Says the patent microcosm

Summary: Quality assessment of US patents is becoming a hotter topic now that PTAB cleans up the mess and courts oftentimes reject asserted patents (which are, on average, considered to be better)

THE truth of the matter is, many patents are of low quality, but those that end up in court are typically the better ones. A few days ago we stumbled upon this post from Finnegan, Henderson, Farabow, Garrett & Dunner LLP. It tracks PTAB's progress reassessing particularly bad US patents:

The Patent Trial and Appeal Board issued 61 IPR and CBM Final Written Decisions in December, including decisions following remands from the Federal Circuit, cancelling 852 (73.83%) instituted claims while declining to cancel 301 (26.08%) instituted claims. Patent owners conceded only 1 claim (0.09%) through motions to amend or disclaimer in cases reaching a final decision. For comparison, the cumulative average rate of instituted claims cancelled in IPR and CBM Final Written Decisions is about 75%.

"investors have become much more sophisticated. It’s no longer simply about patent quantity. It’s about quality," Finnegan wrote in another new article. Here's the relevant part:

Not so long ago, during investment rounds, the IP due diligence typically conducted by a VC or CVC involved little more than asking, “How many patents does your company have?” A satisfactory number provided in response often ended the inquiry.

Today however, investors have become much more sophisticated. It’s no longer simply about patent quantity. It’s about quality. Now, investors often spend the time between the term sheet and the closing carefully assessing the blocking power of the investment’s patent portfolio. They examine the IP with a magnifying glass. If investors are not satisfied with the strength of a patent portfolio, they walk away from the deal or lower the valuation.

It's difficult to say exactly or even roughly just what proportion of patents at the USPTO can actually survive a court's test, but we are guessing that number is not high. 5 days ago the PTAB-hostile Anticipat took note of PTAB's reassessment of (often throwing away) obvious patent applications. Here are some interesting numbers:

Obviousness is by far the most common rejection that gets appealed to the Board. This particular ground of rejection does not draw attention for being reversed at the low end or on the high end. Over the past year and a half, the 12,000 obviousness decisions were wholly reversed about a third (34%) of the time. 43% of the time are at least partially reversed. One might expect these rates to be uniform across tech centers. They are not.

Alice and software patents are of most interest to us. Many patents are just software patents disguised as "AI", "cloud", "on a device", or "over the Internet" etc. "Anecdotally, I've found cross-examiner citing of the same prior art to be rare," wrote this patent attorney a few days ago. "Exception: In internet art units certain examiners definitely have their "everything under the sun on the internet" patent that is 1000 paragraphs long that they cite all the time."

This class of patents is unfortunately far too common. Why are such patents still being granted?

Several days ago Professor Michael Risch wrote about why many terrible patents get granted. Risch is worth reading because he is not part of the patent microcosm and at the same time he's familiar with the subject at hand. Here's his latest: [via]

Michael Frakes and Melissa Wasserman have gotten a lot of mileage out of their micro data set on patent examiner behavior over time. Prior work includes examination of grant incentives, agency funding, time availability, and user fees.

Examiners at the Office would be wise to reject by default (if merit cannot be found), but the way things are set up, they get more money and credit when they grant questionable patents. This, suffice to say, is a problem. ⬆