Last week I ran into Troy Lane Williams, founder of PeoplePad, a stealthy Austin-based startup that’s creating some kind of front-end portal for the semantic web. I have no idea what the finished product will look like, but Williams’ previous startup experience has colored PeoplePad’s product and its formation.

Williams may be familiar to readers who recall his involvement in Questia, the pre-Google Books, subscription-based online library that launched in 2001 with $150 million in backing. Questia is still in business, but Williams left in May 2007.

Wisdom from Williams includes:

• Shut up. I had to push to get Williams to talk to me. After a big launch for Questia and watching other complicated technology companies receive media hype they couldn’t live up to, Williams says he’s not letting PeoplePad into the public eye until March 2009. He’ll do a limited beta this fall, though.
• When it comes to the web, free is where it’s at. The last decade has taught Williams that the mass market isn’t going to pay for online informational content, which means advertising is a must, as is getting cheap — but also high-quality — content. He cites Wikipedia as an example.
• The Internet isn’t for passive reading. Williams dealt with large blocks of text at Questia; he even criticizes Wikipedia for being too text-heavy. “We need to structure data around lists and data boxes instead of around snippets, so people can get to the key points,” Williams says. Data doesn’t have to be text and a site shouldn’t restrict itself to “spiderering around the web” for all of its information.
• What’s good for people is better for machines. If it’s easier for people to enter short amounts of data, it’s also easier for machines to read that data using properties of the semantic web.
• Use existing technology if you can. Williams says he’s not out to build new semantic databases or semantic programming languages; instead PeoplePad will use services from companies such as Metaweb or Radar Networks.
• Usability is key to the success of a consumer-oriented site. The semantic web is less powerful if grandmothers, dentists and other not necessarily tech-savvy can’t use it.

After hearing all this, I suggested to Williams that he was building a sort of Mahalo powered by the semantic web, but he says he’s not. Perhaps after Williams wanders down Sand Hill Road this summer looking for his first round of capital, more information will leak out. Readers, any guesses?