Google’s First Software Patent

by Tim Lee on April 24, 2006 · Comments

A couple of weeks ago, Patrick Ross mentioned a talk given in Brazil in defense of software patents:

GWU Law Professor John Duffy defended software patents. A good example of a software patent, he said, is Google’s first patent. It patents their search approach, which starts with a basic search, then involves a search within the search results based on popularity. That innovation has led the company to a market cap of $123 billion, but couldn’t be protected by copyright because the method could be written in a thousand different ways.

Intrigued, I wrote him asking for details. He kindly pointed me to Duffy’s PowerPoint slides.

I’m not impressed. Here, literally, is the slide making the argument about Google:

Patents can make you rich and famous.

Example: Google = $123 billion

Google is built on a a few key software patents.

Now, it’s probably not fair to judge a presentation from its PowerPoint slides. I’m sure Duffy’s presentation was far more nuanced than this simplistic rendering suggests. Still, even taking that into account, this is pretty weak sauce. You can’t just attribute Google’s market cap to its patent portfolio, as though that settles the matter. Duffy’s slide presents no evidence whatsoever that Google’s patent was key to its success. It’s the crudest kind of post hoc ergo proctor hoc fallacy.

It’s hard to see how Google’s story would have been any different in a world without software patents. The slides offer no evidence that any companies have actually been deterred by the existence of Google’s patent from producing a Google knock-off. If you examine the patent in question, it’s far too general to be of much use to someone wanting to do so. What makes Google so great isn’t the general concept of its search, but its superb implementation. They were the first company to take the general concept of treating links as votes and implement it in a way that could scale to billions of pages.

But of course, implementation details are protected by copyright. Which means that in a world without software patents, Google would have gotten along just fine.

Moreover, even if we assume that Google’s market cap is partly attributable to its patents (which I think is unlikely) that doesn’t prove that it’s a good thing. There are all sorts of government programs that make corporations wealthy. Ethanol subsidies have been good for ADM. Defense spending has been good for Northrop-Gruman. New York taxi medallions have been good for those fortunate enough to have one. The policy question isn’t whether these programs serve the companies that benefit from them, but whether they promote innovation and economic growth. I can’t see any reason to think that the availability of software patents makes future Googles more likely. To the contrary, the primary effect of software patents appears to be a kind of rent-seeking, wherein software trolls obtain software patents to extort legitimate companies, and legitimate companies are forced to acquire software patents (and retain lawyers) to defend themselves.

There might be an example of a legitimate software patent out there, but this sure isn’t it.

Comments Posted in: Patents

  • As I wrote on my blog, the Java and .NET divide proves that software patents are not a good thing for software development. Another major fallacy of the argument that Ross and the other guy put forward is that closed source applications don't need patent protection because they can in most cases reliably hide behind trade secret protection. The only people who could actually see Google's algorithms would be the people working on their products who could be bound by a NDA and other measures to keep silent.


    You know what's really funny, Tim? Ross and the others there missed the fact that the real innovation is in how Google made their search system so powerful, flexible and scalable.

  • Cog
    In all likelihood, some variation of PageRank's key ideas are now standard subroutines in every major search engine. Yet I am unaware of any patent licensing fees paid to Google by Microsoft, Yahoo, or Amazon. This doesn't mean that no such fees are being paid; on the other hand, I suspect that the existence of such licensing agreements would have leaked to the public by now. (Certainly, if Yahoo's paying PageRank license fees to Google, then Yahoo would be criminally negligent not to state in their SEC filings that they are paying license fees on their core business to their most successful competitor.)

    Furthermore, Google's revenue is clearly based on ad sales. Patent licensing fees, if Google's even collecting any, constitute an eyedropper-size trickle next to the firehose of ad revenue. Duffy's crazy to give the PageRank patent credit for Google's market valuation. The existence of the PageRank patent might make investors feel better, but its business utility to Google beyond psychological palliative is questionable.

    Finally, as a technical matter, I have heard firsthand from a developer of Nutch (an open source search engine) that PageRank unadorned doesn't get you very far, and that you need to toss in a bunch of other tricks to make your results reasonable. PageRank gets a lot of attention from computer scientists because it's mathematically elegant, and from the public because the core idea captures the imagination. But its importance is probably overestimated.
  • I'll confess, Tim, when you asked for these materials I wondered why you didn't go directly to Professor Duffy; after all, I was just an observer of his Sao Paulo presentation, and as a college professor he is hardly difficult to track down. Now I see why -- you wanted to ridicule his example without having to get the full argument straight from the horse's mouth, nor did you want to have to approach the person you were planning to attack. A blog is not journalism, I know, but the journalist in me is appalled at the recklessness and intellectual irresponsibility of this blog entry, particularly when you acknowledge it's "not fair" to base your criticism on a single PowerPoint slide. When you want to launch a kamikazi attack on a respected academic in the future, do it without my help.
  • Tim
    I contacted you because I happened to have you in my address book, and was already planning to write to you to follow up on the Linux DVD player question. I was also hoping you might have a transcript of the talk and/or a past PFF publication making his case. And I figured you'd give me his email address if you thought it would be better to contact him directly.

    In any event, I've emailed him and look forward to getting more detail about his argument, and if I've misrepresented that argument, you can expect to see an apology in a future blog post.

    Look, I don't have the luxury of working on technology issues for my day job, so I blog in my free time. I just don't have time to track down every potential angle as I would if I were writing as a professional journalist. Whenever possible (as in this case) I try to acknowledge cases where I might be making an off-the-cuff generalization. And when I make a mistake or draw a hasty conclusion, as I have on several occasions in the past, I acknowledge that too and apologize for it. But if I waited until I had time to research a subject exhaustively before posting anything about it, I'd hardly ever be able to blog at all.
blog comments powered by Disqus

Previous post:

Next post: