A modest proposal

October 29, 2004

Summary: USPTO is urged in the strongest terms immediately to set up a web-based system that would allow a filer to e-file, through a web page, a PDF or TIF copy of what the filer would otherwise have faxed to the central fax number at 703-872-9306. Such a system would be extremely popular with filers on its own terms. Such a system would permit a low-stakes accumulation of experience with PDFs that is not easy to obtain any other way. Such a system permits getting "a toe in the water" for the USPTO regarding web-based filing of PDFs and TIFs and will give objective data as to adoption rates of web-based PDF and TIF filings.

Background and statement of the problem: The USPTO is at a crossroads, trying to figure out what to do differently, going forward, so that eventually most filers will e-file their US patent applications rather than paper-filing. Should the USPTO spend more money on ePave, or would such spending amount to throwing good money after bad? Should the USPTO start spending money on outside contractors developing a web-based system for patent e-filing? Should the USPTO start accepting PDF-format files from filers, or should it stick with XML-format patent applications?

Related to this is the problem that the USPTO does not have unlimited money to spend. Money spent on a particular future e-filing experiment is money that will have been taken away from something else. Guessing wrong about an e-filing approach might simply mean money wasted and time lost. Guessing right or wrong about an e-filing approach might make the difference between reaching 70% e-filing in six months or not having reached it even after three years or more.

Two particular questions, if only they could be answered inexpensively and quickly, would make an enormous difference as the USPTO tries to figure out how to proceed:

  1. Are PDFs fraught with risk and uncertainty? USPTO and other patent offices have occasionally run into situations where a particular PDF file doesn't seem to print out at the patent office the way it printed out for the author. Sometimes this is due to the author's (perhaps unknowingly) using some exotic font that is installed on the author's computer but is not installed on the computer at the patent office. Other times, despite all efforts, it does not seem possible to learn what the cause is of non-identical rendering at the author and at the patent office. While these situations are rare (well under 1% of all received PDF files), they prompt concern and anxiety on the part of the patent office as to whether it is a good idea to permit filers to file by submitting PDF patent applications. A related question is whether filers will tolerate the use of multipage TIF files rather than PDF files as a way of avoiding the PDF issues.
  2. If we build it, will they come? Many focus group respondents and list server participants have said that if only they could file PDFs, they would convert over their paper patent filings to electronic. But are their statements predictive as to the future actions of filers generally? Can we even count on the focus group respondents and list server participants to do as they say they will do? In this context, it has been noted that EPO has permitted PDF patent filings for over a year now, and yet the EPO e-filing percentage is still only a few percent, nowhere near the 70% or so that USPTO would like to reach in the near future. A related question is whether providing a web-based interface (as distinguished from a "thick client" such as ePave) will really make a big difference for filers, or whether it will turn out to make little difference in the behavior of filers.

Given these unanswered questions, it would be extremely helpful if USPTO could receive a truly wide variety and large number of PDF files from filers under some low-stakes circumstances (not the filing of high-stakes patent applications on critical dates). This would permit, in very quick order, accumulating substantial experience as to what, if anything, ever goes wrong with PDFs.

It would likewise be extremely helpful if some simple, uncomplicated, inexpensive web page could be set up that would permit assessing the user acceptance level for web-based filing of PDFs. If such an experiment were later to turn out to be a failure (nobody using it) then not very much money would be lost on the experiment.

What exactly is this "PDF problem"? You can see a brief discussion of the PDF problem.

How to choose what sort of work flow to incorporate into such experiments? Among the many possible things that the USPTO might select as the subject matter of a web-based PDF/TIF experiment, it would be desirable to select something which:

Summary: This paper identifies present workflow which, it is suggested, makes an ideal candidate for a web/PDF/TIF experiment. The USPTO is urged to set up a web-based system that would allow a filer to e-file, through a web page, a PDF or multipage TIF copy of what the filer would otherwise have faxed to the central fax number at 703-872-9306.

