He says :
First there are those thousands of ideas and technologies that are being developed by Google employees in the 20 percent of their week devoted specifically for that purpose. That number of new ideas is far too high to be practical and too high even to be considered safe.
Say the Google Geniuses come up with 4,000 business ideas or technologies per year, which is probably around the current number. Let's guess that one percent of these ideas are truly great -- boffo ideas that one could easily build a company around. That's 40 world-beating ideas. And after the 40 absolutely top ideas, let's say there are another 360 ideas that are pretty darned good -- certainly good enough to pitch to the venture capitalists on Sand Hill Road. The remaining 3,600 ideas are, of course, crap, and can be forgotten by everyone except their inventors, whom I'll get back to in a minute.
Assuming Google executives have the insight to know which of their 4,000 proffered ideas are world-class, they then have to decide which 40 to pursue. Such an assumption is giving even Google too much credit, believe me. No matter how smart they are, issues of loyalty, prejudice, and the odd hangover will result in some less-good ideas making it through but mainly great ideas being eliminated. But this doesn't really matter because the larger question is: How many good ideas can Google pursue vigorously per year? The number isn't 40. It isn't even 20. The number of ideas that a company the size of Google can throw all its weight behind per year is about 10, of which five will probably not be the right ones.
I'm not so sure ... "too many good ideas" looks like the kind of problem that Google ought to know how to turn into an opportunity. (Maybe someone's already is using their 20% time in this direction?)
Firstly, "sort the good stuff from the rest" is Google's raison d’être. There must be dozens of variations on internal prediction markets / social routing networks / ranking algorithms which Google's internal community could use to discover the great ideas.
Second, Cringely's "the number of ideas that a company the size of Google can throw all its weight behind per year is about 10" smells suspiciously like "hit" based, rather than long-tail, thinking. This is a company that specializes in a) having smart, entrepreneurial people inside, and b) supporting large, liquid markets for micro-businesses (via AdSense, Google Checkout etc.) Why, having encouraged 4000 flowers to bloom, would it force them through a vicious filtering to become an official Google product? The answer to the problem of everyone has a product they want to launch ought to be "launch them all", with various degrees of Google official endorsement for those which the company wants to promote.