A Case For The Total Abolition Of Software Patents

This paper makes the case for the total abolition of software patents, based on their lack of promotion of innovation. This abolition may seem like a very distant policy possibility, but this paper is concerned with principles, not feasibility - the star we should be shooting for, not the design of the rocket.

One immediate question is: how does one define a software patent? Where is the boundary? Various suggestions have been made1, but actually, this question is not as important as it appears, for two reasons. Firstly, if we can demonstrate that there is a group of clearly identifiable patents which are harmful, or harmful when enforced in particular situations, then we can adopt the principle that such patents should not be granted or should not be enforceable, and where one draws the exact line between them and other patents becomes a secondary, practical, definitional issue beyond the initial principle. Secondly, some methods proposed for dealing with the problem of software patents do not actually require one to define what a software patent is. For example, one proposal is that one could change the law such that no program written to run on a general purpose computer could ever be said to be infringing a patent. In this case, you need a definition of “general purpose computer”, but you don’t need one for “software patent”. Given these two points, I don’t intend to spend time on definitional issues.

Innovation

This paper examines the benefits or otherwise of software patents via the lens of innovation - whether they help or hinder it. There are other measures one could use - for example, whether they promote the profitability of one country’s industries over another’s - but those are not concerns of ours. Increased innovation is both a major internet policy goal, and also the explicitly stated goal of patent laws in the US and many other jurisdictions - “To promote the Progress of Science and useful Arts”.

Individual Effects

In order to examine the circumstances in which a patent on software could be good for innovation, the following table lays out the path by which a patent could possibly be positive in this way, along with all of the other outcomes along the way which would actually have a negative effect on innovation, or lead to no noticeable positive effect.


The Narrow Road to Patent Goodness

Traps: Ways Badness Can Happen

Person or company has innovative idea

Person or company has non-innovative idea and patent office allows a patent anyway

Inventor pays to patent idea

Inventor has insufficient funds to pay for a patent, and so cannot take advantage of legal protection offered

Inventor implements idea in product

Inventor discovers idea is not necessary for the product they end up making after R&D is finished, and patent fee is wasted

Product is better than it would otherwise have been, or exists where it would otherwise have not

If this criterion is not met, there has been no useful innovation

Patent prevents other company or person using the same idea, with inventor taking them to court if necessary

  1. No-one else wants to use the idea (patent fee wasted); or
  2. inventor does not have time and money to sue other users of patented idea (patent effectively useless in protecting inventor); or
  3. inventor sues but other party counter-sues using their patent portfolio to force a cross-licence (money wasted and no protection gained); or
  4. inventor sues but patent invalidated (inventor ends up paying for patent office and/or lawyer incompetence); or
  5. inventor sues and wins but damages awarded do not cover time and effort used in litigation

Product is more profitable than otherwise, by at least the monetary and time cost of both the patent filing and any litigation

If this criterion is not met, the inventor is not economically better off through having had a patent, which makes filing it an unwise choice

Patent is broad enough that it cannot be simply worked around

If this is not true, no competitor is restricted and the patent fee is wasted

Patent is narrow enough that it does not prevent innovation in related fields or in fields where inventor is not producing products

If patent restricts related innovation, then its value in "promoting innovation" becomes questionable

Patent is not used for customer lock-in or other consumer-negative behaviours

Patent is used to limit or tax interoperability in an industry standard

On expiry, patent is useful to world in teaching patented technique

Patent is worded in such a way that no useful knowledge can be gained from it, or the technique described is by then obsolete


Each patent which makes it down all or most of the road could be seen as a net positive for the case for software patents. Each one which falls into one of the traps is a net negative of some magnitude, depending on the particular trap in question. The question then is: how many software patents travel the narrow road to patent goodness, vs. the number who fall into one of the many traps? What score does the system as a whole acquire?

I say “could be seen” above because even though a patent which makes it all the way down the road - a “positive patent” - will promote innovation, it may have other negative effects in other areas. For example, because patents confer (small and localised) monopolies, albeit for a limited time but a very long time from a software perspective, you can get the problems associated with monopolies - reduced consumer choice and increased consumer cost. Whether you get them depends on the dynamics of each market and is a complicated question, but it is reasonable to say the the conferring of monopoly makes these negative outcomes more likely than if a monopoly is not conferred. So even the patents which have positive effects on innovation may have negative effects elsewhere.

Systemic Effects

The effects noted above are individual effects - measured differently for each patent on a per-patent basis. But before we make our overall assessment, there are also systemic effects which need to be considered - those which arise due to the existence of the system as a whole, and of the body of patents as a corpus. These are almost exclusively on the negative side. One can imagine a possible systemic positive if the collected set of patents became a repository of knowledge which was useful for software engineers in mining techniques and ideas for future development. But given the speed of innovation in software and the long lengths of patents, and the fact that software engineers are generally highly discouraged from reading patents (certainly live ones), this systemic positive has not emerged and seems very unlikely to.

The following systemic negative can be observed, over and above the individual per-patent negatives caused by patents falling into the traps above:

  1. The cost in time and money to the industry of patent-related hassle: patent searches, patent clearance, wrangling over patent policies in standards bodies, and interventions in the patent process to prevent poor patents from being granted;
  2. The cost of defensive patent filing and litigation, where people have to spend tens of thousands to millions of dollars just to acquire weapons and deploy them in an effort to stay in business;
  3. The fact that software as a field has so many niches and facets that patent examiners are unable to acquire sufficient knowledge of the field to properly apply the inventive step test;
  4. Open source implementations, with all the positives they bring, are often excluded from any market where patents are present because their distribution model is incompatible with common licensing fee structures;
  5. The slowness of issuance means that independent invention happens multiple times; all patents effectively become submarine.

