Michael Braly

We have reached a point in time where it is possible to store and search all of the world’s digital media at a reasonable cost, but we haven’t reached the point of being able to find relevant information. Search is not a solved problem.

Ease of use is a huge barrier to advanced search techniques. Even “advanced search” isn’t really advanced enough. Relevancy is hard to measure, so you can never prove that it is in fact relevant. Also, relevancy continually changes as the data searched changes. Never use anecdotes to show how good or how bad a search engine is. You can always come up with good examples and bad examples for search engines.

The new Amazon search in a book was done by actually scanning books, OCR’ing them and then indexing them. So far in six months, they have scanned 120,000 books, 35,000,000 pages and are actively working with 190 publishers. To OCR the books, they used idle disaster recovery machines. At first they thought of creating a screen saver application to run on employees machines, but were concerned about the effect on the network.

They didn’t use the electronic versions of the books because they want a user to see what the book looks like and get a feel for it. They made the search inside feature the default because they want everyone to use it. The books that have been indexed are selling 9% more than non-indexed books.

In the future, they are envisioning you being able to take a digital picture of your bookshelf and have the search engine present give you the ability to search the text of your books.

They have developed a security algorithm that allows a user to browse a few pages, but not read the whole book or copy the whole book. The question was posed, “If I figure out the algorithm, couldn’t I just read the whole book?” Mr. Manber replied, “YOU could not figure out the algorithm.” Geeky laughter ensued.

Amazon is working on Personalized Searches that remembers your past queries and stores them on their servers.

Jeff Bazos’ favorite saying is, “This is still Day 1.”