Word, Google Docs, Pages,
and LibreOffice.

a word processor. The commenter
expressed concerns about how to rectify
such a situation and stated that knowing
that there is a problem and being able
to fix the problem in a timely manner
may be two completely different things.
Response: If there is an instance in
which an error occurs, the EBC should
be contacted for investigation at 1–866–
217–9197 (toll-free), 571–272–4100
(local), or [email protected]. The EBC is
open from 6:00 a.m. to 12:00 midnight
ET, Monday through Friday.
Comment 54: A few commenters
noted that the USPTO places the
responsibility on the practitioner to
check the generated PDF for accuracy.
One commenter wanted to confirm that
the authoritative document will be the
USPTO-generated PDF rather than the
DOCX that was submitted. Another
commenter felt that USPTO-generated
PDFs remove the applicant’s ability to
control accuracy, and applicants who
choose to guarantee accuracy by filing a
self-generated PDF should not be
penalized with increased fees.
Response: The authoritative
document will be the PDF that the
USPTO systems generate from the
DOCX. The filer has always been
responsible for the accuracy of the
documents being submitted. According
to surveys conducted by the USPTO, the
majority of applicants use wordprocessing software, such as Microsoft
Office and LibreOffice, which can
produce a DOCX file.

The "yearlong study" makes no such finding that "order and accuracy [of TEXT as opposed to
drawings] could not be preserved," and especially not for patent applications. (Complex and free-form
docs, like newspaper display ads, present problems. Patent applications, which must be a single
column, do not.) This is at best misleading. An inference of outright fabrication would be justified.

khammond on DSKJM1Z7X2PROD with RULES3

The "yearlong study" never mentions DOCX--the suggestion that it made a finding "With DOCX,
the Office is..." is at best misleading.
Comment 51: One commenter wrote
that PDF files are easier to manage when
filing, are better for long-term archival
use, can be generated in text-searchable
form, will not require fragmented filings
using both PDF and DOCX files, carry
fewer concerns with respect to malware
and viruses, and carry no licensing
concerns. The commenter expressed
that the DOCX file format is intended
for facile editing and by design is not
suited for archival purposes, will
require fragmented filing with different
file formats, will require archiving of
files in multiple file formats, carries
increased risk of malware and viruses,
is no better than other editable file
formats, and carries some uncertainty
regarding licensing status.

VerDate Sep<11>2014

21:09 Jul 31, 2020

Jkt 250001

validated and converted to PDF by
USPTO systems in a consistent manner.
The USPTO continuously performs
rigorous testing to ensure that document
integrity is preserved. To date, the
Office has not received notifications of
any issues resulting from the filing of
applications in DOCX format. If there is
an instance in which an error occurs,
the EBC should be contacted for
investigation at 1–866–217–9197 (tollfree), 571–272–4100 (local), or ebc@
uspto.gov. The EBC is open from 6:00
a.m. to 12:00 midnight ET, Monday
through Friday.
Comment 53: One commenter asked
about a situation in which the USPTO’s
rendering engine has changed the result
relative to what the practitioner saw on

PO 00000

Frm 00027

Fmt 4701

Sfmt 4700

generate PDF documents for the
uploaded DOCX files. At this time,
applicants are encouraged to review the
PDF documents before submission. The
step of applicants reviewing their selfgenerated PDF is being replaced with
their review of the USPTO-generated
PDF document. The amount of time
required by an applicant to review the
self-generated PDF is comparable to the
time to review the USPTO-generated
PDF.
As a part of the DOCX intake process,
preliminary validation is performed on
DOCX documents at the time of upload.
The system immediately detects and
supplies the applicant with useful error
and warning messages, allowing for
adjustments to patent applications early

E:\FR\FM\03AUR3.SGM

03AUR3

Page 050

December 2023 Supporting Statement
In its December 2023 Supporting Statement for 202309-0651-002,
https://www.reginfo.gov/public/do/DownloadDocument?objectID=135781702
the Patent Office claims that DOCX presents information correctly, “in a manner that is
independent of software, hardware, and operating system:”

There is no statement in any standards document or any other source to support a
claim that a DOCX is presented in an “independent” manner.
•

Perhaps the PTO means that if one reads the sentences hyper-literally to refer
only to “text,” and ignores the other content that governs correctness of a patent
application. In that case, while not literally false, the PTO’s statement is
materially misleading.

•

If the sentence is read at face value, it’s just not true.

•

In either case, the PTO breached its obligation of candor to the tribunal.

•

The earliest that significant error checking is possible is 18-month publication.
The bulk of that publication will not begin until July 2025.

The PTO claims:

None of these materials were made available to the public during NPRM notice-andcomment, and none have been included in the record of any ICR. The PTO’s “yearlong
study” was not mentioned in the NPRM, and was not made available during the NPRM
comment period. One copy was produced by FOIA; it is still is not available to the
public. Though requested by FOIA two years ago, the Patent Office has produced none
of the other documents implied here. The PTO relies on secret documents.

Page 051

Additional gotchas
•

The Patent Office’s DOCX software doesn’t accept the “standard”
o The standard allows multiple features to be combined: chemical formulas
inside tables, equations inside quote fields, equations in specific fonts, etc.
The PTO’s DOCX software won’t allow these combinations—some yield error
messages, sometimes the PTO appears to accept and then screws it up.
o Word provides a number of features that patent attorneys use to reduce the
likelihood of error: bookmarks, cross references, fields, merge fields, and the
like. The PTO’s DOCX software gives error messages and refuses to accept
patent applications that use these standard features.
o The PTO’s filing system sometimes renders Greek letters (among other
characters) as a box --Greek letters are common in the scientific notation
used in patent applications.
o For wide tables and wide chemical formulae, past practice was to put the
wide tables on landscape pages, and rotate them by 90° in the PDF. That’s
easy for PDF submissions. But the PTO’s DOCX filing system refuses to
accept the document. We have reported this error several times over several
years. The PTO has not fixed it, and we have not figured out a work-around.
o The PTO has no published document that gives a full explanation for which
parts of the “standard” are accepted by the PTO’s DOCX filing system, which
are rejected, or anything else about the PTO’s implementation of DOCX, and
no description of work-arounds. From an applicant’s point of view, the PTO’s
DOCX filing system is completely random.

•

The PTO’s rationale for DOCX filing is that “DOCX is a standard.” But the PTO
doesn’t implement the standard. Implementing only the easy part of a standard
is not standard conforming.

•

A detailed explanation of standard DOCX features that are rejected by PTO’s
software is in our 60-day letter for 202309-0651-002
https://www.reginfo.gov/public/do/DownloadDocument?objectID=135639300 at
pages 6-7 and Exhibit 4 at pages 83-94.

•

Detailed lists of procedural defects in April 29 letter and October 2023 letter for
202309-0651-002 at
https://www.reginfo.gov/public/do/DownloadDocument?objectID=136723102

Page 052

Document tampering
•

PTO software uploads the document from the user, then tampers with it before
storing it. The stated rationales are uniformity of format, and “validation.”
o The PTO’s document tampering introduces error, and erases the audit trail so
there’s no possibility whatsoever of seeking correction
• In Example 8, the $8000 in extra claim fees, the problem arose because
the PTO converts “new line” characters ( ) to paragraph characters (¶).
Perhaps it appeared innocuous, but because of the complexity of the
DOCX standard and the variability of DOCX documents, it’s impossible to
predict all the cases. Any kind of document tampering will necessarily
introduce error.
• In some cases, if a patent claim uses a number (such as “2” in “2 inches”)
the Patent Office’s document tampering misinterprets the meaning of “2”
and starts renumbering paragraphs.

•

The PTO discards the copy the applicant uploads. The only copy that’s stored is
the PTO’s tampered “validated” copy.
o Violates Federal Records Act. The record that should be the “of record” copy
is tampered with, and the original is discarded.
o Violates Federal Rule of Evidence 1002, the “original document” rule.
o The PTO must follow courts and other agencies, and take a “hands off”
approach—the document that the applicant uploads must be the “of record”
copy, for all purposes

Page 053

Public burden estimates
Public comment letters have given detailed work-ups of burden, ranging from $200
million to $600 million (assuming an apples-to-apples comparison, in which
proofreading yields the same level of reliability as PDF). Several of these estimates
have been given to the Patent Office on multiple occasions. The Patent Office has
never even acknowledged any of these estimates, let alone disagreed with them:
•

David Boundy, estimating $200 million
o November 30, 2020, ICR 202011-0651-001, 30-day letter at
https://www.reginfo.gov/public/do/DownloadDocument?objectID=107472802
at pages 3-5
o August 7, 2023, ICR 202309-0651-002, 60-day letter at
https://www.reginfo.gov/public/do/DownloadDocument?objectID=135639300,
at PDF pages 111-113

•

Brad Forrest, estimating $200 million
o November 30, 2020, ICR 202011-0651-001, 30-day letter at
https://www.reginfo.gov/public/do/DownloadDocument?objectID=107472802
at pages 31-39
o December 29, 2020, ICR 202011-0651-001, 30-day letter at
https://www.reginfo.gov/public/do/DownloadDocument?objectID=107472602
PDF pages 12-20
o August 7, 2023, ICR 202309-0651-002, 60-day letter at
https://www.reginfo.gov/public/do/DownloadDocument?objectID=135639300 ,
at PDF pages 99-107

•

Carl Oppedahl, on behalf of One Hundred Seven Patent Practitioners, estimating
$600 million
o August 6, 2023, ICR 202309-0651-002, 60-day comment letter at
https://www.reginfo.gov/public/do/DownloadDocument?objectID=135640900,
at pages 6-10 works up a total of $600 million.

•

Kilpatrick Townsend & Stockton, estimated $600 to $700 million per year
o August 2023, 60-day letter in ICR 202309-0651-002
https://www.reginfo.gov/public/do/DownloadDocument?objectID=135640600
at page 2-3

The Patent Office has never disagreed. Instead, the PTO has simply ignored these two
specific elements of burden. The PTO has never acknowledged the proofreading and
error correction burden, let alone disagreed with the public's estimates.

Page 054

Questions to ask the agency
•

In 1992, 1999, and 2012, Congress amended the Patent Act to make Patent
Office “more like a business.” Congress gave the Patent Office lots of
exemptions from Title 5. For example:
o Other federal bonuses capped at 10%, with a handful allowed at 20%. For
Patent Office, SES officers eligible for 50% bonuses. 35 U.S.C. § 3(b)(2)(B)
o The Patent Office keeps its own fee collections. Fees do not flow into the
general treasury, and are not subject to appropriations oversight. 35 U.S.C.
§ 42(b) and (c).

•

Bonuses are set by a “Performance Review Committee” that does not observe
private sector rules for independent members of compensation committees.
E.g., 88 Fed. Reg. 52131 (Aug. 7, 2023) (all members but one are PTO
executives setting their own bonuses); 86 Fed. Reg. 49001 (Sep. 1, 2021) (zero
outside members).

•

Speculation: what contracts has the PTO entered relating to DOCX? What quid
pro quo to the agency? To individuals?

•

Toward that end, I filed a FOIA request in 2021 for all information underlying the
DOCX rule. FOIA Request F-21-00169 (Jul. 12, 2021).
o The PTO promised a rolling production of 500 pages per month. The PTO
made one production in January 2023 (six months late), and another in
November 2023. The communications with OMB relating to the Paperwork
Reduction Act were blank-sheet redacted, in violation of § 3507(e)(2).
o I followed up with a request for all documents underlying the “yearlong study.”
FOIA Request F-22-00092 (March 28, 2022). Two years later, the PTO has
produced zero documents.

•

Is there a coverup?

Page 059

Public comments and articles
•

At 2019-2020 NPRM phase:
o Seventy-Three Patent Practitioners,
https://www.uspto.gov/sites/default/files/documents/Comment_Seventy_Three_P
atent_Practitioners_092719.pdf
o Carl Oppedahl,
https://www.uspto.gov/sites/default/files/documents/Comment_Carl_Oppedahl_0
81219.pdf

•

2020 triennial ICR for 0651-0032
o there were fifty-four letters, mostly on this topic
o In this Supporting Statement, the PTO avoided the DOCX issue by stating it
would be “premature” to consider burden of DOCX filing

•

At 2023 ICR 202309-0651-002 for a new information collection, limited to transition
costs for DOCX, and leaving the core “completing and reviewing” burden for this
triennial ICR
o there were fifteen letters, all specific to DOCX
o Especially relevant, if you read only one: 60-day letter of 152 Patent
Practitioners,
https://www.reginfo.gov/public/do/DownloadDocument?objectID=135639300
which gives an extended tutorial on the technology issues, and the 30-day letter
https://www.reginfo.gov/public/do/DownloadDocument?objectID=136723102
which discusses the pattern of misleading non-responses to comments in the
PTO’s Supporting Statement
o The two letters of Carl Oppedahl are both well-researched,
https://www.reginfo.gov/public/do/DownloadDocument?objectID=135640200 and
https://www.reginfo.gov/public/do/DownloadDocument?objectID=135640900

•

Articles:
o Julie Burke, The Trains, Planes and Automobiles of Correcting DOCX-Related
Errors, (Mar. 19, 2024), https://ipwatchdog.com/2024/03/19/trains-planesautomobiles-correcting-docx-related-errors/id=174452/
o Julie Burke, USPTO Flexes Its AIA Powers To Make Retroactive Substantive
MPEP Policy Changes (Mar. 23, 2023)
https://ipwatchdog.com/2023/03/23/uspto-flexes-aia-powers-make-retroactivesubstantive-mpep-policy-changes/id=158222/ and Recent MPEP Changes
Complicate the Sticky Wicket of Restriction Thickets (Mar. 14, 2023)
https://ipwatchdog.com/2023/03/14/recent-mpep-changes-complicate-stickywicket-restriction-thickets/id=157729/ both relate to the MPEP guidance issue in
our 30-day letter

New England Intellectual Property, LLC
291 Main Street
West Newbury, MA 01985
978-363-1700
[email protected]
www.newenglandip.com

May 15, 2024

Nicholas A. Fraser and Jason Grimes
Office of Information and Regulatory Affairs
Office of Management and Budget
725 17th St. NW
Washington, D.C.

Re:

ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30day notice at 89 Fed. Reg. 21495 (Mar. 28, 2024)

Dear Mr. Fraser and Mr. Grimes:
Thank you for talking with David Boundy, Bradley Forrest, and me last week. In our
conversation, you asked about the Patent Office’s “auxiliary PDF” modification to the DOCX
rule. I started to answer, but there was insufficient time to provide my comments. As a result, I
would like to present my thoughts on this subject.
The auxiliary PDF proposal has never received the thorough vetting that this proposal
deserves. The Patent Office has never run a rulemaking notice-and-comment for this modified
information collection, only a “stealth” comment period for ICR 202309-0651-002.1 In 2023,
the Patent Office ran notices in the Federal Register to permit an auxiliary PDF, but only “until
further notice,” with no request for comment, no commitments to correctness in the processing of

1

ICR 202309-0651-002: 60-day notice at 88 Fed. Reg. 37039 (Jun. 6, 2023), 30-day notice at 88 Fed. Reg.
66414 (Sep. 27, 2023)

applicants’ filings, and no commitment to long-term availability of correction. Without the
proper rulemaking procedure, the shortcomings of this proposal have not been properly vetted.
The biggest issue with the auxiliary PDF proposal is that there is no legal framework to
determine how this auxiliary PDF will be handled in the future. In the example used in our
presentation last Wednesday, the change of a number from 0.2 to 10.2 is so subtle that it is
unlikely to be detected in a human proofreading. As such, it may be many years until the error is
detected in litigation. At this point, the defendant will assert that they relied on the erroneous
number in the DOCX version and the plaintiff will assert that the auxiliary PDF should control.
The courts will be left without a fair resolution and will have to choose harm on one of two
innocent victims of the USPTO software error2.
It has been hinted by the USPTO that the DOCX version will be definitive for their
purposes, but this only applies to patent prosecution before the USPTO3. It may be imagined
that the USPTO will treat the auxiliary PDF as an Appendix, as is used in some patent
prosecution. In these cases, the Appendix can be used in US prosecution to support future
changes during prosecution. It is unclear if the USPTO will use the auxiliary PDF as prior art if
the error is useful in a rejection of an unrelated patent application. In short, there are lots of
unanswered questions surrounding the auxiliary PDF.
One of the USPTO’s Federal Register Notices4 warns that the auxiliary PDF is deleted
from the official record after a period of time (87 Fed. Reg. 25227 at col. 1-2) 5. This is in
violation of the Federal Records Act, 44 U.S.C. § 3301(a)(1), which requires agencies to retain

2

Because the legal effect of the auxiliary PDF is so uncertain, applicants are left with essentially the same
burden as without -- a detailed character-by-character proofread is the only way to achieve the same level of
reliability as PDF filing. The auxiliary PDF just moves the burden from one place to another: it doesn’t reduce it.
3

When filing through PatentCenter, the Patent Office’s computer system requires the applicant to surrender
any claim that the Patent Office’s tampered DOCX is incorrect through clicking through a pop-up warning. The
legal weight of this disclaimer will not be known until it is litigated.
4
In Filing Patent Applications in DOCX Format, 87 Fed. Reg. 25226 (Apr. 28, 2022), default retention
period for the auxiliary PDF was specified at three years; an applicant must file a separate petition for indefinite
storage. For applicants, the pre-2019 default was indefinite storage of the PDF original filing. Indefinite storage is
essential: a reliable copy may be essential to litigation as late as 26 years after filing the application.
5

The Patent Office also warns that it will generate its own document-tampered DOCX file (misnamed the
“validated” DOCX), and rely on it, and discard the DOCX uploaded by the applicant. Id.

originals, not “defac[e], alter[ ], corrupt[ ], [or] delet[e]” a record, and then discard the original.
44 U.S.C. § 3106.
One of the Patent Office’s notices in the Federal Register, Submitting Patent Applications
in Structured Text Format and Reliance on the Text Version as the Source or Evidentiary Copy,
86 Fed. Reg. 29571 (Jun. 2, 2021), set a rather draconian policy: “Applicants should not expect
to have a request to correct the electronic record granted if the request … is filed more than one
year after submission [of the patent application]” (86 Fed. Reg. 29572 at col. 1). This one-year
deadline has never been rescinded, and it apparently applies to corrections based on the auxiliary
PDF. The Office has never explained how applicants can review, let alone correct, the electronic
record before a patent issues at about 3-5 years, or (for about 80% of patent applications)
publishes at 18 months. This one-year limit was never mentioned in the NPRM or final rule.
The Patent Office has never sought comment or observed any of the other procedures of the
Paperwork Reduction Act for this one-year limit on corrections.
When a U.S. application serves as a basis for foreign filing under various treaties, it
seems doubtful that the auxiliary PDF will be acceptable in most foreign jurisdictions. U.S. law
has a unique feature that makes the auxiliary PDF useful (if burdensome) under U.S. law—a
U.S. patent application can incorporate other documents by reference, and future corrections may
be based on those incorporations. However, no other (major) country recognizes incorporation
by reference. So the auxiliary PDF seems unlikely to support the kind of correction that the
Patent Office contemplates in a way that will allow correction for purposes of other jurisdictions.
In other words, the loss of patent rights imposed by the Patent Office is likely unrecoverable in
most other countries.
The only purpose for the auxiliary PDF is to backstop errors in the Patent Office’s
handling of the DOCX file. When paired with an auxiliary PDF, the DOCX lacks practical
utility and is unnecessarily duplicative. The auxiliary PDF allows proofreading of the Patent
Office’s processing of a DOCX—but does so via a 100-to-1 disproportionate shift of burden to
the public. Terms of clearance should clear collection of the auxiliary PDF and not the DOCX.
The auxiliary PDF is just a mechanism to shift costs back to the public, rather than absorbing the
costs the PTO creates for itself.

As industry leaders, we do not feel that the auxiliary PDF is a viable solution to the
problems created by the PTO’s DOCX information collection.

Sincerely,

Richard A. Baker, Jr.

New England Intellectual Property, LLC
291 Main Street
West Newbury, MA 01985
978-363-1700
[email protected]
www.newenglandip.com

May 30, 2024

Nicholas A. Fraser and Jason Grimes
Office of Information and Regulatory Affairs
Office of Management and Budget
725 17th St. NW
Washington, D.C.

