Surely that only works if the probability of winning a case depends only on the skill of the lawyers, and not on the actual facts of the cases. I imagine a lawyer with no training at all could unravel your plan and make it clear that your hobos had nothing to back up their case.

If facts totally determine the case, then my exploit doesn't work but Eliezer's radical change is equally irrelevant. If facts have no bearing on who wins or loses, and it is purely down to the lawyers, then Eliezer's system turns lawsuits into a coin flip, which is only an improvement if you think that the current system gets things right less than 50% of the time, and you'd also have to show there would be no negating side-effects like people using my exploit. If facts determine somewhere in-between, then there is a substantial area where my exploit will still work.

Suppose I have to put up a minimum of 10k for each hobo lawsuit asking for 1 million; then I need only have a 1% chance of winning to break even. So if cases with lousy lawyers on both sides wind up with the wrong verdict even 2% of the time, I'm laughing all the way to the bank. And it's very easy for bad lawyering work to lose an otherwise extremely solid judgement. A small slipup might result in the defendant not even showing up, in which case the defendant gets screwed over by the default judgement against him. Even the biggest multinational can mess up: consider this recent case where Pepsi is contesting a $1.26 billion default judgement which was assessed because a secretary forgot a letter. They probably won't have to pay, but even if they settle for a tiny fraction of 1.26 billion, how many frivolous lawsuits do you think one fluke like that could fund?

For that matter, consider patent trolls; they have limited funds and currently operate quite successfully, despite the fact that they are generally suing multinationals who can spend far more than the troll on any given case. How much more effective would they be if those parasites could force their hosts to mount a far less lavish & effective defense than they would otherwise?

Also, being English myself, it hadn't dawned on me that the losers-pay rule doesn't apply everywhere. Having no such system at all seems really stupid.

I did some reading; apparently it's long-standing tradition all the way back to colonial times. The author said the Americans likely wanted to discourage litigation, which I suppose is the polite way of saying that early Americans were smuggling indebted IP-infringing scofflaws who didn't want civil justice to work too well.

It also occurs to me that hiring expensive lawyers under losers-pay is like trying to fix a futarchy: you don't lose anything if you succeeded, but you stand to lose a lot if you fail.

I don't really follow?