But Not Every Fund makes $5 for every $1 invested. A fivefold payback is, in fact, remarkable, even in the extraordinary world of VC. According to Venture Economics magazine, the typical 20-year average industry return is around 16 percent annualized—still not bad nowadays, when the average return from stocks and bonds is in the single digits.

Now consider that this 16 percent return comes from a blended average of successes and failures. A general partnership might invest in 10 to 20 companies. The VCs, of course, are betting that the successes will more than pay for the failures; in fact, their idea of a successful company is one that generates a 10-fold return on investment within five years. That's equivalent to a 58 percent return per year.

With that definition of a home run, VCs can hit a few foul balls. If a successful investment gives you a 58 percent annual return, you could have three failures as well and still do better than the VC industry average. You can reduce your risk further by investing a limited amount of money, often as little as $5 million or $10 million, in the earliest, high-risk, high-reward funding rounds and putting in bigger bucks in later rounds when a company is beginning to look like it will succeed.

All this structure seems designed to maximize investment in true innovation. Yet, as our study showed, the very opposite is true. We believe there are four basic reasons that innovation often gets short-circuited.

1. A venture fund has a life cycle.

VC general partners don't work with an amorphous pile of cash; they manage discrete venture funds. For example, in 2003 the Kleiner Perkins Caufield & Byers XI fund was created—the 11th since the firm began. Each fund has a different set of general and limited partners. To protect against conflicts of interest—that is, to make sure fund XI doesn't bail out failing investments from fund X—most VCs won't invest in two different funds if the same company is funded in both.

The general partners of each fund need their investments to pay off within the fund's life span, which is six or seven years. So the general partners look for investments that can generate revenues in two to four years and break even soon after. Their ideal scenario is one in which they sell the company for a lot of money, or it goes public for a lot of money, within the fund's lifetime.

This short life cycle for venture funds has dramatic consequences for innovation, none good. Typically, when a fund invests in a new company, it needs to reserve additional funding for up to three follow-on rounds, just in case the start-up runs out of money, which is inevitably the case. When you total up the bills for management fees; accounting, legal, and other expenses; and reserves for the follow-on rounds, an initial funding round of $200 million to $300 million might involve a $1 billion commitment over seven years. If that $1 billion sum represents the entire fund, the VC may need to start a new fund to invest in even more companies beyond the ones in the previous fund. Thus, the gap between one fund and the one that immediately follows it might be only two to three years.

That's a big problem, from an innovation standpoint. To raise fund [n + 1], institutional investors are going to look at the interim results for fund [n].

They want to see the start-up company booking substantial revenues or showing other signs of progress, such as contracts awarded or design agreements with major customers. In other words, VCs need to see a company end its start-up phase and become a real business in three years at most.

The upshot is that VCs won't look favorably on funding proposals involving years of research—regardless of the potential payoffs. It's not that they are not interested in innovation. They just won't fund innovation that takes time.

A good example from our own current portfolio is InPhase Technologies, in Longmont, Colo., which is developing holographic storage media and hardware. The company had its technical inception at Bell Labs in 1995 and was venture funded in 2000. This past January, it showed a prototype 5-inch drive. Its first product, expected in 2006, will be a 300-gigabyte removable disk cartridge that will be the same size as a DVD drive but will hold 60 times as much data. But the company will have taken 11 years to hang its first dollar on the wall, which is 7 or 8 years too many for InPhase to be attractive to the denizens of Sand Hill Road.

In funding InPhase, we decided to forgo the traditional short-term window, because we saw a huge potential payoff for investors. Naturally, there's commensurate risk. Corporate research labs, such as those at IBM or Pfizer, and U.S. government agencies, such as DARPA or Sandia National Laboratories, can take this long-term view, but it's rare for venture funds.