Ed Felten on the Search Space
ø
Here are some keywords of Ed Felten’s presentation here at Yale’s search conference:
- Talks about what search is
- Search is broader than we think that it is
- Three steps, processes or elements: (1) observe, (2) analyze/learn/, (3) serve users
- Observation of information, either crawled in the web or the university library, real world
- Put information in a DB, so that it’s available in electronic/digital form. Analyze, index, learn, model that information; put some sort of value on top of it
- Index, model, … built from it allows you to serve users, answering queries, answer questions
- Broad definition, it’s not only search engines such as Yahoo, it also includes Google print, fixture sites (e.g. baseball statistics), attributes of P2P file sharing systems; also applies to consumer reporting organizations, e.g. choicepoint
- Interesting issue among others: Question whether search is internal or external to the world it is studying. Eg. Ebay has searchable search engine inside for objects/auctions. Component that is part of the service they provide. Bidder’s Edge tried to build external search; eBay wasn’t happy. Other interesting case: Grokster, e.g., has/had internal search engine. BitTorrent didn’t provide a search engine, provided only transfer and got significant legal advantage.
- Other interesting aspect: analyzing and learning brings the value, e.g. Google’s PageRank is the value added. Analysis step is where the heavy thinking happens and value is created. Interesting, b/c legal challenges have not challenged analysis element, but are challenge to crawling = observation stuff (e.g. eBay v. Bidder’s Edge).
- Decentralization and P2P design. Complex issue. If analyzing and learning is key, but “observation” element of the search process is the target of law, it’s likely that we try to decentralize the observation part.
- In sum, search is broad, we’re very early in development of this technology.
Update: More on the other panels here.