I choose to take partial credit, in an odd way, for a curious Notice of Proposed Rulemaking that the USPTO publicized today. The Notice proposes to change the Rule that defines what filing date you get when you e-file something about patents at the USPTO. The Rule presently contains some word salad that works out to something like “you get what time it is at the e-filing server in Alexandria, Virginia when you click ‘submit'”. The proposed new Rule contains some word salad that works out to something like “you get what time it is in the Eastern Time Zone when you click ‘submit'”. I think my April 1, 2016 blog post, in which I joked that the USPTO had moved its “contingency” patent e-filing server to the Denver patent office, is a partial cause of this Notice.
The backdrop to all of this is that when the USPTO launched its EFS-Web patent e-filing system, The USPTO realized that the responsible thing would be to provide a backup e-filing system that USPTO customers could use any time the main EFS-Web system were to crash. With this in mind, the USPTO with some fanfare announced the establishment of its “contingency” EFS-Web system. The idea of the “contingency” EFS-Web system was that if the main EFS-Web system were to fail, users could simply switch over and start using the “contingency” system. There would thus be no interruption of service.
The first big time that this “contingency” system became highly visible to users was on May 14, 2014 when the main EFS-Web server crashed for some eighteen hours. What astonished users was that the “contingency” server crashed for exactly this same eighteen-hour period. They crashed at the same time and they were restored to service at the same time.
The next big time that this “contingency” system became highly visible to users was during the massive system crash in December of 2015, in which the main EFS-Web server and the “contingency” EFS-Web server both crashed and were unavailable for some six days.
It turned out (I am not making this up) that the people at the USPTO in charge of these two servers had plugged them into the same massive surge protector. When the surge protector failed, this cut power to both servers. I would have thought that what would have happened next is that somebody would then have gone around to every Best Buy in a twenty-mile radius to get however many surge protectors were needed to provide temporary power to the various devices that comprised the two servers, and plug everything in that way. But no. Six days passed before spare parts could be found to repair the single massive surge protector and only then could the two servers be restored to service.
I had done traceroutes to the two servers months before, and I had independently established that the two servers were also served by the exact same Internet connection. A single errant backhoe could have cut off both servers’ connections to the Internet.
Common sense, of course, dictates that you never put a backup server like this in the same building with the main server. You never plug the two servers into the same power supply. You never connect the two servers to the Internet in the same way. Ideally you put the backup server in a different state, connected to a different power company, connected to a different Internet backbone service provider. This is the sort of common sense that a schoolchild would offer if asked. But no, the USPTO system designers had chosen to put the “contingency” EFS-Web server in the same room with the main EFS-Web server, connected to the exact same power supply, connected to the exact same Internet connection.
Which then prompted my April 1, 2016 blog post in which I stated that the USPTO had relocated its “contingency” EFS-Web server to the Denver patent office. What made it so easy to miss that it was an April fool’s joke was that of course it would have been eminently sensible for the USPTO to do exactly that.
Then on November 7, 2016 the main EFS-Web server crashed and, sadly, the “contingency” server crashed simultaneously. This was the third highly visible simultaneous failure of the two servers. On August 15, 2018 there was yet another crash of the main EFS-Web server and, with it, the “contingency” server. This was the fourth highly visible simultaneous failure of the two servers. I blogged about this on August 29, 2018, using the photograph at right to emphasize my point about what redundancy is supposed to mean.
I like to imagine that my efforts to shame the USPTO into relocating its “contingency” server to some location that is geographically remote from the Alexandria campus are at least partly responsible for today’s Notice of Proposed Rulemaking. As the Notice explains, the USPTO is apparently now in 2021 starting to come around to the idea that maybe it ought to move the “contingency” server somewhere else. The Notice says:
The USPTO’s physical servers that receive electronic submissions are currently located in Alexandria, Virginia. However, in order to enhance resiliency, the USPTO is in the process of providing servers in Manassas, Virginia, and in the future may provide servers outside of the Eastern time zone.
Yes! after some seven years of efforts by many outside the USPTO, including me, to shame the USPTO into moving its “contingency” EFS-Web server to a location geographically distant from Alexandria, the USPTO is finally now “in the process of” actually doing so.
Manassas is a mere 32 miles from Alexandria, and I think most system designers would say that this is an irresponsibly small amount of physical distancing for a reasonable contingency server. But at least it is less bad than the way it was for the previous seven years, with the contingency server having been in the same room with the main server, plugged into the same power supply as the main server and connected to the same Internet feed as the main server.
The main point of the Notice, however, is that the present word salad of the present Rule more or less defines the customer’s Filing Date according to what time it is at the server when the customer clicks “submit”. And if and when the USPTO were to do the right thing and move the “contingency” server even further away, the new location might well be outside of the Eastern Time Zone. At which point the USPTO clearly wants to deny any customer the ability to forum-shop as a way to get a better filing date. The USPTO does not want the customer who has missed the midnight cutoff in Alexandria to be able to shift over to the “contingency” server located in some further-west time zone (for example in the Denver patent office as I stated it to be) as a way to snag a same-day filing date.
So the rulemaking proposes to shift to some new word salad that works out to “you get the time that it is in the Eastern Time Zone when you click ‘submit'” no matter what time it happens to be at the place where the server is that actually receives the customer’s submission. And yes I choose to take partial credit for this. I like to think that the photograph appearing above helped.