What is the "central fax number"? The anthrax scares of a few years ago prompted the USPTO to try to wean filers away from mail filings when possible, and fax filings turned out to work well. A fax filing received by the USPTO is a filing that is free from any anthrax worries. A fax filing received by the USPTO need not be irradiated to kill anthrax spores and need not be swabbed to test for ricin. Irradiated filings sometimes get brittle and sometimes crumble when USPTO personnel try to scan them into IFW. The anthrax scares prompted the USPTO to set up a number of fax servers, one for each technology center. Subsequently, effective December 1, 2003, the USPTO decided that all patent application related correspondence transmitted by facsimile would henceforth be directed to a central facsimile number, 703-872-9306, with a few exceptions. As of that date, replies to Office actions including after-final amendments that are transmitted by facsimile are required to be directed to the central facsimile number.

Faxes sent to the central fax number are not printed out. Instead, each fax is captured as an image file and is viewed by a USPTO person on a computer screen.

flow diagram
(Click on the figure for a larger view)

For each fax, a USPTO person splits up the image file into several parts. For example, a response to office action will typically be split up into a "claims" part, a "remarks" part, and sometimes other parts. The USPTO person then figures out what serial number the filing relates to, and posts the parts to the Image File Wrapper for that serial number.

How do paper filings get processed? Paper filings may be hand-carried to the USPTO or may be sent via Postal Service to P O Box 1450. Along the way they may be irradiated to kill anthrax sports, and swabbed to test for ricin. Eventually each such paper filing reaches a USPTO person. The USPTO person inserts section dividers into the paper filing, with bar-code markings denoting which portion is "claims", which portion is "remarks", and so on. The resulting stack of documents is inserted into a scanner. The USPTO person determines the serial number to which the filing relates, and the TIF files that result from this scanning are posted to the IFW for that serial number.

What is the proposed new system? The proposed new system does nothing more nor less than provide a web-based mechanism permitting a filer to e-file exactly what the filer would have faxed.

As may be seen in the figure, the filer would submit a PDF (or multipage TIF) file to the web page created by USPTO for this purpose. The image file would then be treated within the USPTO in exactly the same way, and by exactly the same people, who presently handle fax filings. The workflow and process steps and exception processing would be identical. Whether it is a multipage image file from a fax server, or a multipage image file from a web server, would make no difference in the steps later performed by USPTO personnel.

Minimal setup cost. It should cost approximately nothing to set up such a server. The EFS list, for example, has easily a dozen participants who could each design and build such a system gratis in a day or less, given access to full documentation as to how the present fax-filing-to-IFW system works. Little or no new training for USPTO personnel would be needed as the human processing steps for processing an image file that happened to arrive by web would be identical or nearly identical to the human processing steps for processing an image file that happened to arrive by fax. All that would happen is that some of the USPTO personnel who previously processed fax filings would switch over to processing web filings. Probably they could switch back and forth between the two types of filings quite easily, and indeed the USPTO person might not even need to know whether the image file he or she is processing came from fax or from the web system.

Minimal ongoing cost. The internal cost to the USPTO to process web-filed PDF responses would be no greater than the internal cost to the USPTO to process the present-day fax-filed responses. Indeed as filers migrate away from fax filings (as they are very likely to do) and toward web-based PDF filings, the USPTO would save money by not needing to have as many telephone lines and modems dedicated to serving the peak levels of fax filings.

A megabyte of fax filing ties up a USPTO telephone line for around 13 minutes. That same megabyte, e-filed, may only tie up a USPTO data line for ten seconds.

Paper filings would get converted to electronic. At present, some filers who have a choice between paper filings and fax filings will choose paper because it is awkward and error-prone to try to fax large numbers of pages. But with an option of web-based PDF or TIF filings, many of these same filers would cheerfully do whatever is needed to assemble the filing in PDF format for e-filing, so as to avoid photocopying and postage costs. The availability of an immediate confirmation from the web server, as compared with the days or weeks of wondering whether the Postal Service delivered a paper filing, would likewise motivate many erstwhile paper filers to e-file.

