2. VCs act like businesspeople, even when they have a technical background.

Engineers who work with VCs for any length of time are inevitably frustrated by what they see as the VCs' limited ability to understand revolutionary technology. Combined with the VCs' strictly bottom-line orientation, the result is an inability to accurately access technological risk.

In fact, it would be difficult to argue that VCs are ignorant of engineering and other technical areas. A review of the backgrounds of 180 general partners at 20 leading VCs shows that 64 percent of general partners have undergraduate engineering degrees [see below table, "A Matter of Degrees"].

But 64 percent also have MBAs, while only 29 percent have master's degrees in engineering or science. So by a wide margin, it seems that the business training of the average general partner exceeds his or her understanding of technology, and that for the people who have both, the technical background supports a business outlook, not the other way around.

As a result, most VCs are more comfortable with business plans that are logical extensions of existing technologies. They're also good at conducting reams of due diligence. The typical VC firm today has lots of junior associates who love poring over market and growth projections and cash-flow forecasts.

VC investing is all too often a mechanical process of reviewing business-school checklists. The dearth of venture capitalists who can really understand fundamental research and who eagerly talk to brilliant researchers with exotic, extraordinary ideas is one of the key challenges facing the industry. Unfortunately, the average Ph.D. scientist or engineer knows little about business, and in our experience, most VCs really don't want to talk to people like that.

VCs do have venture partners with technical backgrounds, whom they can call on for due diligence advice. However, these people are generally consulted only when a project gets past the initial screening process. Many innovative technologies are rejected well before then.

The case of the physicist David Huber shows that while mainstream companies can't recognize a good idea when they see it, neither can venture capitalists. In the early 1990s, when he was with General Instrument Corp. (now part of Motorola Inc.), Huber developed some interesting technology for multiplexing different wavelengths of light onto a single optical fiber. General Instrument decided that the technology was "not strategic" with regard to its business plans and gave Huber 18 months to find a buyer for the group or be shut down. In the end, the group did not shut down (partly due to our intervention), and the company eventually went public. But it came perilously close to extinction, reflecting the level of risk aversion that prevails at most VC firms.

Huber's start-up was Ciena Corp., of Linthicum, Md., which today is a $300 million business, despite the collapse of the telecom industry in 2001. The roster of A-list VCs who passed on the company is embarrassingly long.

3. VCs can't distinguish between smart and lucky.

The opposite of the "nerdy Ph.D." is the "serial entrepreneur"; VCs hate funding the former and love funding the latter. Serial entrepreneurs write good business plans and assemble complete business teams. There is a basic assumption that the serial entrepreneur is smart rather than lucky. So, having a track record of exactly one success, the same physicist who couldn't get funding for Ciena (David Huber) got lots of money for his next company—Corvis Corp. (now Broadwing Corp.), in Columbia, Md.

Being in the right place at the right time does wonders for your apparent intelligence, but the bankruptcy courts are filled with entrepreneurs who made millions the first time, then doubled down on Start-up B with loans secured by the assets of Success A. By the way, investors who bought Corvis, Huber's second company, at the IPO price of $11.8 billion have lost over 90 percent on their investment.

Even assuming that an entrepreneur is smart as well as lucky, serial entrepreneurs—almost by definition—do logical extensions of existing technologies. After all, it's smart to go with what you know. So while Start-up B may be successful, it's unlikely to be disruptive and therefore transformative. For example, we gave Huber's first company, Ciena, a rating on our innovation scale of 2, but Corvis, which develops long-haul optical networking equipment, where there are many alternatives already available, got a 3. The lightning rod of raw innovative brilliance rarely strikes the same technologist twice.

Compounding the focus on serial entrepreneurs is an overemphasis on parallel investing. VCs love to invest in deals that are fashionable. No one likes to invest in anything that seems daring. As a result, we see lots of indistinguishable deals for whatever is hot. For example, after the success of a few storage-networking companies, notably EMC Corp., in Hopkinton, Mass., and Veritas Software Corp., in Mountain View, Calif., more than 50 different investment opportunities in different niches of the same field were venture funded between 1998 and 2001.

The problem with this groupthink is that fashionable companies, again by definition, are going to be companies that are variations on the same technology themes; they are, at best, evolutionary. Arguably, if we want innovation, we need to replace serial entrepreneurship and parallel thinking with a willingness to judge a start-up on its merits, disregarding track records and the hot idea du jour.