For an answer here, see

     http://news.ycombinator.com/item?id=5353799

For a more direct answer:

Pick a problem that needs solving. For the solution, use 'innovation'.

Then want the 'innovation' to be ideas, essentially applied research, that can be presented and evaluated usually just on paper. So, as in peer-reviewed published papers, the ideas should be "new, correct, and significant", and, for innovation in business, powerful and valuable for solving the problem. Moreover, we should be able to check these criteria mostly just on paper.

Then we measure the innovation by the ROI of the solution it provides.

Is that the only way to make money? No.

But this approach to innovation has one heck of a good track record for delivering terrific solutions for challenging problems via the US DoD since the start of WWII, and there the batting average is much higher than for information technology venture capital.

Moreover, and definitely should not be lost, and for 70 years has been understood with great clarity by both the US DoD and Congress, the real foundations of such innovation are pure and applied research as in our top two dozen or so US research universities (where about 60% of the annual budget comes from NSF and NIH for US national security and health care) with more in Federally Funded Research and Development Centers (the FFRDCs) and then more at Raytheon, United Technologies, GE, Lockheed-Martin, Boeing, General Dynamics, Hughes Electronics, etc. In total, Congress authorizes big bucks for all that work, and Congress 'gets it'. However, Silicon Valley wants to laugh at research and then turn and run away afraid of it.

US information technology (IT) venture capital has had a 'business model': Wait for 'traction' to be significant and growing rapidly but, still, for some reason or other, the company still not very profitable, and hope that then the founders will need money for the 'big build out', hiring bizdev and marketing people, setting up the sales force in the 'channels', etc. for the 'execution', or be so eager for "cash now" for 'the good life' that they will sign a bad business deal. Alas, as reported at AVC.COM recently, the ROI has been lower than an index fund.

Also there is an incongruity: A small information technology startup with traction significant and growing rapidly is in a much better position than millions of US Main Street entrepreneurs who make it without an expensive 'build-out' with bizdev, marketing, channel people, business expertise from former management consultants, etc.

For venture partners able to evaluate innovation, i.e., applied research, just on paper, YC is exceptionally good. Then there's Wegner at Union Square, Morgan at First Round, Metcalf at Polaris, Bayless at Sevin-Rosen, and a few more. There are, what, all together in all of the US, 20?

And then there is the list of venture partners who prefer to invest in college dropouts.

In college, at universities, at several laboratories, at GE, IBM, JHU/APL, and more, I have worked with many people good at doing and evaluating applied research. Alas in US venture capital, that list of 20 is about all I can find that is comparable.

With the ROI south of the index funds, it looks like there will be some changes that Darwin would understand.