Allowance Rates

My first unsolicited feedback!

A former examiner reached out. At first I was worried because I've been pretty hard on that profession. Turns out it was a delightful conversation. Production quotas are a scourge. It's not at all surprising if the documents are routinely not read. Sad and shameful, but not surprising… also not entirely the examiner's fault because it's the system that caused this. If anyone wants to chat, feel free to reach out. I'm particularly interested in talking to folks who don't agree the system is broken. I'd like to hear your perspective.

The short version

Stick with me, data is hard and confusing. I'll try and make it simple.

If you don't make it through, here's the TL;DR. Patents are way too easy to get. Or maybe it's hard not to get a patent. And there's really nothing you can do to stop a bad patent application.

Stats are hard to analogize. The US Patent Office (the gatekeeper) allows 4 out of every 5 applications. To put it another way, only 1 out of 5 applications fail to become a patent. You're about as likely to drop your phone in the toilet^† as you are to not get a patent from the USPTO. Sorry, apparently I'm a fan of the toilet humor.

Allowance Rates

As always, please check my stats & math. I'm playing fast and loose with the analysis.

How many patents (that were filed 2013 or later) have issued? That's a big number. I'll focus on Utility patents. ~3.6 million (~4.4M applications)

How many applications were killed (abandoned, failed to respond, etc.)? ~0.9 million

That alone is pretty stark: Out of the ~4.4M applications that we can classify, ~3.6 million became a patent. Only ~0.9 million failed (basically the applicant gave up). That's a lot. And that's the bar. If you keep at it, you're probably going to get a patent. How many of those 900,000 failed applications were given up on for financial or other reasons vs. a definitive no from the USPTO? I don't know. It's almost like the USPTO doesn't actually say no, they just let you give up.

My point is: if you file an application and are prepared to go the distance, you're almost 80% likely to get a patent. Nearly 80%!!! 4 out of 5 applicants agree, getting a patent is easy — way too easy. Those are good odds and no wonder this garbage patent snuck through.

How does that impact good patents?

We can argue about what constitutes a good or bad patent, but do you really think 80% of ideas are good? I know 80% of mine aren't. And is there even a notion of a good patent? Go on Shark Tank and they love to ask if you have a patent. That's it, basically a checkbox. The reality is it's harder to not have a patent (at least if you give it a shot). The patent itself is the credibility, not what the patent actually covers (that only matters if you have millions of dollars). That's a problem with how society views patents. That's why I'm fighting this bad patent.

The fundamental premise of this whole site is that it's far too easy to get a patent (whether it's good or not) and it's way too hard to dispose of bad, poorly examined patents.

If the USPTO has faith in their examination process, they would certainly be interested in the flaws I have pointed out in this particular examination. If 80% allowance is legitimate, and not just a ruse to generate recurring revenue in the form of maintenance fees, then my case should certainly be an outlier worth fixing, right?

Effectiveness of Third-Party Preissuance Submissions?

On that feedback call, the question came up about the overall effectiveness of Third-Party Submissions. We wondered: how many Third-Party Submissions actually tipped the scales? The premise is simple. Someone, somewhere, thought a patent application was previously disclosed (prior art), and went through the process of filing a 3rd party submission. It's not hard to do, but it's one of those pointless busywork things that gain you nothing. If you go to the trouble of filing a Third-Party Submission, chances are you have a good reason to do so. That seems like a self-selecting group. I would expect the allowance rates to be wildly lower for applications with Third-Party Submissions… wouldn't you?

That's a testable theory. All the data is public.

I had already done some analysis on this particular examiner, so I just needed to expand the workflow to cover a larger dataset.

• Applications filed 2013 or later — Important AIA date (where Third-Party Preissuance Submissions became a thing).
• Applications that were allowed — Clear. The unallowed ones are more interesting, but harder to mine the why.
• It gets a bit fuzzy in recent years because some applications may be pending.
• Patents issued in 2013 were filed long before that, so the file date is key.

Do 3rd party submissions matter?

Not nearly as much as I expected (and nowhere near what I hoped).

Here's the high level: if you file a patent application, you are overwhelmingly likely to get a patent. The train is rolling and nearly impossible to stop. It might not be a good patent, but how do you tell a good patent from a bad one? Someone else has to spend a crap ton of money. I've already made the case that having a patent, even if it's garbage, muddies the waters and creates confusion in the market. This might need a dedicated follow-up post.

Let's look at the 3rd party submissions because that's a tiny fraction of the total and barely significant (yes, statistically significant, but barely).

Filed ≥2013	Reached allowance	Not allowed	Pending (excl.)	Allowance rate
All	3,652,084	931,463	756,106	79.7%
3P	5,131	2,055	1,268	71.4%

At first glance that's an 8-point drop — 79.7% down to 71.4%. It looks like something. But that's across all applications. Let's check my math. This is exactly the kind of number that falls under lies, damned lies, and statistics.

Most of that 8 points is a mirage (it's more like 2%)

71% allowance is still bonkers. If you file an application, you are overwhelmingly going to get a patent, whether a member of the public pulls the only lever available or not. Whether there is pertinent prior art or not.

On paper, applications with a 3rd party submission allow less often, and there are some reasons that might be real:

1. The 3rd party submissions work — the prior art actually sinks some applications.
2. People file them against applications that were already in trouble.
3. The group of examiners is just more diligent or less averse to rejecting patents.

Let's ask where these submissions get filed — because they don't get filed evenly.

