Joff Wild

The good news is that after my blog on Friday about the wholly inaccurate figures supplied by End Software Patents concerning the cost of software patent litigation in the US, the organisation has removed the offending stats from both its study and its press release. The bad news is that it has replaced them with much higher figures that it fails to explain or to contextualise; so that now we have a claim that the annual cost of software patent litigation in the US is actually over $30 billion.

End Software Patents gets to this amount by relying on work done by James Bessen and Michael J Meurer. This involves accepting that Bessen and Meurer are correct in assuming that 25% of all US patent litigation is software-related; and then accepting a subsequent claim by the two that the average cost to a company involved in patent litigation was $28.7 million in 1992; an amount which equates to $43 million in 2008. By multiplying the number of alleged software suits by 43 million, End Software Patents gets to $30.4 billion.

What End Software Patents fails to point out is that Bessen and Meurer make no distinction between the types of patent they are talking about. Instead, their figures relate to all patents – from software, through mechanical to pharmaceutical. Crucially, the organisation also fails to state clearly that the amounts quoted by Bessen and Meurer actually relate primarily to falls in market capitalisation when suits are announced, not to the direct costs of fighting suits or to the lasting effect such litigation has on a company’s share price or its financial performance. And at no stage do Bessen and Meurer examine any potential benefits in engaging in patent litigation, such as the maintenance of market share, the protection of product lines or the generation of licensing revenue.

So, when End Software Patents states in its revised press release that “Software patent litigation costs US companies $30.4 billion annually”, what it actually means is this: “We have taken the findings of a study that looks at the overall cost of a patent suit being inaugurated in any field of technology to extrapolate that software patent litigation costs US companies $30.4 billion every year. We have done this by accepting that if you look at the behaviour of share prices when a suit has been announced you will see a dip amounting to an average of one half of one percent of overall value. What we have not done is differentiate between various types of patent to see, for example, whether there is a difference in what happens to the share price of a pharma company that is engaged in litigation over a blockbuster drug and the share price of a large technology company that is sued by a patent troll. Furthermore, we have not reported that the median cost found by Bessen & Meurer was actually $2.7 million and we have not sought to explain why the median is so very different to the mean or to consider that the difference may actually be accounted for by the way markets react to litigation in the pharmaceutical industry, where the actual or potential loss of a patent relating to a blockbuster drug can see major falls in a company’s share price. And on top of all of that, we have decided to assume that there can be no positives in getting involved in litigation and have decided that the costs it creates are nothing more than a waste.” All of which is less snappy, obviously, but maybe a bit closer to the truth. 

The bottom line is that even if you accept the findings of Bessen and Meurer without question, it is very difficult to see how you can, in good faith, get to a statement that: “Software patent litigation costs US companies $30.4 billion annually.” Bessen and Meurer do not say this and they do not provide figures that can allow anyone else to say this. What they do show is that patent litigation in the US can be very expensive, not only directly but also indirectly – which will hardly come as a surprise to anyone.

A pharma company that loses one half of a percent of its overall market value on the announcement that it is being sued over a patent it owns could see its value drop by tens of millions of dollars. But what has that got to do with software patents? Likewise, we know that the average share price drop quoted by Bessen and Meurer amounts to one half of one percent, but do we know how that breaks down by industry? I have looked, but I do not see them providing any answers.

None of this is to say that software patents are a good thing or that it does not cost much to litigate them. Clearly, there are major concerns about the extent to which they have been granted in the US and the expense involved in defending claims of infringement, especially when any transgressions are usually unintentional.

What does concern me, however, is that we have in End Software Patents an organisation that is in search of easy headlines and so is not overly concerned with producing accurate figures. Instead, it seems to me, it is looking for big numbers that will help it to get publicity. It knows that most of the journalists and editors who read its press releases will not read its study; but it knows that, even if they do, they will not have spent much time covering IP issues and will be unlikely to find anything wrong in what is being claimed. In this way, the big number becomes an accepted part of media coverage and the debate is distorted. I do not think that is the right way for any organisation to behave. If there is a strong case against software patents, and many would say that there is, surely it can be made through reasoned argument and reliable data. I am afraid that so far I have not seen either from End Software Patents. In my eyes, therefore, it is an organisation that currently lacks any credibility.