The Gillmor Gang

with guest Adam Bosworth

January 14, 2005
71 minutes, 32.6mb, recorded 2005-01-14
Adam BosworthAdam Bosworth, now Google's VP of Engineering, joins The Gang this week to discuss his vision for the future of search architecture. "How do you handle data that's much less known up front and where the query is by relevance?" Adam asks. Most of today's databases are built on the relational model, but most of today's queries are not. Instead they're looking for keyword precision, location and semantic context -- not a textual or numeric match. The relational model is designed for use when both the data and the queries can be anticipated, but in today's world, neither are typically known in advance.

Adam suggests that the same divide-and-conquer architectures used to make web servers more scalable could be used in search. He envisions data routers that will know which back-end servers have which knowledge and will query servers asynchronously according to the liklihood of getting the best results.

The discussion then turns to the topic of attention and the technology and politics of knowing who's reading what on the Internet. XML-based RSS and Atom have created both the challenge and the opportunity.

Today's Gillmor Gang:

Steve Gillmor, contributing editor, ZDNet
Doc Searls, senior editor, Linux Journal
Michael Vizard, editor-in-chief, CRN
Dana Gardner, senior analyst, Yankee Group
Jon Udell, lead analyst, InfoWorld Test Center

And our special guest:

Adam Bosworth, VP of engineering, Google

[You can also hear other editions of The Gillmor Gang.]


This free podcast is from our The Gillmor Gang series.