(Update: the 6001 customers did eventually get their registration numbers back two weeks later, on May 24, 2022. See blog article.)
(USPTO published an explanation of sorts. Blog article.)
Orwell’s 1984 imagines a dystopian future with a “memory hole”, which Wikipedia defines as
any mechanism for the deliberate alteration or disappearance of inconvenient or embarrassing documents, photographs, transcripts or other records, such as from a website or other archive, particularly as part of an attempt to give the impression that something never happened.
The morning of Tuesday, May 10, 2022 seemed like an ordinary Tuesday morning for trademark practitioners in the US. Just like any other Tuesday, about six thousand US trademarks got registered. But by Tuesday afternoon it became clear that this was no ordinary Tuesday. By today (Friday the 13th), we see that the Trademark Office has gone down the memory hole, and has “disappeared” six thousand and one US trademark registrations.
My firm happens to be handling one of the 6001 affected US trademark applications. For me the first visible event was at 1:39 PM Eastern Time when IP Badger reported to me that our application number 90007982 had gotten registered as US trademark registration number 6721048. I did what I normally do next, which is to go into TSDR to pick up a PDF of the registration certificate so that I could send it to the client (a family-owned business in Germany that makes orthopedic shoe inserts). But the PDF was not there in TSDR. I figured there must be some delay in the posting of the PDF certificates to TSDR, and I made a mental note to come back later to finish this reporting task for the client.
I am fortunate to be part of the e-Trademarks listserv, a community of about one thousand of the best trademark practitioners in the world. When something odd is going on in USPTO’s systems, the usual way that practitioners learn of it is through one or another of the practitioner listservs. At 4:10 PM Eastern Time, alert listserv member Kevin Grierson posted this:
Two of my client’s trademarks registered today. The registration number and other information are showing in TSDR, but there is no registration certificate available in the “documents” tab and the marks do not yet show as registered in TESS.
I generally grab a PDF of the certificate from TSDR the day the mark registers and forward it to my client; can’t recall [any previous incidents of] not being able to do that the same day the registrations issue.
And of course what Kevin reported was the same puzzling thing that I had encountered just a couple of hours earlier that same day. I, too, found myself unable to “grab a PDF of the certificate” from TSDR to forward to my client.
So just to keep score, on registration day ( May 10, 2022), the registration number and registration date were posted in TSDR but were not yet visible in TESS.
And the evening and the morning were the second day. On Wednesday, May 11, things had flipped. Now it was TESS that said the applications were registered, providing the registration numbers and registration dates, but the registration numbers and registration dates had disappeared from TSDR.
And the evening and the morning were the third day. On Thursday, May 12, the registration numbers and registration dates, visible the previous day in TESS, had disappeared from TESS.
At this point it was clear to the practitioner community that something was seriously wrong at the USPTO with the thousands of trademark registrations that had issued on Tuesday, May 10, 2022. Whatever had gone wrong, its effects extended at least to the TSDR system and the TESS system. And yet nobody at the USPTO had fessed up to this big problem. There is a web page called USPTO Systems Status and Availability where the USPTO normally admits to system problems and outages. But there was no mention on the USPTO Systems Status and Availability page of the missing PDF registration certificates, nor any mention of the registration numbers and registration date appearing and then disappearing in TSDR and in TESS. No mention on May 10, no mention on May 11, no mention on May 12. Surely at this point everybody in the Trademark Office, all the way up to the Commissioner for Trademarks, was aware that Something Was Seriously Wrong, but even after two days, nobody at the USPTO had fessed up to this big problem.
Fortunately for those who do not like memory holes, information sometimes gets captured and preserved. In this case, alert listserv member Ken Boone captured a list of the 6001 trademark registrations and the corresponding application numbers before the Trademark Office sent the registrations down the memory hole. He provided the list to me and I have posted it here. As you can see, the registration numbers run from 6720667 (granted to application number 79216524) to 6726667 (granted to application number 97229583).
And the evening and the morning were the fourth day. Today, May 13, the Trademark Office slipped a one-page document into each of the 6001 application files. You can see it at top right. In each of the 6001 application files, there is a “Note to the file” that has been backdated to May 11, 2022. (Yes, the “memory hole” behavior of the Trademark Office extends to misrepresenting by two days the date upon which this document got slipped into the 6001 application files.) Each of the 6001 documents is identical. Each document says:
Note to the file: Application status mistakenly changed to registered on May 10, 2022.
Of course it is pretty clear what went wrong. The USPTO had announced on May 2, 2022 that in a cost-cutting move, it would soon cease issuing physical trademark registration certificates. The last physical trademark registration certificates would get mailed out on Tuesday, May 31, 2022 and beginning on Tuesday, June 7, 2022, registrants would have to make do with mere electronic (PDF) registration certificates. The spin was that this supposedly benefits trademark applicants because:
The transition to electronic registration certificates will give trademark owners easier access to their registrations and allow them to receive their certificates quickly after their trademark is registered.
What went wrong, of course, is some bad luck about the Trademark Office’s preparations for the big transition on June 7. The Trademark Office is clearly doing “trial runs” of its new money-saving internal workflows, and the trial run for May 10, 2022 was a complete failure. The legacy workflow was something like this:
- USPTO’s publishing contractor Reed Tech prints the physical registration certificates and stuffs them into envelopes for mailing.
- Reed Tech provides a data file to the Trademark Office with the registration numbers and matching application numbers, and the Trademark Office uploads the data file into TSDR and uploads the data file into TESS.
- Reed Tech provides the PDF registration certificates to the Trademark Office, and the Trademark Office uploads the PDF registration certificates into TSDR.
Who or what is Reed Tech, you might ask? Reed Tech is a government contractor that in 1970 managed to convince the USPTO to divert the printing of issued US patents away from the US Government Printing Office. Founded in 1966 as International Computaprint Corporation (ICC), it landed the patent printing contract in 1969. Those who do patent searching are aware that full-text searching of US patents first became possible for US patents issuing in 1974; this was because ICC, then branded as Reed, was monetizing its exclusive access to the text of granted US patents as the typesetter and publisher of those patents. (Any would-be competitor would be able to obtain such text only through expensive and error-prone OCR or hand-keying activity.) Reed made enormous amounts of money through its US patent full-text databases that were marketed through commercial services such as Dialog and Orbit. In 1995, Reed launched Reedfax, a for-profit service by which a law firm customer could order up copies of US patents and US trademark registration certificates and receive them by fax.
Reed’s near-monopoly on patent data began to crumble in 1998 when Internet activist Carl Malamud successfully shamed the USPTO into making US patent data available free of charge on the USPTO web site. Malamud had presented the USPTO with an ultimatum — release the patent data free of charge or he (Malamud) would do it for them. He threatened to purchase the data (at considerable cost, given USPTO’s unfriendly pricing for such data) from the USPTO and publish it himself. As described in a June 25, 1998 New York Times article:
The decision to make the data base freely available is a result of a fierce debate that has gone on for years between public interest advocates who argue that Government information should routinely be made available on the Internet and companies [like Reed that have non-arm’s length access to the information].
Indeed, the Clinton Administration’s action came less than a week before a deadline imposed by Carl Malamud, an independent Internet pioneer, who in May said he planned to purchase the data from the Patent and Trademark Office and make it publicly available if the Government failed to act.
Shortly after this, the USPTO released its PatFT search web site for searching the text of issued US patents, and this was the beginning of the end for the companies such as Reed that had previously made a lot of money charging customers for the ability to do full-text searching of US patents. I have to imagine that Reed has since then lost almost all of the very lucrative revenue streams that used to flow from its position as publisher of US patents and US registration certificates. I also have to imagine that Reed has figured out that nobody else can effectively bid for its contract with the USPTO for this publishing work, and has been quietly continuing to extract as much money as it can from its captive customer.
Now in 2022, the USPTO has made decisions to cease the the provision to its customers of physical US patents and physical US trademark registration certificates. The customers will have to make do instead with PDF certificates. The USPTO will save incredible amounts of money, not having to pay Reed to print the documents and not having to pay Reed to mail them out.
Back to the legacy workflow. To completely cut Reed out of the action, of course the Trademark Office would need to develop its own internal processes for generating its own PDF registration certificates. This is the sort of thing that computers are supposed to be good at! The Trademark Office already possesses all of the source information for such certificates and would merely need to incur some modest one-time costs to write the software to generate the PDF files.
This is, of course, the sort of thing that a half a dozen competent graduates of any computer science school could accomplish in a single weekend, if given some pizza and soft drinks, and they would have part of Saturday and all of Sunday left over.
But that’s not how the USPTO does software projects. The USPTO puts these things out for bid, and the bidding process is rigged so that almost no normal company could ever even submit a bid. There are half a dozen beltway bandit government contractors that are the only companies that are capable of satisfying all of the many bidding conditions imposed by the USPTO. The result is that even the simplest software project ends up costing at least the high six figures, and takes many months.
I have to imagine that this past Tuesday was The Big Day for the Trademark Office to try, for the first time, to cut Reed out of the action for everything except the physical printing and mailing of the registration certificates. The Trademark Office would, for the first time, push the “go” button on the software from the beltway bandit that was going to generate the various PDFs and lists of registration numbers that would need to get uploaded into TSDR and TESS.
Had this software been designed by my dream team of six randomly selected recent CS grads, I expect it would have worked right on the first try. But instead, the software crashed. And if it had crashed completely, in a sick way that would have been ideal. Instead, the beltway bandit software crashed in a sloppy way. The software worked well enough that on Monday evening it prepped a batch job for TSDR so that the registration numbers and registration dates did successfully load into TSDR at about midnight, and were visible early in the morning of Tuesday. The software worked well enough that on Tuesday morning it prepped a batch job for TESS so that the registration numbers and registration dates did successfully load into TESS at about midnight, and were visible early in the morning of Wednesday. By mid-day Wednesday, however, enough phone calls had arrived at the Trademark Office from worried customers that eventually somebody at the Trademark Office figured out that the batch job that was supposed to load the PDF registration certificates into TSDR had crashed. The PDFs had never gotten uploaded to TSDR. Maybe there was some scrambling around and maybe what eventually revealed itself is that the software that was supposed to actually generate the PDFs from the Trademark Office’s internal systems had failed to actually generate the PDFs. Or maybe at the last moment, somebody at the Trademark Office had noticed that the software was buggy and had generated PDFs in which characters with diacritical marks had gotten changed to smiley faces, and in which the logo marks had been halftoned out of recognition. Or nobody at the beltway bandit company had taken into account that some registrations have enough characters to spill over to a second page, and so had failed to provide for the formatting of a PDF having a “page 2”. Those of us who are outside the USPTO will probably never learn just what contracting blunder or bidding-spec mistake or system design error led to the things that went wrong this week. But anyway, the externally observable result is clear and unequivocal. On Tuesday, May 10, 6001 trademark applicants were told “this is your new registration number”. And on Friday, May 13, those 6001 trademark applicants were told that the events of May 10 had not actually occurred.
So what to do?
There’s what I think most people that I hang around with would have done. And there is what the Trademark Office actually did.
I think that if you pick at random any one of the people I hang around with (professionally on the e-Trademarks listserv, for example), the person would have done something like this:
- Step 1. Promptly post a status message at USPTO Systems Status and Availability fessing up to the fact of the PDF registration certificates being missing from TSDR, and promising that more would be posted soon.
- Step 2. Figure out what the remedial steps would be. Post the PDF files a few days late, and apologize for the delay? Generate the PDF files all over again, this time for issuance a week later, using the same old registration numbers that had already been communicated to the applicants? Generate the PDF files all over again, this time for issuance a week later, using all new and higher-numbered registration numbers, and telling applicants to disregard the registration numbers that had already been communicated to the applicants? Give up and pay Reed to clean up the mess? But anyway, step 2 would be to figure out what do do in the way of remedial steps.
- Step 3. Promptly communicate the remedial steps to the applicants.
As I say, there’s what I think most people that I hang around with would have done. And there is what the Trademark Office actually did. What the Trademark Office actually did was try to tough it out and avoid embarrassment. What the Trademark Office actually did was (as of today) fail to say anything public at all about what went wrong or what it plans to do next. The six thousand and one registration numbers got flushed down the memory hole, first from TSDR and then from TESS, as part of a Trademark Office effort to try desperately to pretend that the registration numbers had never existed.
Even now, the six thousand and one trademark owners are in a position of not knowing when, if ever, they will have restored to them the registration numbers that were given to them on May 10 and then clawed back from them on May 11 and May 12. All they have is the content-free “Note to the file” which raises more questions than it answers.
The problem here of course is that there are trademark owners who actually care when their applications get registered. Sometimes it is because a trademark owner needs to be able to enter the registration number into a nearly-completed court complaint, and head down to the courthouse, to seek preliminary injunctive relief against a counterfeiter. Sometimes it is because a trademark owner needs to get a trademark registered with US Customs (which you can only do once you have received your registration number) so that US Customs will start intercepting imported counterfeit goods. Sometimes it is because a trademark owner needs to be able to invoke Article 6quinquies of the Paris Convention in some foreign trademark office, and perhaps needs to do it in a hurry.
I figured maybe I ought to contact the USPTO to see if they wish to comment, before clicking “publish” on this blog article. Maybe, for example, the Office of the Chief Communications Officer (the office in charge of press, media, and general communications) would be willing to comment. The Office’s web page says:
For all media inquiries, contact the Office of the Chief Communications Officer main phone line at 571-272-8400.
I dialed that number, and the recorded greeting, here quoted in its entirety, was:
Thanks for calling the USPTO. Press 1 to sign into voice mail, or press 2 for the dial-by-name directory.
I guess I will not succeed in getting any comment from the Office of the Chief Communications Officer.
(Update: I also see that the OCCO web page says that in addition to contacting the OCCO by phone, one may send inquiries by email to a Paul Fucito, press secretary, or to a Mandy Kraft, deputy press secretary, so I also tried that. Neither of them got back to me.)
I will now click “publish” and I will send a link to this blog article to some people in the Trademark Office.
Very interesting, and very funny! – if only it hadn’t caused so many problems for so many people.
I suspect that by now, there are a bunch of circular-firing-squad e-mails describing the problem and attempting to deflect blame to a rival or external party. Are those not subject to FOIA?
I wouldn’t be at all surprised if the contractor hired to do the PDF creation process was Reed. It should have been a very easy procedure for them, since they already do the PDF creation process for the legacy workflow, and they’d have an incentive to try to keep that cash cow alive as much as possible. The PDFs would be different, however, with some sort of authenticity encoding built-in. I have a suspicion that authenticity encoding infrastructure was what might have fallen apart.
I just did a quick random sample of several applications in that list. Unlike your application, which was published for opposition in February, so could reasonably have been registered on May 11, quite a few of my random sample had apparently not yet been published for opposition, so the “Note to file” would be correct: that application should not have been marked as registered. I found others in your condition, however: published for opposition in February, so ripe for registration. My sample was small, but it makes me suspicious that there really were a bunch of registrations that issued incorrectly that day and that the PTO couldn’t immediately figure out which ones should have registered and which ones shouldn’t, so just quietly pulled everything that registered that day. Regardless of what happened, there is no excuse for the PTO not publishing something somewhere explaining that they’d had a massive SNAFU,
I have loaded the 6001 cases into IP Badger. Yes prompted by your comment, I spot-checked and yes there were definitely some cases that had not yet been published for opposition.
I will try to follow the 6001 cases as best I can, in IP Badger.
Did you check if those records NOT published for opposition were for the Supplemental Register? I just reran my search of those 6001 serial numbers. All but 182 were published for opposition on or before 22 February 2022. The 182 without a published for opposition date entry are for the Supplemental Register.
Curiously, the 6001 registration numbers are consecutive numbers. From my experience at the USPTO, that would be very unusual. The normal quality review process typically would remove 1 or 2 registrations per 1000, leaving those registration numbers unused. Well, that’s what I recall, but it has been a while.
Just got a batch of e-registration certificates via email. The client was copied on the email and a copy of the email is in the public file showing the email header with the client’s personal email address. The USPTO’s assurances regarding shielding the email address from public view, TMEP 803.05(b), is out the window.
Thank you for posting your comment. Your comment inspired this blog article.