The costs or other negative effects of each of these points need to be considered in an overall assessment of the whole.

Why Is Software Special?

I remember attending a software patents event at the start of this century, when someone refuted a point I had made, saying “Well, unless you want to argue that software is sui generis…”, continuing in a way which made it clear that he thought that was a foolish notion. Some of the points above might be said to apply to all patents - so what makes software special?

Software, if not sui generis, is at one end of a very long continuum, and it’s the end where patents have least positive and most negative value. That continuum could perhaps be labelled “cost of development per unit idea”. The far end of the continuum might be a pharmaceutical, where the cost of development of a single drug treatment (or “idea”) can be in the hundreds of millions of dollars. The near end of the continuum is software, where the cost of development of a single idea might be $100 - half an hour of the time of a decent programmer presented with a new and interesting problem. A “product” in the pharma industry contains a single idea, or perhaps two or three at most. A “product” in the software industry contains thousands upon thousands of ideas. Or to compare to a different industry - it costs say $50M to build a car plant to make a car, but $50K to build a software system with the same number of parts as the car has. Same number of ideas, 1000x smaller cost.

One could also imagine another continuum, labelled “likelihood of independent invention”. The idea of a patent is that it promotes innovation because the one person who has a great idea doesn’t keep it secret, but publishes it for everyone to read, and in return they can’t use that information until the patent expires or they get a licence. But if people are independently inventing the same ideas without help from the system, then the system is not necessary. Again, pharmaceuticals are at the far end of the continuum here - independently working out that imanitib mesylate is a good treatment for leukemia2 seems highly unlikely, and to work it out independently and conduct the necessary trials to prove it would be highly expensive. By contrast, deciding that when implementing a spreadsheet, you should make a data structure indicating which cells depend on the values of other cells and recalculate the cells in that order seems fairly common sense3. Two people independently implementing spreadsheet software may well come up with this solution to the problem without the need for one to read (and license) the patent of the other.

So patentable ideas are on one or more continua. However, the patent rules for all points on the spectrum are the same - pick an idea, write it up, $10k or so to file, 20 years protection. And international agreements like TRIPS make it hard to vary these conditions to better match the situation.

Software, being at one end of these continua, is the clearest case of a field in which patents are unnecessary and unwelcome. Where and how the patent system draws one or more lines to create different conditions for different circumstances is an implementation detail; but this paper asserts that software is a clear-cut case.

The Future

Those in internet policy who “fight for the user” might also be said to be those who try to be a focal point for the prevention of individually-small but large-in-total distributed losses in the face of concentrated gains. According to Boldrin and Levine:

[T]he lobbying effort and power of IP’s advocates are bound to be much stronger than those of IP’s opposers. This follows from the fact that, once some kind of even marginal IP protection is introduced, extending it will yield substantially higher per-capita rents to the few holders of the right than reducing it would for the much larger number of non holders: the rent of the monopolist is a lot higher than individual consumers’ deadweight loss.

The losses from the patent system are distributed among billions of users (indirectly), the open source community, and innovative individuals and small firms. The gains are concentrated in the hands of patent lawyers4, USPTO employees, patent trolls and large companies. (Although even the latter category is split on the usefulness of software patents, with their opinion generally being related to how many of them they hold.)

Boldrin and Levine again5:

The initial eruption of small and large innovations leading to the creation of a new industry – from chemicals to cars, from radio and TV to personal computers and investment banking – is seldom, if ever, born out of patent protection and is, instead, the fruits of highly competitive-cooperative environments. It is only after the initial stages of explosive innovation and rampant growth end that mature industries turn toward the legal protection of patents, usually because their internal growth potential diminishes and the industry structure become concentrated.

They go on to explain that pressure for a stronger patent system comes from older firms, not from new innovators. Therefore, even if we currently think that software patents are a nuisance to innovation but not a substantial brake, over the next decades as the software industry becomes more mature we can expect to see pressure to strengthen patent protection grow, and patent problems become more numerous. If we are going to abolish software patents it needs to happen soon, before those countervailing forces become too strong.

Conclusion

Very few software patents make it down the road to patent goodness. The vast majority cost the company money to file and then lie gathering dust, acting only to provide a vague chilling effect on innovation in that area for those brave enough to do a patent search. A few become famous and tie up an enormous amount of industry and lawyer time and money. And almost none achieve that fabled goal of protecting the small inventor from the large rapacious company which wants to “steal his idea” and leave him penniless. If you add to that poor hit-rate the negative systemic effects from having a software patent system, it seems to me that the conclusion becomes obvious. Innovation in software proceeds at a rapid pace, and does not need the “encouragement” of patents. Software patents are a problem for the industry now, and will only be a bigger one in the future. We should work to end them.


[1] E.g. the one in the never-passed USA SHIELD Act.
[2] http://www.google.com/patents/US7300938 (Gleevec)
[3] http://www.patents.com/us-4398249.html
[4] No offence meant to any patent lawyers reading this document :-) But it remains true that if there were no patents, you would need to retrain into a different part of law.
[5] https://research.stlouisfed.org/wp/2012/2012-035.pdf

Original URL: http://www.gerv.net/writings/phinehas/