Conversion of paper filings to e-filings saves money for the USPTO. Every time somebody paper-files a filing, this requires USPTO personnel to scan the paper filing at USPTO expense. If that same filing is e-filed, then the scanning burden is off-loaded to the filer and away from the USPTO. This saves money for the USPTO. The filer, however, will not mind assuming the scanning burden, because (a) the filer probably would have scanned the filing anyway for photocopying purposes or for internal image management purposes, (b) the filer gets to have an immediate acknowledgment rather than waiting to see if the Post Office delivers the package, and (c) the filer saves money on postage.

This system would be a "crowd pleaser". Many filers have said they wish they could file information disclosure statements electronically even where the references are non-patent references. This system would provide that capability and would instantly be popular for that reason alone.

This system provides a low-stakes way to gain experience with PDF formats. Where the e-filing of a patent application is concerned, any failure is unacceptable to a filer. Loss of a filing date will often lead to irreparable loss of substantive rights for a client. Many would-be e-filers hold back from e-filing precisely out of concern for professional liability in the event of some failure of the e-filing software or system.

In contrast, the mere filing of responses to office actions has far lower stakes. While some responses are filed on the last possible date, many of them are filed well in advance of the last possible date. This means that if a filing went awry (due, say, to some problem with fonts in PDFs) there is often time to spare for re-filing the response (e.g. by fax or on paper or by means of a PDF generated with different fonts) and avoiding any loss of substantive rights. So far as responses to office actions are concerned, even an e-filing on the last possible day is, truth be told, not very scary, because if the e-filing goes wrong the filer can always simply drop a paper filing in any mailbox. There is no need, for example, for a 24-hour post office; the papers can be dropped in any mailbox with a mere certificate of mailing, prior to midnight local time.

This system permits filers to identify readily any problems with PDF rendering. The filer can simply log into IFW and see the actual images as posted into IFW. If the fonts are no good, the filer will be able to see it, and will be able to take corrective steps without the great fear of irreparable loss of the client's substantive rights that would attach to bad fonts in the filing of a patent application. Meanwhile any such incident can be studied in a cooperative way between the filer and the USPTO and the cause of the difficulty can be explored.

What if the file is not "electronic"? This experiment offers its promise chiefly in that a filer who has e-filed a PDF file will later be able to check in private PAIR and in IFW to see the actual images as posted to IFW. If there are problems with fonts or other PDF rendering problems, this will be apparent to the filer when the images are viewed in IFW. But this works only if the serial number to which the PDF file is posted is an "electronic" file. By this is meant a file that has a "location" in PAIR of "electronic". A file is "electronic" if:

(Some types of files, for example reissue files, are not "electronic" regardless of filing date.)

Probably the user web page for this system should:

Should this system actively enforce a prohibition against use with non-electronic files? I suggest the answer is "no". After all, even now a fax filer who faxes a paper for filing in a non-electronic file is taking some sort of risk. If a page of the fax is inserted backwards or sticks to a previous page or scans poorly in the filer's fax machine, the filing will be defective and the filer might not learn of the problem for months if ever. This is a problem that filers do not seem to feel is a very big problem, given that many filers choose to fax-file rather than paper-file. Such problems are sometimes detected by the Examiner, and are sometimes simply fixed later with a Certificate of Correction. The risks associated with a PDF e-filing are, I suggest, comparable in magnitude, namely rather small and can generally be detected and fixed later about as readily as fax risks. In any event the number of non-electronic files is diminishing with time, and the filer who wants to be cautious can simply not use the web-filing system for non-electronic files.

Does this system need PKI? It is commonplace in many e-commerce systems, including those systems created by patent offices, to assume that the system must include and make use of filer cryptographic certificates. This system, however, does not need such PKI complexity. Consider, after all, that a filer who files a paper response to an office action need not prove his or her identity using a cryptographic certificate. Anybody can stick an office action response into an envelope and mail it to the USPTO without anything more than an pen-and-ink signature. Anybody can fax an office action response to the USPTO without any need for a cryptographic certificate. The cryptographic goals and assumptions for web-based submission of responses to office actions ought not be at the level of inter-bank funds transfers, but ought instead to be at the level of purchasing a book from Amazon.com. Protection against eavesdropping is all that is needed.

