i think the most interesting takeaway, that none of the comments touch on, is PG thinks there are a finite number of good ideas (even though this is currently larger than the number of good startups.) this probably means that (he thinks) that the number of good ideas is a simple function of the number of technological breakthroughs in a given year. so, in fact, it might be possible to comprehensively list all of the good ideas, in a RFS, for example.