Re:

ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30day notice at 89 Fed. Reg. 21495 (Mar. 28, 2024)

Dear Mr. Fraser and Mr. Grimes:
Thank you for your consideration of my May 15th letter and for your question.
Even if the USPTO maintains the PDF version of the submitted patent application in
perpetuity and declares it to be “evidentiary,” there are still two major sources of burden that
have to be estimated and booked.
A patent is only as valuable as its ability to be enforced. Some patents are high stakes
with hundreds of millions of dollars at stake, thousands of jobs, and the fate of one company or
the other. For small companies, survival is often tied to patents—patents are often the only
securitizeable asset to obtain startup capital and are often essential to the preservation of a new
market niche against free-riding by market incumbents. As I will explain below, the “evidentiary
PDF” proposal does close to nothing to attenuate uncertainty in enforcement. That post-issue
uncertainty forces the burden onto applicants to re-create certainty during agency proceedings.
Any patent attorney will assure you that a large fraction of the work in a patent application is

taking steps to reduce uncertainty confounders—this would simply be another task on that list.
To achieve the same level of certainty as applicants had with the PDF status quo ante will
require essentially the same extensive proofreading and error correction as all the past letters
have explained—mid-nine figures.
There are two important legal factors.
First, the USPTO has no authority to promulgate regulations that have any effect after a
patent issues. The USPTO’s primary rulemaking statute, 35 U.S.C. § 2(b)(2)(A), only allows the
USPTO to promulgate regulations to govern proceedings in the Office. Courts regularly deny
any post-issue effect to USPTO regulations. The USPTO has no authority to displace the text of
the patent itself by simply declaring something else to be “authoritative” or “evidentiary.”
Second, competitors are entitled to rely on the text of issued patents. Because the patent
owner is the only party with the ability to detect and correct ambiguities, the risk of noncorrection falls on the patent owner. The Federal Circuit held that “the PTO [has] no authority to
correct the claims retroactively,” a “certificate of correction is only effective for causes of action
arising after it was issued,” “major errors are subject only to correction by the PTO” (not by
courts), and a court has no authority to correct any significant error that could have been
corrected by certificate of correction. Novo Industries, L.P. v. Micro Molds Corp., 350 F.3d
1348, 1356-57 (Fed. Cir. 2003). Though it’s always possible that courts would treat this kind of
USPTO-created error differently than the Novo facts, until that happens, the law is certainly clear
enough to put applicants on notice that the consequences of the USPTO’s error in an issued
patent almost certainly fall on the applicant, unless and until the applicant undertakes
proofreading and error correction. Because of the asymmetric risk (not to mention the
malpractice risk for an attorney that skips this step), that proofreading has to happen for every
patent, not just the ones where the USPTO actually introduces an error.
All the evidentiary PDF proposal does is create a permissive structure for the correction
of patents; it does nothing to reduce the burden of proofreading every application to find the
errors, and the burden of filing the paperwork to actually seek the correction. Because the
evidentiary PDF proposal does nothing (at least nothing legally reliable) to reduce the risk during

the time between the issue of the patent and the issue of the certificate of correction, that burden
of proofreading and correction has to be undertaken before the patent issues.
Likewise, the PTO offers no procedural mechanism to do the proofreading. The first
time the applicant has a copy of the text that will be used in the published patent is upon
issuance. There’s no way to do the proofreading until it’s too late.
In the example used in our May 8 presentation, the change of a number from 0.2 to 10.2
is so subtle that it could well have gone undetected. If the error remains when a patent is to be
enforced (typically by licensing, sometimes by litigation), one party will assert that they relied on
the PTO’s published text, and the other will assert that the evidentiary PDF should control. On
the case law as it stands today, even the proponents of the evidentiary PDF proposal
acknowledge that it only gives “a little better chance,” far from certainty. As a practical matter,
the evidentiary PDF proposal does nothing to reduce the legal obligation of careful proofreading
and error correction at the earliest possible time. That’s a nine-figure burden, and the PTO has to
book it.
The second source of burden in the evidentiary PDF proposal is that it opens the door to
gaming the system by deliberately submitting different documents in an attempt to avoid full
examination.
It should be facially obvious that any proposal to have two legally effective versions of
the same file is bad public policy and an invitation for expensive problems in the future.
A third issue with the evidentiary PDF proposal is that the USPTO, in the June 6, 2023,
Federal Register Notice, states that the acceptance of the PDF version “is being extended until
further notice.” This implies that the auxiliary PDF solution is temporary and that the USPTO
can end this program at any time without following the rulemaking process. To go the extra
step—from simply retaining the auxiliary PDF to promoting it to “evidentiary” is a nonprocedural rule that requires full notice-and-comment rulemaking. The evidentiary PDF
proposal has never been exposed to an information collection procedure, and without that, no one
can rely on it in either direction. 44 U.S.C. § 3512. At the very least, the evidentiary PDF
proposal should go through the full rulemaking and information collection clearance process.

Thank you again for considering our concerns,

Sincerely,

Richard A. Baker, Jr.

PTAAARMIGAN
PATENT AND TRADEMARK ATTORNEYS, AGENTS AND APPLICANTS FOR RESTORATION AND
MAINTENANCE OF INTEGRITY IN GOVERNMENT
MAILING:

P.O. BOX 590372, NEWTON MA 02459

E-MAIL:

[email protected]

June 4, 2024
Via Email

[email protected];

Nicholas A. Fraser and Justin Grimes
Office of Information and Regulatory Affairs
Office of Management and Budget
725 17th St. NW
Washington, D.C.
Re:

ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day
notice at 89 Fed. Reg. 21495 (Mar. 28, 2024)

Dear Mr. Fraser and Mr. Grimes—
This is a reply to your email of May 28 asking whether the USPTO’s June 2023 guidance
“achieves the same thing as if it was designated as the evidentiary copy.” That also begs the
question, whether the “auxiliary PDF is permanent and is evidentiary copy forever” variant on
the USPTO’s DOCX information collection can be cleared. I promised it to you yesterday, but
it’s a group letter that took longer to put together than I expected.
The answer is that neither the “evidentiary PDF” nor the USPTO’s June 6, 2023 guidance
significantly reduce burden of DOCX filing. As Mr. Baker put it in his letter of May 30, “It
should be facially obvious that any proposal to have two legally effective versions of the same
file is bad public policy and an invitation for expensive problems in the future.”
Mr. Baker’s letter of May 30 (and his follow-up email of June 1, with the Oklahoma case
attached) explained the two factors that, in combination, mean that the “evidentiary PDF”
proposal doesn’t work at its end point: (a) when a printed patent has errors, courts tend to lay the
consequences on the patent owner, and do not require the competitor/infringer to read past the
text of the issued patent to resolve the error. (b) The USPTO has no authority to make any
promise of “evidentiary” status post-issue, or to promise how other countries’ patent offices will
treat errors created by the USPTO. Because of those two features, patents are uniquely
intolerant of error, and the “evidentiary PDF” proposal gives essentially no relief from the
proofreading and error correction burden. That burden was estimated in the letters of ICR
202309-0651-002, at $200 million to $600 million per year (see letters of Boundy, Oppedahl,
and Cretsinger). Having the “evidentiary PDF” sitting silently in the file, while the USPTO
prepares the legally-effective patent from an unreliable DOCX, doesn’t reduce burden at all.
This letter will explain why the “evidentiary PDF” proposal also leaves in place much of
the burden at the start point, at the time of filing.
Patent and Trademark Attorneys, Agents and Applicants for Restoration and Maintenance of Integrity in Government
PTAAARMIGAN LLC | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 2
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024
PTAAARMIGAN is a not-for-profit § 501(c)(4) social welfare organization, that
educates and advocates on behalf of patent and trademark attorneys, agents, and applicants.
I.

Background facts

To understand the full implications of both the base DOCX proposal and the “evidentiary
PDF” variant, and why both are flawed, it helps to have all the basic facts laid out. In order to
get your successor up to speed, here is a consolidated exposition of the relevant facts.
Many of the people in our group were software engineers earlier in our careers. We
developed a two-page “user requirements specification” and have attached it as an exhibit to this
letter. It might be useful to review that first, before you review this list.
•

The PDF standard is a voluntary consensus standard (OMB Circular A-119). The very
first sentence of the PDF standard, ISO standard 32000-2:2020, states the central design
goal: “PDF enables users to exchange and view electronic documents easily and reliably,
independent of the environment in which they were created or the environment in which they are
viewed or printed.” In contrast, the DOCX standard is designed for editing, not for

archival or interchange purposes. In order to allow editing on different users’ computers
with different software installed, the DOCX standard (ISO 29500) leaves many dozens of
characteristics as “implementation defined” or “unspecified.” In its Supporting
Statement for ICR 202309-0651-002, the USPTO claimed “DOCX presents documents,
including text and formatting, in a manner that is independent of software, hardware, or
operating system.” The USPTO gave no “supporting record” for this statement. The
DOCX standard makes no such promise (indeed, it promises the opposite), and empirical
evidence that many users have brought to the USPTO’s attention (see our handout to you
of May 8) show that the USPTO’s representations to you are unequivocally false.
•

Many patent applications are initially drafted by the client or inventor, not the patent
attorney. The client/inventor does things their way, making full use of the features of
whatever word processor they use. The constraints of the USPTO’s “Quick Start Guide”
aren’t known to this person.

•

As the Patent Office acknowledged, about 20% of filers do not use a DOCX-based word
processor (Libre Office, Word 2003, Word Perfect, Apple Pages, Google Docs, etc.)
These word processors have features to export DOCX files, but that feature tends to be
low reliability. The DOCX file often loses meaning-significant formatting information,
equations, special symbols, and the like.

•

The applicant has to generate a DOCX file and a PDF file simultaneously. Whatever
word processor the applicant is using, generating two files that are supposed to be
identical is error-prone—if the phone rings during the process, the user might introduce a
discrepancy between the two versions.

•

Many patent attorneys want the inventor to review the exact document that will be filed
as the patent application. Historically, that’s been done by giving the inventor a PDF—
it’s guaranteed to be the same on all computers. But review of the exact document is
PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 3
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024
simply impossible if a DOCX appears differently on the inventor’s computer, the
attorney’s computer, and the PTO’s computer, and if the USPTO changes what’s
uploaded before storing it.
•

The USPTO’s DOCX intake software does not accept all DOCX files that conform to the
ISO standard. The USPTO has never disclosed what software it uses, so we assume that
it’s a home-grown DOCX scraper. In order to reduce its own costs, the USPTO skipped
out on implementing the full standard. Inventors and patent attorneys use word
processing features that reduce error, such as bookmarks, style-ref cross-references to
automatically copy of text from one place to another, automated cross-references to
ensure consistent references to outline section numbers, landscape pages, and the like.
The USPTO’s DOCX intake software didn’t implement these. Instead of accepting the
full range of the standard, the USPTO’s DOCX intake software gives error messages and
refuses to accept perfectly “standard” DOCX files. So the user has to do hand-editing to
“dumb down” the patent application, or simply give up use of error-proofing features. In
either case, the USPTO forces a change from “existing recordkeeping practices” and
forces applicants to introduce risk of error.

