Precisely - the difference is between "worthless", which we instinctively know isn't true, and "won't pay money for":

http://journal.dedasys.com/articles/2007/04/26/ideas-are-worthless

In short: for practical purposes, PG is right, but wrong in his theory.

"How would a market for ideas work, in any case? How could you prevent someone from listening to a bunch of ideas and just walking away with the ones they like?" -- your blog post

Secrecy and contracts?

Which makes the transaction costs higher than the idea is probably worth. Execution is very important, and in an ideas market, there would probably also be a lot of crappy ones. So, you have to sign a bunch of NDA like contracts, risk being sued if you ever develop anything like the things you may have seen, and probably also have to wade through a lot of junk.

While it might be sort of doable, it's probably not worth it until we have a magic pill that lets you forget the last 5 minutes of your life completely and totally.

"Which makes the transaction costs higher than the idea is probably worth."

These costs could surely be lowered if there was demand for ideas in the first place.

I think this is more evidence that ideas by themselves are generally worthless.

It's evidence that the value of ideas, on aggregate, is low (which makes sense), but doesn't point to a value of 0, just lower than the transaction costs inherent in the attempt to create a market for a more or less non excludable good. Markets do best when there is sufficient information, and so a market based around secrecy and potential lawsuits isn't going to be a very good one. Look at software patents, for instance... it's not a very nice or healthy market.

see

http://news.ycombinator.com/comments?id=24338

do a google search on "patent trolls"

I don't think patents are legitimate.

Please tell me why you think they are if you disagree.