The only cryptographic security employed in the purchase of a book from Amazon.com is mere SSL (secure sockets layer) security. Such security is signaled to the user by a protocol identifier of "https://" rather than the more common "http://". This is precisely the level of security that makes sense for a web-based system for responding to office actions.

What should there be in the way of a certificate under Rule 8? It is suggested that the USPTO would define a "certificate of web-based submission" that would work out to the exact same thing as a certificate of faxing or certificate of mailing. This would be by means of a partial waiver of Rule 8 to the extent necessary to permit this sort of certificate. It is important to realize, of course, that this certificate procedure would almost never actually be needed. For example, at the present time a fax filer who has received the fax receipt from 703-872-9306 can be quite confident that the USPTO will not later claim never to have received the fax. For a fax filer, once the fax receipt has been received, it does not matter much whether there was a certificate of faxing under Rule 8. Likewise, under the proposed system, for a web filer, once the web tracking number has been provided to the filer (through the web interface), it is assumed that the USPTO will not later claim not to have received the web-filed file. Thus, once the web-submission receipt has been received, it will not matter much whether there was a certificate of web submission under waiver-modified Rule 8.

Importantly, however, the waiver creating the certificate of web submission procedure could optionally be worded to allow the filer to state (in the certificate) under penalty of perjury the time zone in which the filer is located. This, together with the Virginia time stamp on the filing log, could be used to calculate the date to which the filing is entitled. This would remove the present time-zone penalty associated with being west of Virginia.

Should there be a requirement that the entire web-based submission is a single image file? This is an interesting question and user comments will be helpful on this point. For simplicity of design of the system, it might be desirable to require that everything be in a single file (i.e. a single PDF file or a single TIF file). This would make it easier, in some sense, to match up submitted files with tracking numbers. This would also have the salutory effect of forcing many filers to physically scan paper originals as a way of building up a single PDF file containing all of the submitted pages (e.g. Form PTO-2038 and the response to office action) which would completely avoid any of the PDF risks associated with licensed and missing fonts.

On the other hand, permitting the submission package to be two or more image files has some advantages. One "advantage", such as it is, is that the package might include, say, a scanned Form PTO-2038 along with a character-printed PDF of a response to office action (containing, say, a virgule-style signature). This would increase the chances of a particular filing including troublesome fonts, thus accelerating the accumulation of experience with troublesome PDF files. Given that this is one of the chief goals of this proposed system, then allowing multiple-file submissions is probably the best choice.

Further arguments in favor of multiple-file submissions include:

Probably the best course, then, is to permit attachment of two or more image files in the web filing process. The system would then need to keep a log that would group the image files together in accordance with the tracking number assigned to the filing.

Should there be a limit on file size? Yes, by definition there must be some limit on the total size of the filing package, as well as some limit on the size of any one file in the filing package. The reason is simple -- the hard disk on which these submissions are stored has finite size and no single filer should be permitted to use up the entirety of that hard disk. (It is of course a mirrored hard disk but still there needs to be a limit on file size.) This number could be adjusted based upon experience. The present USPTO ten-megabyte limit for e-filed patent applications might be a good starting value.

How sophisticated should the system be? The guiding principle for this experiment should be to Keep It Simple. The simpler it is, the cheaper it will be to design and implement. The simpler it is, the less there will be to go wrong. The simpler it is, the easier it will be to beta-test. The simpler it is, the more easily it can fit seamlessly and smoothly into the existing process flow for fax-filed responses to office actions. This entire system ought to be designed within a month, beta-tested within a a second month, and released to the public within a third month. Simplicity and avoidance of fancy features are the key.

