The Register has a disappointing news story Patent Quality has fallen, confirms Euro examiners. It reports that nearly 1,000 patent examiners have written to the European Patent Office’s (EPO) Administrative Council (their board) to say that to follow the management requirements would actually result in them breaching their duty and guidelines.  This is about as damming as it gets for the senior management of the EPO.

The whole patent system relies on patents being a high-enough standard. If low standard patents are issued, the innovation-chilling “land-grab” aspect of patents overcomes the innovation-stimulation aspect. Many credible commentators on IP and patents think that the innovation-stimulation aspect does not work at all in areas where research, testing and tooling-up are cheap, such as with Software products.  We used to think that the only people who bad patents benefited were IP lawyers (bless ’em: necessary!), but it seems that we can add patent office executives to that list too (also very necessary people.)

What we saw in the .COM bubble was that bad IP creates bad decision making and bad investments.  Never give a sucker an even break. More particularly, it diverted attention away from real innovation.  It disenfranchises honest people who come with an idea, then regard it as too obvious to be patented, only to have someone subsequently patent the idea.  Lay juries are easily bamboozled, and the language of patents is so far removed from technical language of the particular areas as to render technical experts useless.

Of course, the software patent industry now has nothing to do with innovation, and everything to do with getting a portfolio of confusion: bad patents actually play into this, because the more confusion there is about how low the bar is for a patent, the more uncertainty is available to allow cross-licensing.  This on top of the notorious use of patents to suppress innovation that would damage or be internal competition to a company’s product ecosystem.

It sounds like the Administrative Council needs to get the credibility of the EPO under control. Failing that, the European Parliament needs to act.  The US patent system became a laughing stock because of exactly this same issue, in the 1990s and 2000s, and has been hardpressed to improve: that generation of junk patents are a cancer that continues to afflict business and innovation.  An investor in a successful company can suddenly find their stock value wipe out because of retrospective royalty payments based on some hicks in a jury making a bad call.

I have had aspects of Schematron, which I invented and has become an ISO standard, patented several times, in ways I consider trivial and bogus.  Sometimes the flaw is entirely in mis-statments in the motivation pre-amble, which are not really susceptible to protest:  they will say “Schematron does not allow assertions in Klingon, but we do, therefore we patent this completely excellent set of elements” which actually do nothing other than what Schematron does.  Junk patents waste the time of examiners, who may fail to give due consideration to excellent patents that will enrich the community.

What management is saying is that the time taken to examine a patent should be a constant.  O(k).   But surely the time taken to examine a patent properly must be at least linear to the number of patents. O(n).  Fortunately, computer searching can help so to trim  it; but it is still at least O(n) and therefore prone to be linear: if you grant twice as many patents, you need to double the amount of time examiners spend figuring out what the patents mean.  Especially given that patents examination needs to count expired patents and rejected patents too, to be credible: a new patent for something that was patented 40 years ago is just institutional incompetence.

Having read quite a few patents, and having acted as an expert witness in a multi-million IP patent law-suit, it seems to me that it is quite likely that that adequate patent examination is actually an O(n²) task, because you don’t need to just search every earlier relevant patent and whatever prior art and compare it to the patent-applied technology, you also may need to cross-check between those in order to determine what the patent is about. Patents have their essential claims, but these are so broad that you need to check other granted (or disallowed) patents to see what the concrete innovative steps are that have actually been patented.

The EPO crowed in 2016 that “The EPO granted approximately 95 900 patents in 2016, an increase of 40% over 2015, and the highest ever number. This increase was possible following a series of reforms improving quality processes and enhancing efficiency, which ensure high productivity without compromising on quality.”  In the period 2015 to 2017, there has been an increase in absolute numbers of patents granted by about 50%, but with an increase in applications of  This makes no sense at all, given that patents applications only increased of 3.5% from 2015.  How could an improvement to the EPO processes result in an improvement in the quality of applications? Unless the claim is that previously the EPO spuriously rejected 46% of patent applications?  It simply is not credible.

Alarm bells need to be ringing: it lookes like there is a serious problem here.

The EPO website’s 2017 report says the noble words:

The credibility of the entire patent system is based on certainty, for the benefit of inventors, society and the economy. All applicants rely upon the quality of the EPO’s products to be confident about the decisions they take. This is why quality remains the EPO’s first priority, and why quality is at the heart of all aspects of our business: staff, tools, procedures and law.

What this letter from the patent examiners says, absolutely incontrovertibly, is that this statement is near complete rubbish, and we should give it no credence. Whatever gains there have been (well done) have been swamped by self-serving management metrics: a bait-and-switch on patent application success.  Patent examiners are professional whose lives are dedicated to determining what is factual. They are the ones in more a position to know that any management statistics (the number if patent applications withstanding challenges says absolutely nothing about the actual quality of the result, one way or another, for example), and certainly not customer satisfaction, where the customers are deemed to be the applicants and not the community.

Now I am sure that AI and search technologies can help improve patent searching and processing time. But unless the result is not that more patents are detected and rejected as junk, their success criteria are just wrong. It fails a sanity check. (It fails what we call in Australia “The Pub Test” of propositions that would be regarded with derision by disinhibited but rational fresh people outside the bubble of the people making the statement.)

There are not 50% more ideas in the last two years than any two years previously. It just simply is not humanly possible that we have somehow raised a new generation of super-geniuses who suddenly are flowering with bold novelty that no-one has thought of before.  I could not have more scorn, and neither should you.

The thing is, I appreciate the patent system: I get why and where it can work, and where some temporary monopoly rights are effective and warranted to getting innovation out the door. But we should not be fools  in our enthusiasm for what became a Ponzi scheme in the .COM crisis (the last people holding the junk patents took the loss, and it effectively locked up and buried many good ideas.)  I cannot help but think of Bill Clinton’s 1996 unironic formulation on abortion policy, that abortions should be “safe, legal and rare” when thinking of what an ideal global policy for patents might need to be, as a necessary corrective to an era of unfortunate junk patents.

I also recommend the blog quoted in The Register’s article, from the super-respectable Wolters-Kluwer IP blog.  That blog points out that currently 15% of challenged patents fail the challenge: this is not a sign of success of the initial grant, but a sign of the failure of the process.  They should be looking at an order of magnitude better than that, in order to be credible.  15% wrong in challenged cases may mean even more wrong in cases no-one notices: we simply cannot say, and there is no reason to assume otherwise, apart from making bland assurances.