The patent game that's pending disaster

There's nothing as unsettling as the realisation that something you thought you owned belonged to somebody else all along. It's true of a whole load of things -- cars that the previous owner didn't finish paying for, that second-hand stereo liberated by the local tealeaf and flogged through eBay, the present from a friend that wasn't his to give. Such things belong firmly in the class of vague worries we don't like to think about because they're too difficult to guard against and besides, we'd never get any sleep.

But such things happen. Ask the makers of digital cameras, scanners and picture editing programs, who have just been told to cough up or else. The wielder of these threats, Forgent Networks, is waving a wad of patents that it claims covers aspects of the JPEG picture format -- a format that until now was accepted as freely usable. It's an international open standard of some maturity: it's hard to think of something safer. But then, you'd have said the same of the humble hyperlink before BT decided to lay claim to the idea: it doesn't look as if BT will win that one, and it's not clear whether the Forgent claim will stand up to scrutiny. But it'll be an expensive discovery in both cases, whoever wins.

The real question is -- is it possible to write any new software these days that you know won't hit patent problems later? Imagine if the same situation held for the written word. For something to be worth reading, it needs to be novel -- either with new facts, or drawing together old ideas in an interesting way. You may have already thought about the similarities between publishing English prose and publishing software, but I'm betting it's worth putting into print here. It's not a perfect analogy, but it has its virtues.

So far, so good. But if the same laws applied to both English and software, my employer couldn't publish the previous paragraph without checking whether someone had already patented the idea of using analogies between software programming and writing English. And, since it's not possible to see the details of patents pending under US law, even a full check wouldn't prevent future problems.

This would leave publishers with a tricky choice: either publish and be prepared to absorb the cost of any subsequent licence claim or court case, or not publish at all. Even if a publisher adopts the Wellington strategy and loudly declaims "Publish and be damned", you can bet your bottom Enron share that the potential liabilities had better appear on the balance sheet somewhere.

Here's the real well of poison. After years of battling with the arrant stupidity and greed of big companies sticking small companies for royalty payments on dodgy software patents, we know that grotesque absurdity is no defence for the small guys. No, they have to go to court and risk death by legal fee. Would you invest in a small software company knowing that at any moment, they could be hit for arbitrary licence fees and restrictions, and that there's no reasonable way to guard against this?

In an economy where any hint of undisclosed liability is more poisonous than a badly cooked puffer fish, this realisation will be death to anyone hoping to innovate their way to success. Patents were designed to protect and encourage innovation. A good patent makes a good idea public while protecting the inventor for a limited period of time: nobody could make that claim of Forgent's action.

You can't in all honesty blame those trading in patents and hoping to put the squeeze onto hapless victims. It may be distasteful, it may be dangerously short-termist, it may even be immoral -- especially in the parallel world of biotechnology and genome therapy, where the consequences of patent-powered monopoly are truly life-threatening -- but it ain't illegal. Even vultures have to eat, and there is any number of dot-com shells lying around where the only asset left for the investors is intellectual property.

The solution is to sort out software patents. The current situation is a mess, caused by companies cottoning on that the US Patent Office would grant software patents without fully understanding what was being discussed or doing a proper check for prior art -- the get-out clause that says you can't have a patent on something someone's done in public. There are uncounted thousands of the things lying in the mud of the law like mines, ready to explode under anyone unlucky enough to tread on them in the course of their intellectual prospecting.

To defuse them means to accept that the Patent Office got it wrong, be far more rigorous in issuing new software patents and come up with a saner way of checking the viability of existing ones. In particular, a way to limit the liability of those who fall foul of bad patents needs to be thought up, and fast. Otherwise the recovery of the IT industry from the miseries of the moment will have to be put on hold, perhaps forever. How's that for an unsettling realisation?

To have your say online click on TalkBack and go to the ZDNet UK forums.