•

The USPTO does not store the applicant’s uploaded DOCX. Instead, the USPTO
mangles it in the name of “validation” and forcing the DOCX into a uniform format.
This step introduces errors. We don’t know exactly what the USPTO does today.
however, in the past, the USPTO would replace “new line” characters ( ) with
“paragraph” characters (¶). This introduces errors of paragraph numbering, etc. Changes
make automated file compare difficult. The Federal Records Act and Federal Rule of
Evidence 1002 and basic principles of reliable software engineering require that agencies
store the original version of documents; the USPTO’s DOCX intake software doesn’t.

•

The USPTO’s DOCX intake software attempts to “validate” the patent application, but
gets fooled, and won’t accept valid patent applications. For example, it requires a page
with the word “CLAIMS” at the top. Some attorneys used to use the page heading “C L
A I M S” Or sometimes there’s a typo, “CLIAMS” A warning makes sense. But the
USPTO’s DOCX intake software implemented these as hard errors.

•

The USPTO makes this DOCX available for review, but that review occurs on the user’s
computer. The problem is that a DOCX file appears differently on different computers,
depending on software versions installed, add-ons for equations, and the like. Therefore,
that review tells nothing about how the USPTO’s computer will read or understand the
very same DOCX file. (The USPTO books no additional time for this review, and offers
no other acknowledgment that any burden even exists.)

•

During that review, the USPTO’s computers have a 30-minute timeout. If the review
takes more than 30 minutes (either because of the detail or because the phone rings), the
timeout will log the user out of the system.

•

The USPTO stores that mangled DOCX for its own processing, and discards the user’s
upload The USPTO scrapes the text out of the mangled DOCX, and that's the text that

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 4
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024
will be used for examination and printing. The user’s original upload is not stored in any
form.
•

ISO approved standard 29500 for DOCX. But Microsoft doesn’t implement the standard
(see Microsoft’s 900-page list of departures from the standard
(https://msopenspecs.azureedge.net/files/MS-OI29500/%5bMS-OI29500%5d.pdf, linked
in the May 8 handout, page 39). Microsoft continues to evolve its implementation of
DOCX. Microsoft has indicated that in a few years, it will stop offering desktop Office,
and will only offer software-as-a-service Office 365. In the future, Microsoft will be able
to (and almost certainly will) make changes that introduce new features into DOCX files
that are not accepted by the USPTO’s DOCX intake software. The way the USPTO
implements DOCX intake and the way Microsoft implements DOCX creation are on a
collision course.

•

The USPTO likewise does not store the auxiliary PDF that the user uploads. Instead, the
version that the USPTO proposes as “evidentiary” is a USPTO-generated version that has
been flattened to a fairly-low-resolution halftoned bitmap (just a picture of pixels, with no
text stored—it’s so poor that it can’t be reliably OCR’ed). The USPTO will discard the
original uploaded PDF. That flattened version of the applicant's PDF will be stored
permanently but never practically used by the USPTO unless and until the applicant
relies on it to seek correction. The USPTO’s acknowledgment receipt does not
acknowledge the applicant’s uploaded PDF; it gives a signature (fingerprint or hash) of
the flattened file, so if the USPTO makes an error, there’s no way to prove that a
replacement that might be offered later is the same. Any full-text content of the
applicant’s PDF file won’t be available to the applicant from the USPTO’s copy because,
as part of the USPTO’s transformation of the applicant’s uploaded PDF, the USPTO
flattened that content away.

•

The USPTO's last word on the subject (June 2021) is that corrections of USPTO errors
would only be permitted for a year. We assume that the “evidentiary PDF” proposal
envisions that the USPTO will run a Federal Register notice that (a) the evidentiary PDF
option will be available permanently, not “until further notice,” and (b) the applicant can
seek correction up to the end of the patent's life, and (c) that error correction must use the
normal mechanisms for correction, and the USPTO will bear none of the burden of
seeking the errors it introduced or self-correcting them.

•

Under the “evidentiary PDF” proposal, the burden of proofreading and filing the papers
to obtain corrections will still fall entirely on the applicant. Given that the USPTO
flattens the applicant’s uploaded PDF file, and the “evidentiary PDF” is not the file that
the applicant uploaded, the applicant must also proofread the flattened PDF file. Again,
it is entirely the applicant’s burden.

•

The USPTO has never put the “evidentiary PDF” proposal up for notice-and-comment
under either the Administrative Procedure Act or the Paperwork Reduction Act. In ICR
202011-0651-006, the USPTO’s Supporting Statement states it would be “premature” to
seek any clearance for any variant of any DOCX proposal. In ICR 202309-0651-002, the
USPTO only estimated “adjusting existing ways to comply” burden, and ignored

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 5
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024
proofreading and error correction burden. The public has provided the same estimates
multiple times, most recently in ICR 202309-0651-002 (see letters of Boundy, Oppedahl,
and Cretsinger), and those estimates run $200 million to $600 million per year. In this
ICR 202304-0651-001, the USPTO doesn’t mention DOCX at all, let alone offer
estimates for comment. The USPTO has never offered any estimates of its own, and has
never identified any analytical flaw in the public’s estimates of nine-figure burden.
•

The relevant statute (35 U.S.C. § 254) is absolutely clear that correction will have no
effect for some situations, only for “causes thereafter arising.” We see no plausible
mechanism that corrects all errors for all purposes back to the time the USPTO created
them.

•

Normally, it’s the applicant’s burden to make the record clear; it’s not an infringer’s
burden to figure out ambiguities. The normal inclination for courts is that ambiguities are
resolved against the party that had the opportunity to repair the ambiguity, and that
inclination is especially strong in patents (as the example in Mr. Baker’s email of May 31
demonstrated).

•

The statute only gives the USPTO the authority to govern how the record is interpreted
“in the Office,” § 2(b)(2)(A), not by courts. The USPTO has attempted in the past to tell
courts how to treat the record, and the courts have declined the USPTO’s invitation. E.g.,
Hill-Rom Co. v. Kinetic Concepts, Inc., 209 F.3d 1337, 1341 n. 1 (Fed. Cir. 2000). So,
we won’t really know what it means for the PDF to be the “evidentiary copy” in cases
where it differs from the USPTO’s typesetting until the issue gets litigated. In the
meantime, applicants must operate on the worst-case assumption.

•

Many U.S. patent applications form the basis for international filings under international
treaties. These treaties specify that a first-country filing has the same effect as if it were
filed in the second country on the first-country’s date, under the second country’s law. In
other words, correctness of initial U.S. filing will be measured under the lowest common
denominator of Chinese law, Japanese law, Korean law, German law, U.K. law, etc.
Other countries’ law tends to be even stricter than U.S. law. For example, to our
knowledge, none of these treaties or national laws recognize appendices, incorporation by
reference, or auxiliary documents. Further, national patent offices share patent
documents using an agreed internal electronic form, which will be derived from the
USPTO’s scrape of the DOCX file, errors and all. We see no plausible theory under
which the USPTO’s “evidentiary PDF” proposal provides any reasonable assurance that
the “evidentiary PDF” will displace that electronic document share, or will in any other
way be treated as “evidentiary” by any other countries’ patent offices, let alone all other.
The U.S. Patent Office has no authority to make promises on behalf of other countries.
Until there’s an international treaty or cross-recognition protocol, during U.S. phase,
patent applicants must operate on the worst-case assumptions to make sure that
international rights are not compromised due to USPTO error. None of the USPTO’s
Federal Register notices or information collection paperwork have even acknowledged
these international implications, let alone offered a reliable path.

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 6
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024

•

We phoned a number of courts and agencies to learn their practices for publication of
their decisions, and we called legal publishers to see what form they receive. All the
information we received (including from the USPTO itself) was the same: tribunals make
their decisions available to publishers as PDFs. The USPTO’s needs for patent
application intake is exactly the same as legal publishers’ need for intake of decisions for
publication (in fact, the USPTO’s are somewhat simpler, because of limits on format of
patent applications, 37 C.F.R. § 1.52). PDF works reliably for all decision publishers.
The USPTO has never suggested that their needs are any different.

•

During 2019 notice-and-comment, the public suggested that “PDF/A” (the “archival”
variant of PDF) would serve the USPTO’s needs better than DOCX. The USPTO’s
“response” was only to evade.

II.

The “evidentiary PDF” and June 2023 guidance proposals assume a counterfactual

The “evidentiary PDF” proposal assumes a counterfactual—that the USPTO accepts all
DOCX files. The USPTO doesn’t. The USPTO claims that the reason DOCX is a useful filing
format is that it’s a “standard”—but that’s one of the representations that the USPTO has made
to you that is true only in a hyper-literal-but-misleading sense. True, ISO approved a standard.
But Microsoft doesn’t implement the standard (see Microsoft’s 900-page list of departures from
the standard (https://msopenspecs.azureedge.net/files/MS-OI29500/%5bMS-OI29500%5d.pdf,
linked in the May 8 handout, page 39). More importantly, the USPTO’s DOCX intake software
doesn’t accept the standard. Instead, the USPTO accepts only a small subset of the ISO
standard. When a patent application uses “standard” features that the USPTO hasn’t
implemented, the USPTO gives an error message and refuses to accept the upload. Files that are
refused (therefore showing that the USPTO's DOCX intake software is incompatible with
“existing reporting and recordkeeping practices,” § 3506(c)(3)(E)) arise from many sources:
•

Many patent applications are written largely by the inventor/client (several tens of
thousands per year). We can’t control what they do, what features they use, or what plugin extensions they use (especially for technical subject matter). Many of the DOCX files
we get from clients use sophisticated Word features. The USPTO hasn’t implemented
many of these features in its DOCX intake software; they generate error messages.

•

Patent attorneys tend to be technologically savvy, which means they’re sophisticated
users of Word and use features that the USPTO doesn’t accept. For these attorneys, the
deficiencies of the USPTO’s DOCX intake software affect 100% of filings. For example,
in my patent applications, I use many bookmarks, cross-references, and the like to ensure
that all the parts stay consistent as I edit a document. These features are important to
prevent substantive errors. It works great when I file in PDF. However, when I try to
upload to the USPTO DOCX system, I get many error messages, and the USPTO refuses
to accept my filing. I have tried “dumbing down” my DOCXs. I found that the
handwork took more than $400 worth of time, and replacing the automatically generated
(and therefore guaranteed-correct) features with hand-typed replacements was the risk of

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 7
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024
error that just isn’t worth $400. Because patent attorneys tend to be technologically
sophisticated, my experience is broadly shared.
•

DOCX files output from Word Perfect, Libre Office, and Google Docs tend to have these
features that the USPTO won’t accept. According to the USPTO’s own estimate, about
20% of filers don’t use Word and are therefore exposed to this risk.

•

The threshold for “too sophisticated for the USPTO’s systems” is pretty low: even
landscape pages for wide tables cause the USPTO’s systems to refuse acceptance (it
works fine in PDF—just turn the page sideways so that all pages in a file are in portrait
layout).

•

The diagnostic messages given by the USPTO’s DOCX intake software are noninformative. The error message is often only a few words, with an inscrutable code—
entirely cryptic. Further, the USPTO’s DOCX intake software never tells where the
offending feature is and never tells what to do about it. Often, it’s some invisible field.
We’ve had reports on our attorney email lists of people spending hours trying to find the
problem. As often as not, the attorney gives up and files in PDF.

•

Ironically, the most disadvantaged patent applications are the most technologically- and
commercially important.

The upshot is that there will always be situations with no viable DOCX path, and PDF filing will
be the only option. At a guess, we estimate this at about 5% where there is no DOCX path, not
even theoretically. We estimate that for 20% of all applications, no DOCX path (even with
“evidentiary PDF”) makes sense on a cost-benefit balance-of-risks analysis. The USPTO cannot
estimate the number of such cases because their existence is hidden—the applicant gives up
trying to file in DOCX and just reverts to PDF. In its 0651-0089 Supporting Statement, the
USPTO told you, “The USPTO is simply not seeing the unreliability asserted by the
commenter.” That’s sample bias, not data. In its Federal Register notices and Supporting
Statements, the USPTO has consistently misreported the error reports from the public, or has
reported them in sentences that are true in only a hyper-literal reading but misleading under a
normal reading. When the USPTO tells you there’s no practical problem, that’s an Information
Quality problem, not data.
III.

The “evidentiary PDF” and June 2023 guidance leave many burdens unaddressed

“Evidentiary PDF” reduces some problems (which tend to be the low-impact ones) and
leaves others untouched (which tend to be the high-impact ones). Almost by definition,
“evidentiary PDF” acknowledges that the USPTO cannot reliably scrape data out of a DOCX.
(In contrast, there are several reliable off-the-shelf software products for scraping text out of
PDFs that the USPTO could use.) We’ve brought that proofreading and error correction burden
to the USPTO’s attention multiple times; they continue to book it at zero. Your email suggests
that the USPTO continues to do so. Making something possible doesn’t make it free or easy.
Here are the various failure points that continue to create proofreading and error correction
burdens:

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 8
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024

•

For users who don’t use Word, the “export to DOCX” feature simply isn’t reliable.
There’s no commercial value for a software vendor to export reliably (import yes, export
no). See the Library of Congress page in our May 8 handout at page 38. The USPTO’s
survey estimated that using non-Word word processors accounts for about 20% of filers.
Our observation is that 20% is concentrated among small entities. This will be a source
of errors. True, the “evidentiary PDF” proposal makes it easier to correct errors. Still, it
does nothing to reduce the proofreading burden to identify those errors and does little to
reduce the burden of preparing the papers to ask the USPTO to make the correction.
Proofreading and error correction are a burden that must be estimated and booked, even
for the “evidentiary PDF” and June 2023 guidance variant proposals.

•

Microsoft updates Word frequently. Microsoft has announced that as the future
mandatory migration path will be Office 365, the Software-as-a-Service offering, which
will be updated even more frequently, with even less notice. When Microsoft changes
Word in a way that adds new content to DOCX files, the USPTO will be caught flatfooted and, for several days or weeks, won’t be able to handle patent applications
correctly. Again, the “evidentiary PDF” proposal is no remedy at all if the USPTO’s
DOCX software won’t even accept the application—or accepts it with errors. The only
alternative is to pay the $400 penalty fee.

•

For users of Word, if the applicant is using one of the features on Microsoft’s 900-page
list of departures from the standard (https://msopenspecs.azureedge.net/files/MSOI29500/%5bMS-OI29500%5d.pdf, linked in the May 8 handout, page 39), there’s no
guarantee that the USPTO can handle it correctly. If the USPTO’s DOCX intake
software refuses to accept an application, neither the “evidentiary PDF” proposal nor the
June 2023 guidance accomplishes anything. Again, the only alternative is to pay the
$400 penalty fee.

•

Often, the attorney prosecuting the application is not the one who wrote the application.
The later attorney may not realize that there are discrepancies between the evidentiary
PDF and the USPTO’s version. So, from a liability point of view, when the later attorney
takes over mid-stream, does that later attorney have to compare the evidentiary PDF to
the USPTO’s record? Is that now a required part of the intake for a transferred case? Is
it something that all later attorneys should do when responding to an Office Action for a
case they didn’t write? Maybe even for ones where there’s no change of attorney?

•

Suppose after a patent has been issued, an attorney now notices a discrepancy between
the evidentiary PDF and the issued patent. What’s the process, and what’s the estimated
burden of resolving the issue? During NPRM, ICR 202011-0651-006, and ICR 2023090651-002, and this ICR, the USPTO estimated this all at zero.

•

Suppose a defendant charged with patent infringement checks the record and finds that
the issued patent does not match the evidentiary PDF. Can the defendant use that
discrepancy to invalidate the patent? Why didn’t the patent owner have a duty to seek
correction? If the USPTO has given this meaningful thought, it hasn’t been disclosed in
any request for public comment.

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 9
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024

•

The recordkeeping to make sure that the PDF and the DOCX reflect identical versions is
genuinely difficult. If the phone rings or there’s some similar distraction while preparing
the final copies for filing, there’s a real risk of introduced error. And over time, there’s
no ready software tool for coupling files to ensure the applicant’s recordkeeping is
accurate. If you need a numerical estimate, I’ll poll my colleagues. I personally catch
myself making analogous errors several times a year. I have checklists and processes to
catch them. But most attorneys don’t have the education in error-resistant engineering
that I do and won’t do things like I do. Errors will get through, and version skew will
create real problems.

•

The USPTO’s DOCX intake software changes the input and introduces errors. That
introduced error is the most likely explanation for the “Example 8 $8000 fee” example
(see May 8 handout, pages 12-27). As any software engineer will tell you, as long as the
USPTO changes its inputs (thereby violating the Federal Records Act and Federal Rules
of Evidence), they’re going to introduce errors. High-reliability software starts with
storing the user’s input exactly as received (see our “requirements spec” attachment).
The “evidentiary PDF” variant won’t change the applicant’s burden to detect and correct
these errors. “Evidentiary PDF” makes it easier to prove the need for correction but does
little to reduce the burden of detecting and filing the paperwork to correct the USPTO’s
errors.

•

A DOCX file has no single meaning—the ISO standard says so (see the May 8 handout at
pages 36-39). Every implementation can do something different with the same file and
claim to be “correct.” So, there’s really no reference standard and no way to prove that
the USPTO’s computers were in “error.” The “evidentiary PDF” proposal provides a
reference standard to calibrate against, but that just creates the burden of proofreading
and filing the paperwork to get the correction made. The USPTO charges different fees
for USPTO-created errors vs. applicant-created errors (35 U.S.C. § 254 vs. § 255). The
applicant will still have to convince the USPTO that their rendering of the DOCX was
“mistake incurred though the fault of the Patent and Trademark Office,” § 254—which
can’t be done because the ISO standard is so loose that there’s no uniform definition of
“correct.” In Example 8 of our May 8 handout, you can see the overwhelming burden of
proof the USPTO puts on applicants who attempt to show, when fees are at issue, that the
error was the USPTO’s.

•

Every computer professional knows the acronym GIGO—garbage in, garbage out.
Unreliable input guarantees unreliable output. It’s like a squishy toy—the USPTO can
move the error from one place to another, but the USPTO can never get unreliability and
errors out of a system if the USPTO starts with unreliable input.

As you note, after the patent issues, what the USPTO calls the “evidentiary PDF” will be
available as a reference standard for error correction during litigation. But the errors are still
there. Because the USPTO transforms the applicant's original PDF file, there may also be errors
in the evidentiary PDF. And litigators will litigate. It’s hard to estimate the burden for a lowprobability-high-severity event, but having been a litigator earlier in my career, I can say with

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 10
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024
great certainty that the cost of even one of these fights will be greater than the annual savings
that the USPTO claims for itself.
The “evidentiary PDF” is at best a necessary element in establishing a discrepancy
between the PDF and the DOCX. But it is not a sufficient condition, because it keeps the entire
burden on the applicant to do the proofreading and to request correction. The USPTO has not
yet told us what will be required in a grantable petition to correct (i.e., how burdensome it will be
to seek correction), and the USPTO also has not committed to the ultimate effect of a grantable
petition—and lacks authority to do so. The relevant statute is absolutely clear that correction
will only have “thereafter arising” go-forward effect, and does not correct errors back to the time
the USPTO created the error. How can the USPTO guarantee that corrections will be binding
immediately for all purposes, including pre-grant publication, examination, grant, and all
downstream and derivative records and tools? Simple—the USPTO can’t. That leaves the
burden on the applicant to carefully proofread, and to call for corrections at every step.
The bottom line is that the USPTO is using the evidentiary PDF as a camouflage for the
inherent unreliability of DOCX. The evidentiary PDF variant is only a diversion from the core
issue—DOCX’s unreliability—and instead burdens applicants with a significant amount of
additional work during and after intra-USPTO proceedings.
IV.

The copy that the USPTO proposes to designate as the “evidentiary PDF” is not the
original—the USPTO’s implementation of the “evidentiary PDF” increases burden

The USPTO accepts an uploaded auxiliary PDF—but immediately transforms it by
flattening it to a halftoned bitmap, throwing away the full-text form. (a) That makes the
evidentiary PDF really difficult to proofread—it can only be proofread by eye, not by machine.
(b) It makes it impossible to do a fingerprint (hash-value) check to confirm that the document
one downloads later is the same as the document uploaded. (c) It violates the Federal Records
Act.
The USPTO might contend that the evidentiary PDF maintains the same content as the
applicant's original PDF, with the only difference being the format (the evidentiary PDF being an
unreadable version of the original). However, this assumption demands that applicants trust that
no other transformations have occurred and that the flattening process did not introduce errors.
Due to this transformation, applicants are compelled to verify the evidentiary PDF against their
original PDF file, imposing a significant additional burden.
The “evidentiary PDF” proposal and June 2023 guidance both mean that the applicant
has two proofreading burdens, to proofread the USPTO’s mangled DOCX and the flattened PDF.
And the final proofread, of the issued patent against the flattened bitmap PDF, must be by eye,
not by computer, because the USPTO threw away the text-based form.

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 11
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024
V.

For many patent applications, proofreading during intra-PTO pendency is
impossible

For about 15% of applications, while the application is pending, it’s absolutely
impossible to get a meaningful copy of what the USPTO thinks it received. 35 U.S.C.
§ 122(b)(2)(B)(i) permits an applicant to request that applications not publish before they are
issued. For those applications, the USPTO makes nothing available that reflects what the
USPTO thinks it received. The USPTO makes available an XML file, but that XML has so
much junk in it that it’s not useful in any practical sense for proofreading.
For these applications, the examiner will be looking at one thing (and potentially raising
rejections), and the applicant will be looking at a different thing. There’s already plenty of
talking past each other because of things one knows and the other doesn’t; this will be a total
train wreck.
The “evidentiary PDF” and June 2023 guidance variants offer no reduction to this
problem.
VI.

The USPTO’s only estimates for cost savings concede that the cost shift is
disproportionate

The only estimate the USPTO has ever given for its cost savings is $3.15 per application
(in the NPRM). Multiplied out by the number of responses, the USPTO’s cost savings are well
under $1 million per year.
In contrast, under the “evidentiary PDF” and June 2023 variants, patent applications will
still require careful proofreading. The public’s estimate for the proofreading, to achieve a ceteris
paribus comparison to the status quo ante reliability of PDF filing, are in the ICR 202309-0651002 comment letters ($200 million to $600 million). The USPTO has never challenged either
these estimates or any of the underlying methodology.
VII.

The PTO has a legitimate need for full-text filings—most applicants file them today
as PDFs

We fully agree with the USPTO that there’s a legitimate problem that needs a legitimate
solution. The USPTO needs a reliable intake of full-text documents. The issue is whether the
USPTO should do what every other agency and legal publisher does—accept full-text PDFs,
under a voluntary consensus standard—or whether the USPTO should build its own software to
accept a highly irregular, poorly controlled form, DOCX. We have suggested several times that
the documents that applicant file today—full-text PDFs—should be a near-complete solution to
the USPTO’s needs, if the USPTO would simply stop flattening them to bitmaps. We have also
suggested that PDF/A—the “archival” variant of PDF—is a complete solution to both the PTO’s
needs and the public’s. The USPTO has never disputed either conclusion or any of the
underlying assumptions.

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 12
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024
We have no doubt that the USPTO’s sunk costs for its DOCX intake software are very
large. DOCX is a fantastically complicated way to do a task that’s much simpler in PDF, as the
USPTO’s own contractor found (see the May 8 handout at pages 29-32). The DOCX standard is
several times the size of the PDF standard (I uploaded both DOCX standards at
https://www.reginfo.gov/public/do/PRAViewDocument?ref_nbr=202309-0651-002 PDF is
designed for precisely the purpose the USPTO wants, PDF is the form that every court—and the
USPTO itself—uses to publish their decisions, and all the legal publishers have reliable PDF text
intakes. We have no direct insight into the USPTO’s software internals, only our intuitions as
software engineers (one trained at Michigan and MIT, one with a PhD in computer science from
NYU, another with a PhD in physics, another with a software engineering career before
switching to law) based on external observation. But we can tell a lot as outsiders—Congress
didn’t have to have on-site inspectors to recognize that Boeing’s engineering wasn’t up to snuff.
We are as sure as we can be that if the USPTO replaces its DOCX intake software with an intake
from text-based PDF, the USPTO can switch to a solution that largely relies on commercially
available or open-source software components and services. We are reasonably sure that the
USPTO’s costs will be better two years from now if they give up the DOCX wild goose chase
and do this job the same way every other agency does. Our understanding of the law is that it’s
the USPTO’s burden to show otherwise—we’ve raised this issue repeatedly, and the USPTO’s
response to comments has either mischaracterized the comment or answered with a non sequitur.
The USPTO claims to have relied on “internal IT considerations and initiatives, external
sources, and industry practices,” but they’ve produced none of it publicly. For every document
we’ve obtained by FOIA, the USPTO’s characterization has been either materially misleading or
outright false. The law requires OMB to withhold clearance until the USPTO puts its documents
up for public vetting as a “supporting record.” § 3506(c)(3). The USPTO has refused to do so
on multiple occasions.
This is an information collection with a really low impact on the agency, under $1 million
per year. And now they’re asking you to approve nine figures of burden in an ICR they didn’t
publish for comment, based on hidden documents, after years of avoiding fair response to public
comments. As far as we can tell, their only argument is sunk costs. Please don’t reward
misbehavior.
VIII. Further rulemaking and information collection procedure is required
The USPTO has never proposed the “evidentiary PDF” or June 2023 proposal for
comment. In fact, the USPTO’s public statements have been to the contrary. For example, in
June 2021 (after the conclusion of the last triennial review of 0651-0032 in May 2021), the
USPTO stated that the ability to correct would only last one year, Patent and Trademark Office,
Submitting Patent Applications in Structured Text Format and Reliance on the Text Version as
the Source or Evidentiary Copy, 86 Fed. Reg. 29571, 29571 col. 3 to 29572 (Jun. 2, 2021). We
see no change to that one-year limit on the right to seek correction. The evidentiary PDF doesn’t
do any good if the right to seek correction lasts only one year.

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 13
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
June 4, 2024
IX.

Conclusion

This ICR should not approve the burden the USPTO has never put up for public comment
and didn’t request now. This is a nine-figure rule. It ought not be negotiated in a secret overtime
period by a half-dozen people just because the USPTO skipped out on the required procedure
during regulation play. If the USPTO wants a clearance, please ask them to do what every other
agency does—start at the beginning with a 60-day notice that gives honest burden estimates that
cover the full spectrum of burden and a fully disclosed supporting record that allows public
vetting and a Supporting Statement that gives candid answers to the public comment letters, and
support everything with a record that meets Information Quality guidelines. This is not a big
deal for the agency—they delayed implementation for three years, and now the cost for them to
accept the worst form of PDF is under $1 million annually. We conjecture that the USPTO will
derive all the benefits and impose none of the costs if they switch to text-form PDF as the
submission form, perhaps even PDF/A (the archival form).
A § 1320.12(f) rescission would be appropriate. The $400 penalty fee should be
rescinded so applicants can file in any form they find convenient.
We may be reached at the address below, if you have any questions.
Very truly yours,
PTAAARMIGAN

Very truly yours,

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

Exhibit A

A “requirements specification” given to USPTO
staff in January 2023

EXHIBIT A

Agenda for DOCX discussion
Correctness first. Every computer engineering student is taught that correctness may (nearly)
never be compromised for “efficiency” or cost.
a. This is especially true here—the law of new matter makes post-filing correction of errors nearly
impossible. A patent application or patent could become worthless because of a textual change
(no matter how small) introduced by the filing system.

1. Principle 1: The file that the applicant uploads must be the authoritative document.
a. No wiggle room. No changes, no “validation” if that means change or replacement. The
applicant’s uploaded bits are authoritative, and must be retained indefinitely.
b. These are legal documents, evidence for litigation. The best evidence rule applies.
c. Humans are not well-equipped to review for small changes—any scheme that changes the input
in subtle ways and asks for human review is an inherently flawed design.
d. Inevitably, the best-engineered system will still run into errors. When the same file is uploaded
months or years later (either during intra-PTO examination or during litigation), the upload must
give exactly the same checksum. No modification on the fly.
e. The system must be designed to minimize the path between file upload and authoritative
archival storage, in order to minimize the chance for screw-up.

2. Principle 2: What You See Is What I Get. The file specification (perhaps ISO standard)
for the authoritative file must guarantee that anyone, anywhere, any time, that opens or prints
the authoritative document will see the same rendering in content and visual layout.
a. Litigation parties must be left with no ambiguity about exactly what was filed—small
ambiguities will become big litigation issues.
b. DOCX documents appear different when opened on:
i. Microsoft Word, Libre Office, Google Docs, Mac Pages, the PTO’s DOCX filing system,
and the PTO’s FOIA system.
ii. English text written in non-Latin-alphabet editions of Word (Word for Israel, Japan, etc.)
screws up when opened in the American edition of Word.
iii. Microsoft Word for Windows, Word for Mac, different releases of Word, the same version
of Word on computers with different plugins and other software installed
c. The PTO estimated that 20% of users use word processors that don’t generate DOCX

3. Standards matter—but only if the guarantees of the standard match up to the correctness needs
of the application.
a. Principle 2 implies that a usefully-applicable standard is one that is designed to ensure
portability and uniform reproducibility. A standard that is designed to adapt the document to
each computer it’s opened on, that has a great many “implementation defined” parameters, or
that is designed to encourage extensions and plug-ins, is not a useful standard for this purpose.
b. If the PTO justifies its choice based on a “standard,” the PTO must implement the entire
standard. Standards are two-sided agreements. The PTO can’t implement part of a standard
that’s convenient for the PTO, and forbid the part that’s convenient for the public. The PTO can
enforce valid regulations, e.g., 37 C.F.R. § 1.52(b)(2) (nonscript type fonts, single column, etc.),
but may not rule out “standard” features just because they’re hard to implement.

Within those three principles, all options are on the table.

4. Legal requirements. Any implementation must meet the requirements of the Paperwork
Reduction Act. Rulemaking must observe the procedures of the Administrative Procedure Act,
Paperwork Reduction Act, Regulatory Flexibility Act, and Executive Order 12866.
5. For preparation, it will be helpful to review three of the public comment letters. All three
explain technological flaws in the PTO’s DOCX proposal; the Oppedahl letter is most detailed. The
Seventy-Three letter and Boundy letter explain legal process errors.


Carl Oppedahl, Comment letter (Aug. 12, 2019)—this letter explains several attempts to file
patent applications using the PTO’s DOCX filing system. In the Final Rule notice in the Federal
Register, the PTO stated “To date, the Office has not received notifications of any issues
resulting from the filing of applications in DOCX format.” This representation was submitted to
the Office of Management and Budget and to the Small Business Administration for their ex
parte proceedings to review the rule just before publication in the Federal Register.



Seventy-Three Patent Practitioners, Comment letter (Sep. 27, 2019) explains technological
failures, costs, and legal process failures.



David Boundy, comment letter (Nov. 30, 2020), presents two cost workups for the DOCX rule.
Both total about $200 million per year. (The PTO was required to generate similar estimates,
under Executive Order 12866, the Paperwork Reduction Act, and the Regulatory Flexibility
Act—to date, the PTO has not done so for any of the three.) In the PTO’s response to OMB of
May 25, 2021, the PTO conceded that it had not observed the procedural requirements of the
Paperwork Reduction Act, and has no plans to commence compliance.

6. A way forward: Standard ISO 14289-1:2014 appears to be a good starting point for further
discussion. It’s popularly known as “PDF/UA” for “universal accessibility.” To our knowledge,
PDF/UA satisfies all criteria for all parties:
a. The document text is presented in logical reading order. The document’s logical structure and
organization is organized into sections, paragraphs, lists, tables and so on.
b. Problematic content is prohibited, including illogical headings, the use of color/contrast to
convey information, inaccessible JavaScript, etc.
c. Our initial investigations suggest that PDF/UA meets Principle 1. PDF/UA has been a feature of
free and low-cost PDF writers for many years. Microsoft included an accessible PDF writer in
Word 2016.
d. PDF/UA has the portability, and identical appearance guarantee of PDF, thereby satisfying our
Principle 2.
e. It’s an ISO Standard, an extension layered on top of PDF 1.7. The U.S. Government Access
Board (the agency that implements the Rehabilitation Act) recommends PDF/UA-1 as one of
several “appropriate standard[s]” for full accessibility. We conjecture that the PTO’s needs are
largely coextensive with this accessibility and archiving standard.
f. We conjecture that the PTO could contract with the vendors of free or low-cost PDF writers to
ensure that their output meets any criteria the PTO needs.

Standard 19005 for PDF/A, for "archival", may have desirable properties as well.
Public comment letters suggested both PDF/A and PDF/UA. The PTO has never
published any document that suggests bona fide consideration of either alternative.

PTAAARMIGAN
Patent and Trademark Attorneys, Agents and Applicants for Restoration and
Maintenance of Integrity in Government
MAILING:

P.O. BOX 590372, NEWTON MA 02459

E-MAIL:

[email protected]

September 5, 2024
Via Email
Nicholas A. Fraser
Office of Information and Regulatory Affairs
Office of Management and Budget
725 17th St. NW
Washington, D.C.
Re:

ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day
notice at 89 Fed. Reg. 21495 (Mar. 28, 2024)

Dear Mr. Fraser—
This is a follow-up to earlier letters about a proposal by the U.S. Patent and Trademark
Office to penalize those who file patent applications in PDF form, and instead require filing of
patent applications in DOCX, with a $400 penalty for PDF. Three new facts are relevant.
We understand that the USPTO no longer seeks approval of the information collection as
originally proposed in the 2019 NPRM, 2020 Final Rule, or January 2024 60-day notice. We
understand that the USPTO has modified its ask, to treat the “auxiliary PDF” as the evidentiary or
authoritative copy. The USPTO has never requested comment on that variant, and this letter
raises new facts. PTAAARMIGAN therefore requests that you consider these comments, even
though submitted outside the regular order.
•

•

•

We understand the current state of play to be as follows:
The USPTO has never published anything in the Federal Register—let alone a
rulemaking—to implement the “evidentiary PDF” proposal (perhaps a sub silentio
confession that the USPTO knows it has no such authority, see the Loper Bright
discussion below).
The USPTO’s last definitive statement in the Federal Register on the “authoritative” or
“evidentiary” copy is “the USPTO considers the validated DOCX file(s) submitted by the
applicant to be the authoritative document and that applicants may rely on the validated
DOCX file(s) as the source or evidentiary copy of the application to make any corrections
to the documents in the application file.” Filing Patent Applications in DOCX Format,
Notice, 87 Fed. Reg. 25226, 25227 (Apr. 28, 2022). The USPTO has never stated in the
Federal Register that the PDF would be “authoritative” or “evidentiary.”
The most recent Federal Register notice, Extension of the Option for Submission of a PDF
with a Patent Application Filed in DOCX Format, 88 Fed. Reg. 37036 (June 6, 2023)
Patent and Trademark Attorneys, Agents and Applicants for Restoration and Maintenance of Integrity in Government
PTAAARMIGAN LLC | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 2 of 7
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
September 5, 2024

•

observes none of the requirements for rulemaking or information collection, and makes
none of the relevant commitments. The June 2023 notice states that the USPTO is
“extending the option” to file an auxiliary PDF, but the USPTO pointedly avoids any
promise about what the USPTO will do with that auxiliary PDF, and avoids any promise
of legal effect. The USPTO has never promised in the Federal Register that the auxiliary
PDF will be treated as “authoritative” or “evidentiary,” never promised to attribute any
legal significance to it (beyond mere “option to submit”), never promised any legal effect
extending throughout the 26-year life of a patent (including post-issue). The USPTO only
says that the “option” to file will exist “until further notice”—a commitment to nothing.
The USPTO’s most recent statement on retention of the PDF was a commitment to
destroy the PDF three years after a patent is granted. Filing Patent Applications in DOCX
Format, 87 Fed. Reg. 25226, 25227 (Fed. Reg. Apr. 28, 2023). But the economicallyimportant part of a patent’s life is post-issue licensing and enforcement until the patent
expires, plus the six year statute of limitations. Patent owners require that the USPTO
preserve all federal records, to ensure they’re available to correct USPTO mistakes for 26
years from filing.1 The June 2023 Notice only makes a short-term promise, with no
permanent commitment to retain the PDF, and no commitment to any legal status for the
auxiliary PDF.

With that background, here are the three new facts that have arisen since the close of the ICR
202304-0651-001 comment period, and how they apply to the “evidentiary PDF” proposal.
1. Loper Bright. For the “evidentiary PDF” proposal to provide any meaningful effect to
patent owners, the USPTO would have to bind courts to credit secondary evidence (the auxiliary
PDF) over the issued patent. But as far as we know, the USPTO has no such authority. Without
that authority, the “evidentiary PDF” proposal solves nothing.
Before the Supreme Court’s recent decision in Loper Bright Enterprises v. Raimondo, 144
S. Ct. 2244 (2024), if an agency made a plausible claim to have authority for a rule, courts would
defer. After Loper Bright, that is no longer true. The USPTO’s main rulemaking statute, 35
U.S.C. § 2(b)(2), “does not authorize the Patent Office to issue ‘substantive’ rules.” Cooper
Technologies Co. v. Dudas, 536 F. 3d 1330, 1336 (Fed. Cir. 2008). We know of no statute—§ 2
or any other—that would authorize the USPTO to govern the post-issue life of patents. We know
of no plausible statutory basis for the USPTO to require courts to place “authoritative” or
“evidentiary” weight on the “auxiliary PDF,” even if the USPTO acted by full regulation, rather
than guidance.
For decades, it’s been the opposite. In the event of any ambiguity, typo, or similar error,
courts put the risk of error on the patent owner, and construe against the drafter. “As courts have
recognized since the requirement that one's invention be distinctly claimed became part of the
patent law in 1870, [courts have construed patent documents] to guard against unreasonable

1

This is a case showing why APA notice and comment and § 1320.9 comment periods are so
essential—if the USPTO had followed the requisite procedures, it would have avoided a mistake that
brings into question the agency’s subject matter competence and respect for the rule of law.
PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 3 of 7
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
September 5, 2024
advantages to the patentee and disadvantages to others arising from uncertainty as to their
[respective] rights.” Athletic Alternatives, Inc. v. Prince Mfg., Inc., 73 F. 3d 1573, 1581 (Fed.
Cir. 1996). In another recent example, Sipco, LLC v. Jasco Products Co., Case No. CIV-19-709PRW (W.D. Okla. 2024), the USPTO examiner required the applicant to stipulate to a limit on
enforceability of the patent. But the examiner’s proposal had a typo—a number was off by one.
The applicant adopted the examiner’s suggestion, verbatim. Nonetheless, the Sipco court bound
the patent owner to the examiner’s mistake—even though the correct number was entirely
obvious in context—and rendered the patent unenforceable in its entirety.
We know of no law that would give a patent owner any lenity for a USPTO mistake in
rendering a DOCX—the most likely outcome is that the cost of USPTO error will fall entirely on
the patent owner. Courts recognize that patent applicants control the record, and competitors
have to make decisions based on that record, so as between them, the consequence of the error
must fall on the patent owner. Courts do not require competitors to identify and read past typos
or similar errors in patents. Courts have been very clear that it is the patent applicant’s burden to
provide requisite notice, so typos, ambiguities, and disagreements between two documents (the
DOCX from which the USPTO prepares the issued patent vs. the auxiliary PDF, for example)
will (in most cases) be resolved against the patent owner. We know of no basis on which the
USPTO would have authority to tip that balance the other way in favor of some “evidentiary
copy” that disagrees with the patent document itself. Loper Bright virtually eliminates any
possible argument the USPTO might have.
Until it’s tested in court, the USPTO’s “evidentiary PDF” proposal reduces burden by
essentially zero. Even under the “evidentiary PDF” proposal, every patent applicant that files in
DOCX must do the proofreading and error correction described in the public comment letters.
Burden estimates for proofreading and error correction are basically unaffected by the “auxiliary
PDF” proposal, as explained in the public comment letters for ICRs 202011-0651-006 and
202309-0651-002 (each of these estimates assumes a ceteris paribus baseline where reliability
and enforceability of future DOCX patents parallels past PDF patents):
• David Boundy, objectID=107472802 at pages 4-5; estimates total $200 million/yr.
• Brad Forrest, objectID=135639300 at pages 99-107; estimates total $200 million/yr.
• Carl Oppedahl on behalf of 107 co-signatories, objectID=135640900 at page 9; estimates
total $600 million/yr.
• Dr. Cathy Cretsinger, objectID=136718302, at pages 7-8; estimates total $600 million/yr.
To date, the USPTO has not responded to these estimates in either an NPRM or ICR comment
period, except to (falsely) state that DOCX is as reliable as PDF. That contention is false, as new
fact 2 demonstrates.
2. New update to Microsoft’s definition of DOCX. Two weeks ago, Microsoft
published an update to its definition of DOCX format, describing “extensions.”
https://learn.microsoft.com/en-us/openspecs/office_standards/ms-docx/b839fe1f-e1ca-4fa6-8c265954d0abbccd “Extensions” in a software product are new features above the features required
by a standard. By definition, extensions are software behaviors that are outside the standard. By
definition, software that uses the extension will generate output data that can’t be read as input by
PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 4 of 7
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
September 5, 2024
other software (such as the USPTO’s DOCX intake software). Extensions are used strategically
by vendors with monopoly power—extensions make it nearly impossible for customers to
migrate to competing products, and create constant costs for competitors who must scramble to
catch up.
Microsoft’s August update to the “extensions” document is characterized as “major” and
“Significantly changed the technical content”—dozens of other changes since 2009 have also
been “major.” Microsoft introduces these extensions into Office unpredictably (for example,
Microsoft updates Office 365 on the fly with no notice to the public). Unless the USPTO has
somehow gotten on the advance notice list so that the USPTO can update its DOCX intake
software before Microsoft releases them, it’s inevitable that the USPTO’s DOCX intake software
will fall behind Microsoft’s. That will translate into errors when the USPTO encounters patent
applications created with updated versions of Word. Further, a change to an extension could
mean that a DOCX file that renders one way today will render differently a year from now. In
litigation, where differences are tested, that unreliability of the evidentiary record will create
chaos.
Microsoft’s “extensions” document proves our point—DOCX is a foundation of shifting
sand, not a voluntary consensus, uniform, reliable, or unvarying interoperability standard for
reliable exchange of documents between different computers. DOCX is not designed to, and
makes no promises that it will, guarantee uniform meaning of a data file for the 26 year life of a
patent. The USPTO claims in its 0651-0089 Supporting Statement, “DOCX presents documents,
including text and formatting, in a manner that is independent of software, hardware, or operating
system” (Responses 2 and 7). The USPTO has never provided any “record supporting” that
claim, 44 U.S.C. § 3506(c)(3), has never explained away the contrary statements in the standard
document itself2, and the multiple software errors the public comment letters. This “extensions”
document proves that Microsoft makes no attempt to preserve uniformity and consistency across
“software, hardware, or operating system” or time. The extensions document shows that the
USPTO’s claims in its Supporting Statements are simply false.
3. New empirical experience with DOCX filing. A number of patent attorneys use nonDOCX word processors in their day-to-day work—Word Perfect, Libre Office, Word 2003,
Hebrew- and Asian-character versions of Word, Apple pages, and others. (The USPTO estimated
this proportion at nearly 20%. 85 Fed. Reg. at 46947, col. 2). Since the $400 penalty went into
effect in January, those filers have gathered further empirical evidence that DOCX files generated
by non-DOCX word processors simply don’t work when submitted to the USPTO, in violation of
the USPTO’s obligation to provide systems “consistent and compatible, to the maximum extent
practicable, with the existing reporting and recordkeeping practices” (§ 3506(c)(3)(E)). Of the

A number of “implementation dependent” features that would prevent reliable exchange of
documents between different hardware, software, and operating systems were given in the very first round
of comment letters, Seventy Three Patent Practitioners, at pages 14-19, and several subsequent letters.
The USPTO has never responded to this comment or acknowledged the existence of the 100-or-so
“implementation defined” features in the DOCX standard document.
2

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 5 of 7
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
September 5, 2024
many fact patterns discussed on the attorney email lists over the last two months, some are
particularly anomalous:
•

A DOCX exported from a non-Microsoft word processor sometimes does not meet the
USPTO’s requirements for form—in the last two months, users have reported that the
USPTO’s DOCX intake system gives a warning of faulty line spacing, font, and the like.
o First, some word processors export DOCX files with invisible control codes or other
“stuff” in them. These files are accepted by Word (so it seems to be correct and
within the standard), but draws an error message from the USPTO’s DOCX intake
software. Sometimes the USPTO error message gives a partial diagnosis (“Remove
content control codes”), but no location. Sometimes the USPTO error messages just
notes that the application has some feature that the USPTO doesn’t know how to
handle—but doesn’t identify specifically what the problem is, or where it’s located.
So these exported DOCX files—which appear to be correct—can’t be filed as patent
applications.3
o Second, some of the “requirements” enforced by the USPTO’s software exist only in
subregulatory guidance—they were never adopted with rulemaking formality, and
shouldn’t be enforced in any event. But the USPTO enforces them anyway.
o Third, the USPTO’s DOCX software apparently has bugs—to all appearances, the file
as observable on the user’s computer complies with the USPTO’s requirements, so
when the software reports a spurious “warning” or displays a rendering that differs,
there’s nothing that can be done to respond to an erroneous diagnosis
The “auxiliary PDF” proposal offers essentially no reduction in burden to those who use a
non-DOCX word processor—it makes the showings easier to correct USPTO error, but
offers no reduction in burden to detect or correct them. (The USPTO has pointedly
refused to book proofreading or error correction burden at anything above zero, so even if
there were some reduction, the USPTO can’t claim that “auxiliary PDF” reduces it.)

•

After any of the errors of the preceding bullet, when the USPTO sends a “notice to file
corrected application papers,” the applicant then refiles identically the same PDF a week
later. On second submission, the identical PDF document is now accepted. This is
“unnecessarily duplicative.” § 3506(c)(3)(B). If the auxiliary PDF is “evidentiary,” then
it should be “evidentiary” for all purposes.

•

Some patent application filings must include (a) an exact copy of an earlier-filed
application (to establish the right to the filing date of that earlier application), (b) a
“preliminary amendment” (to correct typos, idiomatic translations, substantive changes to
the claims, etc.), and (c) a substitute specification that engrosses the amendments into the

As earlier comment letters noted, the USPTO’s DOCX intake software doesn’t accept the
DOCX standard. It accepts only a cramped and poorly-defined subset. Applicants can only determine
what is acceptable on a hit-and-miss basis. The USPTO can’t claim that its proposal is justified because
“DOCX is a standard” and then not implement the full standard—the USPTO’s “heads I win, tails you
lose” approach to standards is not just a violation of Circular A-119, it’s just plain dishonest.
3

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 6 of 7
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
September 5, 2024
earlier-filed application. The USPTO’s application submission software only accepts this
substitute specification in PDF—no set of checkboxes allows submission of a substitute
specification in DOCX format. Even though the application as filed is in DOCX (as
provided by the regulation), and the USPTO’s software only permits PDF filing, the
USPTO nonetheless imposes the $400 fee because the substitute specification is in PDF
instead of DOCX. That’s not what the text of the regulation says.
The USPTO issued a “clarification” guidance document in May—but (i) it’s only a
guidance document, not an amendment to the regulation, (ii) the “clarification” changes
the rules, it doesn’t clarify ambiguity in the regulation (applications and preliminary
amendments are in two entirely different things under different control numbers) (iii) the
USPTO did not include this guidance document in its request for clearance, (iv) the
USPTO enforces this guidance document retroactively, and (v) the USPTO suggests filing
the preliminary amendment at least a day later (which adds about a half hour of
administrative time).
We know of no good-faith explanation for the USPTO’s imposition of the fee outside the
situations covered by the regulation.
•

The added complexity of filing two copies of the same document (the DOCX and
auxiliary PDF) increases the risk of error. This isn’t a theoretical risk—it has actually
happened. On July 25, an in-house patent professional asked the attorney email lists for
advice for how to correct an error—outside counsel had filed a DOCX and PDF that
mismatched. Complexity breeds error. The “auxiliary PDF” proposal doesn’t reduce
burden (as discussed above), it just creates new problems.

In past Supporting Statements, the USPTO claims “The USPTO is simply not seeing the
unreliability asserted by the commenter.” That’s sample bias, not data. When an upload fails, the
user abandons the attempt, and tries something else. The USPTO won’t “see” that failure. The
USPTO’s claimed inability to “see” gives zero information about the existence of the software
problem, but a good deal of information about the USPTO’s commitment to information quality.
Public comment letters acknowledged the USPTO’s legitimate need—patent documents
should be filed in a form that has robust full text. PDF/A as the one and only filing format meets
the needs of both the USPTO and the public, and avoids all the problems of the “auxiliary PDF”
variant proposal. “A” in “PDF/A” is for “archival.” PDF/A is a standardized variant of PDF
(ISO 19005, currently 19005-4), that limits the features that can be used in a PDF to only those
suitable for long-term archiving. External references, hotlinks, etc. are excluded, and all
information necessary to rendering the content and appearance must be embedded in the file.
(DOCX has no such variant—every DOCX file depends on its surrounding environment, at the
very least for fonts, rules for geographic conventions for dates and punctuation, etc., and the
DOCX standards documents outline about 100 features that are “implementation dependent,”
such that uniformity cannot be guaranteed.) The USPTO has never responded to public
comments suggesting PDF/A—except to misparaphrase the comments in order to evade a direct
answer. The USPTO’s public responses on the issue suggest that the USPTO’s mind is so closed
that no consideration or investigation of PDF/A has ever been initiated. The USPTO has
certainly never given any explanation that reflects a moment’s fact-gathering or thought.

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459

PTAAARMIGAN
Page 7 of 7
ICR 202304-0651-001 (control number 0651-0032), Initial Patent Applications, 30-day notice at
89 Fed. Reg. 21495 (Mar. 28, 2024)
September 5, 2024
The USPTO receives several times more post-application filings than initial
applications—and these will continue to be received in PDF. If the USPTO builds robust intake
software to intake PDF/A for all phases of a patent’s life, that solves the USPTO’s document
receipt problem for the entire life of a patent application, not just initial application filings. Based
on what we know as software engineers, we are reasonably sure that PDF/A intake software will
be simpler and more robust than DOCX, and will reduce the USPTO’s costs for a far-larger
proportion of the USPTO’s information collections.
Evasion of rulemaking formalities. The USPTO has never requested public comment
on any “evidentiary PDF” proposal—it’s simply been announced as fait accompli—and it
changes month to month. The silences in the USPTO’s notices support an inference that the
USPTO is intentionally evading the procedures that govern rulemaking, perhaps in the hope that
the USPTO’s silence can paper over the technological impossibility of the USPTO’s DOCX
proposal. The best explanations of the technical issues and pattern of evading rulemaking
procedure in past comment letters include these:
• Letter of 152 Patent Practitioners, objectID=135639300—technological mismatches
between DOCX and practical needs of the public at pages 2-12. These technological
misstatements can only persist because of the USPTO’s procedural failures to offer a
§ 3506(c)(3) “record,” failure to genuinely respond to comments, and failure to observe
information quality principles of “reproducibility” and “objectivity.”
• Letter of Carl Oppedahl, objectID=135640200—explaining the origins and evolution of
the DOCX standard within standards organizations, and Microsoft’s subsequent evolution
of Word. The USPTO’s claim that “DOCX is a standard” may be true, but it’s irrelevant
if the standard doesn’t purport to require the properties relevant to solving the problem.
• Letter of 123 Patent Professionals, objectID=136723102—cataloging the USPTO’s
procedural shortcutting, evasions, and falsified certifications.
We hope that the USPTO will withdraw the DOCX proposal voluntarily. If not, we urge
that OMB to require the USPTO to provide a careful and complete response to all these
comments. The seven letters bulleted in this letter demonstrate that the USPTO’s say-so cannot
be taken at face value--OMB should strictly require a § 3506(c)(3) “record supporting” all
contentions. We believe that no such response is possible, and we suggest that the appropriate
disposition of the DOCX information collection (37 C.F.R. § 1.16(u)) is a § 1320.12(f)
disapproval and order to rescind.
We may be reached at the address below, if you have any questions.
Very truly yours,
PTAAARMIGAN
David E. Boundy

PTAAARMIGAN | [email protected] | P.O. Box 590372, Newton, MA 02459