Technology Centers and Art Units

The USPTO groups its examiners into broad Technology Centers, each split into smaller art units. Third-party submissions pile up in just a few Technology Centers: chemistry, biotech, and pharma (the 1600s and 1700s). Those are fields that probably have big stakes and diligent participants (filers and interested patent owners).

Nearly half of every 3rd party submission lands in those two Technology Centers. And those TCs are hard — applications there allow well below the 80% average even when nobody submits a thing (about 65% and 72%, respectively). I still think 65% is too high of an allowance rate, but that's better than 80% (better because my whole premise is that patents should be harder to get — that makes them stronger and more valuable, not something trivially copied).

So "3rd party submissions vs. everybody" is skewed because of these tough Art Units. Why is there such a thing as a tough Art Unit in the first place? Shouldn't they all be tough? In other words, those Art Units drag down the allowances, but also have a disproportionate number of 3rd party submissions (I told you this was going to be hard to follow).

First, be clear about what these numbers are. 71.4% is the allowance rate for every application that had a third-party submission — all of them, nationwide, every art unit thrown in one bucket. 79.7% is that same global rate for everybody (submission or not). Comparing those two isn't fair, though, because 3PS don't land evenly — they cluster in the tough neighborhoods.

So here's the apples-to-apples version. Leave the 3PS number exactly where it is — it's still that same 71.4% — but swap the thing you compare it to. Instead of the 79.7% national average, measure it against the allowance rate of the other applications in the same art units where 3PS actually get filed. That baseline is 73.6% — lower than 79.7%, because those art units are genuinely harder.

Compared against	3PS allowance	Baseline allowance	Gap
Everybody (all art units)	71.4%	79.7%	−8.3 pts
Same art unit	71.4%	73.6%	−2.2 pts

The gap collapses from 8.3 points to 2.2. Three-quarters of the apparent "effect" was never the submissions doing anything — it was just 3PS coincidentally piling up in already hard-to-approve fields.

My own backyard: Technology Center 2800

The '537 lives in Technology Center 2800 — electronics, semiconductors, sensors.

• This examiner's art unit: 2863
• And my examiner's art unit: 2855
• But oddly not that other one. That was in art unit 3746…?

Why three separate applications that were arguably the same exact thing were spread across 3 Art Units and 2 wildly different Technology Centers is a mystery.

So how does my 2800 neighborhood look?

TC2800, ≥2013	Reached allowance	Never allowed	Allowance rate
All	835,858	112,411	88.1%
3P	703	153	82.1%

Wow. In this corner of the Patent Office, where my case sits, the USPTO says yes 88% of the time. Eighty-eight percent! No wonder a bad claim strolled right through. And the applications that had a motivated stranger's carefully submitted prior art sitting in the file? Still 82%. Even here, the 3rd party submission is barely a speed bump.

Even with a 3rd party submission, applications in this Technology Center still allow at 82% — above the overall 80% average. That alone should raise some serious questions.

The effect basically vanishes

Go art unit by art unit across the whole dataset and the argument "3rd party submissions hurt" (or even weigh on the patentability) falls apart. 3rd party submissions do almost nothing in terms of allowance.

In about half of art units, the applications with a submission allowed more often, not less. The typical art unit shows no difference at all. It's a coin flip. These aren't massive numbers (3PS are barely one-sixth of one percent of filings) so take all of this with a grain of salt.

"How much does a third-party submission actually move your odds?" almost nothing! Somebody reads your application, hunts down the prior art, fills out the government forms, files them — and the needle barely twitches. That train already left the station.

The one lever the public has to shout "wait — look at this" (the exact lever I pulled on the '537) turns out to be a rounding error.

Remember, my issue with this one particular patent is that it doesn't disclose anything novel, and I also believe I thoroughly disclosed all of the relevant claim elements. My issue is they got a patent without actually inventing anything.

To be fair: it does something

One caveat before I close, because "barely moves allowance" isn't the same as "does nothing."

I also checked how hard these applications had to fight before the office finally caved. Two signs of a rough prosecution: a final rejection (the examiner digs in) and an RCE (you pay up to keep arguing past it).

Allowed apps, ≥2013	Final rejection	RCE
All Utility	33.3%	24.7%
All Utility + 3P submission	50.8%	41.1%
TC2800 (my neighborhood)	24.1%	17.6%
TC2800 + 3P submission	35.0%	27.5%

Across the board — or in my own 2800 neighborhood — applications with a submission on file were about 1.5× more likely to draw a final rejection and to need an RCE. Same destination, bumpier drive. Maybe that's an examiner actually pushing back with the prior art — or maybe these were just messier cases to begin with. These numbers can't tell you which. But it's the one place the submission clearly left a mark.

How to get a patent

Just keep after it. You're more than likely to get one.

† About 18% of Americans say they've dropped a phone in the toilet (YouGov, 2022; 2,000 U.S. adults) — right about the same ~20% odds as getting turned down for a patent. ↩

Related dispatches

• Dispatch No. 4 — The first time this idea was good enough to copy — the earlier time this same idea was good enough to copy.
• Dispatch No. 6 — The Third-Party Preissuance Submission — the submission whose data shows it did almost nothing.
• Dispatch No. 9 — The Overly Broad Claim Slipped Through — the overly broad claim that slipped right through anyway.
• Dispatch No. 14 — My Apology to Good Examiners — my apology to the examiners who get it right, and the quota system that fails the rest.
• Dispatch No. 15 — Strapping Cell Phones to Oil Wells — the cell-phones-on-oil-wells origin of the rotation-sensing I disclosed first.
• Dispatch No. 26 — TL;DR; All the system failures — the system-failure recap behind that "80% Non-Original, Legally Approved" billboard.