In contrast, he said, it was much easier to sell Linux in the context of a server: "There's just a few loads, they're fairly simple, they're fairly well-understood, people have much less inertia in upgrading a server than they have in upgrading their desktop."

At one point, the discussion turned to Sun's work to create an open source community around OpenSolaris. Torvalds expressed deep skepticism toward the effort.

"It's generally hard to build a community around a commercial entity that also wants to be in control because everybody else around that commercial entity will always feel like they're at the mercy of Sun," he said.

This dynamic is reflected in OpenOffice, "where the fact that Sun wants to have copyright assignments and exclusive control over the license ends up being something that actually drives away some developers," he argued.

Overall, Torvalds expressed ambivalence toward Sun, which recently has developed an image as a particularly open-source-friendly company. "In many ways, Sun has done a lot of things right. At the same time, they seem to often have trouble going the full last step," he said.

Ian Murdoch, vice president of Sun's Connected Developer group, defended Sun's commitment to the open source model. "An open source project has to have coordination, it's not anarchy," said Murdoch, who earlier in his career founded the Debian version of Linux.

"If you look at the Linux kernel itself, Linus is firmly in control of that, and the only difference is he's an individual and Sun is a company. ... Some of the most successful open source projects are being driven by companies like MySQL, for example," he said.

Torvalds also pulled out his crystal ball, offering predictions on how technology will change in the next five years.

While hardware will be "hugely better," software's power and complexity will grow alongside it, he said. "I suspect things will be about the same speed because the software will have grown and you'll have more 'bling' to just slow the hardware down."

Virtualization is "not that big of a deal," according to Torvalds. "It's been all around for decades, and it's very interesting in niche markets -- I think the people who expected to change things radically are just fooling themselves."

Real change will come from entirely new uses of computers, he predicted.

(James Niccolai in San Francisco contributed to this report.)