Tuesday, March 11, 2008

SearchMe & Search Innovation

Search innovation is all about the indexing and the ranking and the massively scalable infrastructure, not natural language queries, not categorized search results or suggested keywords.

SearchMe raised $31M today from Sequoia. Their slick Flash UI to display previews of search results is getting all the attention, but I want to know what the smart money at Sequoia got behind from an indexing and ranking and infrastructure perspective - I am curious about that.

SearchMe CEO Randy Adams, on Kara Swisher’s blog, says:

“We are no Google, of course, but we are trying something different to provide a new experience for search users,” said Adams. “Most of all, we are trying to innovate in search, which is still largely a text and list experience.”

‘Text & list experience’ is not at all the most glaring hole in today’s search experience. Here is a better acid test: try finding authoritative, non-commercial information about ringtones - http://www.google.com/search?q=ringtones - turns out, as great as Google is at searching - not only is there is still a ton of improvement that can be made, but as a Google employee like Matt Cutts might tell you, improved search experience is a moving target that requires constant maintenance and innovation, particularly with regard to parallel developments in SEO techniques and wide dissemination thereof.

Powerset is doing some really useful work in search, mainly by contributing to the open source distributed computing projects Hadoop and hBase. Powerset also touts some natural language processing work, which is probably, relatively less impactful (‘impactful’ is not a real word). Hakia is doing some useful work putting natural language processing to work on the indexing (not search query construction) side of things. And more approachably, Like.com is forging ahead in image search through their initial, shopping vertical.

Like SearchMe, at blist we know that UI innovation is an important piece of the puzzle, but at the same time, it's only the tip of the iceberg in terms of creating a ton of value for users. I want to know about the 3/4 of SearchMe that is below the surface.

1 comment:

Mark Johnson said...

Though Powerset certainly is excited and using open source resources like HBase and Hadoop, natural language processing is our bread-and-butter. Through technology we've licensed from PARC and the work done by our internal linguists, we're able to do a deep, semantic parse of every sentence we read.

You should come to Powerlabs to try it out and drop me a note if you have any q's.

-Mark Johnson, Product Manager