In a follow up to an earlier post, I wanted to comment more on the talk given by Nelson Minar of Google. First, the audience at Gnomedex includes a broad range of the technology users spectrum, from newbies to hardcore developers and implementers. The talk was probably as technical as it could be for the audience; however I wish it had gone in depth a little more. Maybe I'll get that chance at some other event in the future.
Nelson talked a lot about relevance in search results, and provided a "top level" description of how Google performs text analysis. There are over 100 factors that Google uses when analyzing text to assess a document’s relevance. A few of those factors include link text, font sizes, and proximity of keywords in the text. Nelson referred to some papers written on this subject. I’m going to look for those on the web and if I find links, I’ll post the updates here.
Another interesting subject was Google’s use of commodity hardware in its datacenters (supporting around 10,000 servers) and Google’s development of custom web server software to enhance the performance of their search process. These points made it clear that Google sees its value in the application of new ideas in technology, including the creation of the best hardware and software solutions to support the goals for the site. Many people would think "Why build a web server, there are so many options already?" or "Why build your own servers when you can buy low-end servers from (insert vendor here) cheap?". I am more impressed with Google as a business now than I was before I came to the conference- I think they’re doing the right things to focus on meeting the needs of customers and understanding the value proposition that they can provide to web users through technology innovation.