Patently stupid...

Like anyone, I have my fair share of stop-the-world-I-want-to-get-off moments. But I really did have to check the calendar to make sure it wasn't April 1st today when I read through IBM's patent application for their "solution for providing real-time validation of text input fields using regular expression evaluation during text entry" (20090132950). Wading through all the verbiage, it really does appear that IBM think they have "invented" evaluating a regular expression inside an onChange event handler.

In terms of this specific patent application, I don't think many programmers are too worried that they're suddenly going to be prosecuted for calling RegExp in the wrong place-- needless to say, plenty of prior art is being cited, in case it were needed.

But the case does leave some other more interesting questions. Why does IBM of all companies think it needs to stoop this low? And how can it help us ditinguish the "bleeding obvious" from the "possibly patentable"?

I'm really at a loss to answer the first question. Possibly we're simply looking at a clerical error on the part of an overenthusiastic junior employee. But for a company that is supposedly at the forefront of computing research and invention, claiming patent rights on what amounts to a chain of JavaScript API calls doesn't really help them uphold their reputation.

In answer to the second question, I'm reminded of a comment by Donald Knuth, that he regularly sees patents granted to "solutions" to problems that he sets as undergraduate homework questions. If "how do you validate input as the user types into a web form?" had been posted as a question on one of the various Internet programming forums (and probably it has...), it would almost certainly have been tagged as "smells like homework". Few programmers, I suspect, would have tagged it as "goodness me how novel-- I think we should patent this".

I really have no idea how in the field of software patents you can concretely separate "significant invention" from "doing the bleeding obvious" (though, as in this case, I can sometimes recognise the latter when I see it!). But in order for a software solution to be patentable, I would at least expect to see:

- evidence of significant empirical research necessary to find the solution
- an invention of an actual algorithm, with perhaps some guage of number of API calls per other instructions/lines of code in order to determine if a new algorithm had actually been invented.

I would also like to see severe financial penalities for cases such as this, where a company is clearly attempting to use its abundant resources to abuse the system. The small developers that the patent system was originally designed to protect really have to think twice before committing to the thousands of dollars per year that it costs to keep a patent going. A company such as IBM really has nothing to loose by continually paying for nonesensical patent applications out of petty cash on the off-chance that at some point, one of them will sucessfully fly over the cuckoo's nest.

In the case of this particular patent application, I really really hope for the sake of human sanity that common sense fianlly prevails...