Menu Close

Nonpublication Requests and Their Pitfalls (Including Making Zombies)

Sync and Share All Your Calendars with CalendarBridge

My oversight last week had me wondering how common nonpublication requests are. I couldn’t find this stat in any of they USPTO’s dashboards nor anywhere else on the Interwebs. It’s possibly a stat that could be obtained through a FOIA request, but that’s no fun and I am too impatient for it. Instead I tried back this number out a couple of ways. In a follow-up post I will explore a hypothesis I have that nonpublication requests are underused.

Estimating nonpublication requests based on published allowance rates

The technology center data currently available on the USPTO visualization center shows the following data for FY2019 through January:

USPTO Reported Allowance Rates (FY2019 through Jan)

# Allowances8,96712,8359,72112,68615,62628,00115,19318,049121,078
Allowance Rate w/o RCE66.6%69%75.6%78.5%83.6%86.3%68.272.2%76%

An allowance rate of 76% with 121,078 allowances means 38,450 non-RCE abandonments.

Next, looking at the public PAIR data, I count 111,425 allowances and 33,381 non-RCE abandonments. Both those numbers are about 9% lower than the numbers reported by the PTO. Why?

Since an allowance sooner than mandatory publication is fairly common (e.g., for Track 1 applications and continuations), many of those allowances which have not yet issued would still be unpublished even without a nonpublication request. So it seems that the percentage of applications allowed between Oct 1, 2018 and Jan 31, 2019 which had nonpublication requests is likely a fair bit less than 9%.

Abandonments before mandatory publication for non-continuing applications are pretty rare1. As I discussed in this post, however, roughly 25% of applications are continuing applications. So there could be some abandoned continuing applications among these abandonments. Hence, I think the best conclusion I can draw from this is that 9% is an upper bound on percentage of applications filed with nonpublication requests.

Estimating nonpublication requests based on priority and issue dates

The next approach I took is to look at how many patents issue more than 18 months after their priority date without having a prior publication. Here’s what that looks like:

By Filing Date:

By Grant Date:

In both cases, it is about 3% of patents that issue more than 18 months from their priority date without a prior publication. Of course, some applications with nonpublication requests likely issued earlier (e.g., Track 1 applications with nonpublication requests). Thus, here I think 3% represents a lower bound.

Estimating nonpublication requests based on published applications

Okay, so after I did the above two analyses I realized there is an easier way (I left the sections above as a sanity check and because I think the data there is interesting nontheless).

For the third method, I simply pulled the number of published applications which have nonpublication requests in their file history. Here’s what that looks like:

The red line shows that the number of published applications filed with a nonpublication request was about 5% from 2005 to 2013 and drops off in more recent years…presumably because more of those later applications are still pending, but possibly because fewer nonpublication requests are being filed2.

But wait, what’s that orange line?

Nonpublication gotcha: publication of a child application

Here is a zoomed-in view of the orange line from the previous graph:

These are applications which have a nonpublication request in their transaction history, there is no rescission of nonpublication request in their transaction history, and they are not patented. So why are they in public PAIR? It took me a minute to figure out, but then I realized that these applications are publicly accessible because they have children applications which are published.

Nonpublication gotcha: failure to rescind

The better-known gotcha with respect to nonpublication requests is of course abandonment that results from failure to rescind it before filing a foreign counterpart. In researching this post I came across the great “ant-like persistence” blog where Carl Oppedahl mentions that the USPTO didn’t always notice such failures, resulting in what he calls “walking-corpse patents.” Upon reading that, I of course decided to go zombie hunting.

I pulled up all applications which PAIR lists as having a non-rescinded nonpublication request and a child PCT (I had to settle for PCTs because PAIR doesn’t track foreign filings and I don’t yet deal in foreign data). That returned about 1100 results. I looked up about 20 of them. Most never nationalized the PCT but I think I found one former zombie: 7,835,712 and one current zombie: 7,581,611 Disclaimer: I only skimmed the contents files (I am on a snowboarding vacation in SLC, after all), so perhaps I missed something in there that one of my readers (either of you 😉) will catch.


  1. It seems about 5% of applications are filed with nonpublication requests. But, keep in mind that this is not adjusted for percentage of applications that are eligible for a nonpublication request (e.g., I haven’t accounted for applications that were first filed in a foreign country).
  2. Remember that publication of a child makes nonpublished parents available in public PAIR.
  3. Remember to rescind your nonpublication requests before foreign filing.
  4. Don’t forget to Subscribe

1they would typically fall in the abandonment categories that are down in the dust in the second chart of this post
2It would be interesting to look at this in combination with the percentage of applications that are being foreign filed. But that will have to wait for another day. Subscribe so you don’t miss it.