He also drove the development of Java, co-designed the picoJava and ultraJava processor architectures, and worked on the Jini technology that uses Java to connect computer devices. BusinessWeek's Heather Green recently discussed the state of innovation with Joy. Following are edited excerpts of their conversation:
Q: Do you think -- though it's difficult for startups to get funding these days, and good companies along with not-so-interesting ones aren't getting money -- that innovation overall will be fine?
A: I don't quite agree with that. Assuming that you want maximum innovation, then you need a rational investment climate, where the good ideas can get capital at a reasonable rate. A couple years ago, even the bad ideas were getting capital. This wasn't good, as the good ideas not only had trouble getting the people they deserved, but too many companies doing the same thing were funded, so they all had trouble.
Now, we've gone too far in the other direction, shutting down investment even in some good ideas because the public market and other endgames (e.g. get acquired by Cisco) have largely gone away. Venture capitalists seem to have responded by doing fewer, bigger, deals, but the most interesting and innovative things are often not big-dollar-amount deals.
Q: Is slower innovation fine? Do important things get lost? Is it O.K. for companies to simply say: Now, we're going to absorb the technologies we acquired during the past couple of years?
A: Clearly, society is having trouble adopting all the technology that [the] Moore's-law-type curve is producing. [Moore's law posits that the number of transistors on a chip will double every 18 to 24 months, driving computing power to increase at an exponential rate.] Witness the huge capital and licensing expenditures for things like G3 [third-generation wireless technology] spectrum.
Will people's behavior change quickly enough to absorb this innovation, or will the companies run out of money first -- like, say, Globalstar and, perhaps soon, Ricochet and other innovative networks? If what happens is you "skip steps," then innovation can benefit.
For example, in the personal-computer space, the transition from x86 to 286 to 386 wasted an enormous amount of time and resources. The 286 was not (in my opinion) a very good chip and still didn't let the kind of software being run escape the pitiful address-space architectural restrictions in that time frame. As I remember, Microsoft clearly wanted to go more quickly to 386, but IBM slowed it down by wanting to do the 286 first. This could have been viewed as more conservative but clearly just slowed things down.
Right now, with Itanium implementation of their new IA-64 architecture, Intel has, in my opinion, a similarly "doggy" chip on their hands that's late, slow, and just generally not worth it. I think you'll see it fail in the marketplace in a similar way to [how]...the 286 failed to have much impact. The difficulty in deciding to skip the 286, is that the 386 might not have worked, either, and then you could just defer essentially indefinitely. Intel has this same problem now with the Itanium.
Q: The low cost of capital during the past couple of years led to the creation of companies that didn't have defensible models or technological leads. With VC investing running at a 50% lower rate than last year, are some good ideas getting lost?
A: I think there are good ideas not getting funded because the groups are too small to be of interest to $1 billion funds. I also think that anything that requires infrastructure investment on a large scale is really difficult. But realize that there are often nonrenewable resources involved: If this is spectrum, then it needs to be allocated wisely. If it's people, then the markets are supposed to allocate the bright people. But if the markets are distorted, then the limited number of bright people who can do great things are badly allocated.