In a recent blog post I said “USPTO gets it right — new way to pay issue fees”. Turns out I was mistaken. USPTO did not get it right when implementing its new way to pay issue fees. What a disappointment.
In that recent blog post I noted that USPTO has worked out a new way to pay patent issue fees. The old way was that the filer would hand-key information into Form PTO-85B and would then upload an image of the form into EFS-Web. The later workflow was that some USPTO person would then hand-key the information into the data capture system to appear in the issued patent. This information included the assignee name and city and country, and the “attorney, agent or firm” information.
We try to keep close track of this stuff. We had been seeing that an average of 2.8% of the time, the USPTO people fat-fingered something from the 85B. About 2.8% of the time, we were forced to request a Certificate of Correction from the USPTO to correct the Office mistake.
So in September of 2015 I was delighted to see what I thought was an improvement. USPTO released the “eprocessing” version of Form 85B, in which the filer could paste characters into some blanks on a web-based form. Silly me, I assumed that (a) USPTO had surely set this up so that the filer’s characters would auto-load into the data capture system and (b) the USPTO error rate would drop to zero due to the auto-loading.
Turns out I was wrong. In one of our patents that issued recently, in which we had used the “eprocessing” version of Form 85B, USPTO misspelled one of the items of information on the front page of the granted patent.
Turns out USPTO is even now hand-keying the issue fee information into the data capture system.
Indeed if I review all of the cases in which our firm has paid an issue fee using the new “eprocessing” version of Form 85B, the USPTO screwed it up twenty percent of the time. This is a much higher error rate than before.
What a disappointment!
This is funny in a sad sort of way. So is your recommendation to stick with the old approach?
Well there are two reasons why in our office we are likely to stick with the new approach.
First, the new approach is less work. Fewer mouse clicks, etc. So that is a reason to use the new approach.
Second, people at the USPTO have informally indicated to me that some time down the line, maybe in a few years, the USPTO will get around to making it so that the characters supplied by the filer in the new approach would actually get auto-loaded into the data capture system.
The error rate of 20% under the new approach, as compared with 2.8% under the old approach, might just be caused by the relatively small sample size for the new approach. It’s only been in force for a few weeks.
Silly you. Thanks for this Carl. However, I thought most, if not all, the EFS-Web based forms the USPTO uses are hand-keyed anyway. Perhaps, the USPTO would not have to increase the fees to cover costs if they invested in software that auto-loaded information from all EFS-Web forms instead of investing in software and procedures for the Automated Pre-Examination Search program they want to implement. I mean think about it… The savings in the costs it takes one to manually key in information, and then to process the incoming paperwork to fix USPTO mistakes, and then have a person manually key in the correction (which I’ve had a case where when making the requested correction, another error was somehow entered by the USPTO). Sure, it might cost a few jobs (that are likely contracted outside), but it would result in money to more wisely spend someplace else, e.g., substantive examination and thereby enhance patent quality.
I am not sure why their error rate would go up to 20%? You would think they could at least copy/paste from the electronic form but apparently they are re-typing. I have filed a few cases recently with the e-processing form and will now keep a close eye on the published documents.
Well the error rate under the new approach being 20% may be largely due to the relatively small sample size. The new approach has only been around for a few weeks. So even a single mistake will look like a 20% error rate if the data set is only five issue fee payments. Maybe over the long run the USPTO error rate for data capture under the new approach will drop to match the USPTO error rate for data capture under the old approach.