What about displaying "thumbnails" of uploaded images? USPTO has systems called ETAS and EPAS for submission of assignments for recordation. With each of these systems, after the user has uploaded one or more imge files, the system offers the filer the opportunity to see "thumbnails" of the uploaded images. By clicking on a "thumbnail", the filer can view the entirety of the submitted page. This view uses a native viewer within the filer's web browser, such as a PNG viewer. In this way the system avoids use of a "plug-in" such as Adobe Reader. The PNG image provides an unambiguous indication of how the uploaded PDF file was rendered within the USPTO, and the filer can be imformed that the permanent record within the USPTO will be the image as displayed. This provides a clear line of responsibility on the filer to check the viewed thumbnails.

In the proposed system, such a thumbnail display capability could be included, so long as it does not delay rollout of the system and does not add cost or complexity to the system. Perhaps the code could be ported over from ETAS/EPAS for this purpose.

What kind of acknowledgment receipt should be provided? In the initial system, the acknowledgment receipt should be as simple as is humanly possible to design. Probably it would display nothing more than a tracking number, the date and time, and a page count from the PDF file. Internally the system would log the name of the uploaded PDF file, the file size, the tracking number, the date and time, and the IP address of the filer.

Why does this proposal include multipage TIF as a filing format? As is clear from the discussion above, there is some uncertainty surrounding PDF as a filing format. There is just not enough experience yet to know what sorts of odd things filers might knowingly or unknowningly hide inside their PDF files, that patent offices might not be able to deal with, or how often this sort of problem really comes up. There is just not enough experience yet to know how realistic a goal it might be to try to test for such things in PDF files at filing time so that "problem" PDFs could be rejected. It is easy enough to tell users not to use exotic fonts, or not to use "licensed" fonts, but some users simply will not know what fonts they are using. Some users for example may be lawyers who are starting with files created by clients, where the client may have used an exotic or licensed font unbeknownst to the lawyer. Likewise it is easy enough to tell users only to e-file PDFs that were created by scanning of paper originals, but some users will have PDF files in their possession the provenance of which is not known.

Multipage TIF is a format that sidesteps all of these issues because it is always bit-mapped and is never character-based. If only filers could be convinced to use TIF files rather than PDF files, then these font issues would simply not arise. But there is simply not enough experience yet to know whether filers are willing to learn how to generate multipage TIF files rather than PDFs. This proposed system would permit gaining some experience with filers on this point.

What enhancements could be added later? Once the system is in use and is known to be stable, enhancements could be added:

It is crucially important, though, to hold back from trying to build all or any of these enhancements into the system from the beginning. They aren't necessities. They add complexity to the integration into the legacy workflow. They would cost money. They would delay rollout. The initial system should be nothing more nor less than an identical substitute for a fax machine.

Why not let filers send in the PDFs by email? It will be suggested that it is faster and easier simply to let a would-be e-filer send the PDF file to the USPTO by email rather than by a web interface. Email is, however, a bad idea. First, spammers will send spam to the email address as soon as it becomes known. Second, email is intrinsically insecure with respect to eavedropping, arguably less secure even than faxing. Third, if an acknowledgment receipt is to be emailed from the USPTO to the filer, it might get lost in email or might get deleted inadvertently as spam. Fourth, the filer's email message itself might get lost on its way to the USPTO.

In contrast, an SSL web page is intrinsically secure against eavesdropping. With a web page, it is easy to provide an acknowledgment receipt to a filer that is not at risk of getting lost in the email or deleted as spam.

Will people try to e-file patent applications through this system? It would be important that the USPTO make clear that this system is a Rule-8 system -- a system that only does what faxes can do. Just as the present Rules make clear that patent applications cannot be fax-filed, those same rules would make clear that this system could not be used for e-filing of patent applications.

The web page for this system would need to contain appropriate instructions and cautionary language so that filers would be clear on what they can and cannot file through this system. The most succinct way to say it would be simply to state in large letters that this system can only be used for the same kinds of filings as can be faxed to the central fax number.

Note that this proposed system is just as usable by Mac and Linux users as it is by Windows users. The existing ePave system is unusable by Mac or Linux users. (Yes, I know about the Mac plug-ins that are supposed to permit running Windows applications on Macs. That's not the point.) This proposed system would permit experience to be gained with users who use operating systems other than Windows.

