Wall Street Journal reported today on some changes coming to Google search, but the article seems a bit confused about what they are. The lead item is that "[o]ver the next few months," Google "will begin spitting out more than a list of blue Web links," providing direct answers to questions instead. That's not new at all.The
What is new is that Google's Amit Singhal and team are bringing semantic understanding to search queries. Instead of just parsing keywords in a query like a dictionary, Google will use machine intelligence to interpret the meaning of the query and use that to find the most pertinent results. There are no specific announcements from Google yet, but the implications of this PR messaging are profound.
Outsmarting The Humans
This is bound to shake up the way today's keyword-driven search engine optimization works. The essence of the SEO game is tailoring page titles, URLs, topic tags and body text to the words and phrases people use to search the Web. Google only has to match the keywords in the query to the keywords on the Web using a lexical database. That's relatively easy, and it allows humans to game the system.
Semantic search would add much more intelligence on Google's side of the transaction. The specific wording of the query won't matter so much, because Google will be able to determine the intent behind it. It will determine the probabilities of various meanings of your words and phrases and decide on the fly what results make sense. The process that gets the user to her results will be much less subject to manipulation.
Google's New Kind of Relevance
This doesn't just have implications for SEO, though. Much of Google's existing business - its search ads - are based on keywords. But Google is in the process of shifting to new kinds of relevance signals across the board.
The purpose of Google+ is to create a layer of people, places and things - and a network of their relationships - that is visible to Google search. In order to stay ahead of Facebook and Twitter's interest graphs, Google wants to move toward those kinds of signals for ad relevance. Instead of showing you ads that just match the terms in your search, Google will have to match ads to your search based on your meaning, as well as your context, location, +1s, relationships and so on.
The Google+ identity layer also gives semantic search a leg up. With a more explicit understanding of the thing, place or person to which your search query refers, Google knows better what you mean. It doesn't have to do as much statistical guesswork when you type in a name if you can explicitly identify the person through Google+.
But Is This Really New?
Google still has a semantic search experiment called Google Squared, and as Google Fellow Ben Gomes told us last month, the search team has been working for years to incorporate signals from semantic databases like Freebase. Still, when the results of these efforts eventually result in a change to the very way the Google.com search box understands queries, the Google experience will change dramatically.
With a more intelligent search engine, Google will be better for users. It's not just the interpretation of queries that will improve; the quality of results will be better since they can't be gamed with keywords. This will change Google's ad business profoundly, but that change is inevitable. If Google doesn't become the most relevant, intelligent search assistant, Apple's Siri will.
Again, Google hasn't announced any specifics about the semantic search changes. But recently, the company has been remarkably transparent about changes to search. This week, for the first time ever, Google posted unedited video of a search quality meeting. Let's hope that trend continues.