What else is good about this proposed system? The USPTO has heard many users say they would prefer to e-file patent applications using a "thin client", namely a web-based submission system. But jumping right away into the deep water is unwise; it would be a mistake if the first "thin client" e-filing project undertaken by the USPTO were a project for e-filing of patent applications. Getting it wrong (and there is some risk of getting it wrong) could alienate filers, since the loss of substantive rights is such a pervasive concern.

In contrast, e-filing of responses to office actions is low-stakes. Rolling out a new and unproven submission system (such as a hitherto untried web-based PDF/TIF submission system) ought to be done with something that is low stakes. Later, after months of experience with web-based image submissions, it would be possible to migrate carefully toward augmenting the web-based system to permit e-filing of patent applications. For example one issue is how, in a web-based e-filing system, to collect the bibliographic data. Should this be hand-typed by the user into a web form? Or should the user create the bibliographic data ahead of time in an XML file which will later be attached by the user to the web page for e-filing? These are questions that are not easy to answer at the present time. But experience with e-filing of responses to office actions, and with gradual enhancements of an office-action-response e-filing system, would permit making better guesses as to how bibliographic data would get entered at such time as web-based patent application filing were made possible.

As an example of a "toe in the water" next step, once this proposed office-action-response e-filing system were found to be stable, it could be announced that it is available for (say) web-based PDF/TIF e-filing of provisional patent applications. This could at first be a filing in which the bibliographic data are provided on a scanned cover sheet, processed manually within the USPTO just as it would be for a paper-filed provisional application. Then, gradually, an enhancement could be added that would permit provision of the bibliographic data in some simple XML format. Once the bugs were worked out of this, a next step would be receiving ordinary utility patent applications through the web-based system.

How does the present proposed system fit into the USPTO's planned "portal" system? The USPTO has suggested that in the medium term it may migrate from the present private PAIR system to a "portal" system. I am not quite sure exactly what is meant by this "portal" system but it seems to embrace the notion that a user would "log in" to a "portal" web site and then, once logged in, would move from screen to screen in some fully integrated set of web pages, so that all user functions (filing patent applications, viewing IFW, ordering file wrappers) would take place in this "logged in" environment.

Right now, a would-be e-filer of an assignment for recordation goes to one page, while a would-be visitor to public PAIR goes to a different page. Adding funds to one's deposit account takes place on yet another web page, while visiting private PAIR requires a visit to a still different web page. Checking the status of a pending order for a file wrapper requires logging in to another page. E-filing an IDS isn't even done on a web page but is done by launching ePave.

I have never heard a USPTO customer say that they think this is a problem that needs fixing. It's what browser bookmarks are for. Users do not presently demand that all of these functions take place in a single integrated "portal" with a single login. Putting these functions all into a single integrated "portal" (without addressing the things that keep people from e-filing patent applications now) is not going to make somebody e-file a patent application who was previously unwilling to e-file a patent application.

It is my suggestion that this proposed web-based system for PDF/TIF responses to office actions ought to proceed independent of (and faster than) any "portal" development. This experiment needs to be simple and inexpensive and needs to be rolled out right away. Trying to design in features now that are intended to integrate later with some as-yet-unfinalized "portal" design would merely add cost and complexity and would inject delay.

Benefit to the Examining Corps. Fax filings, by definition, never have resolution better than 200 dots per inch, due to physical limitations of the fax scanning array and the CCITT fax standard. An e-filed PDF or TIF image will (by USPTO requirements) be at least 300 dots per inch and many filers routinely scan and store 600 dots per inch. The practical consequence is less eyestrain for Examiners if the responses they must read are of better resolution.

Saving space in the offsite storage locations. At present, if somebody paper-files a voluminous IDS, USPTO is forced to scan the entire submission and then ship the paper to offside storage. This takes up space. Allowing e-filing of such an IDS saves this space in the offsite storage location.

Boosting e-filing percentages. It goes without saying (but I will say it anyway) that this proposed system for e-filed responses to office actions is likely to be wildly popular. Reasons why it will be popular include: