There is a large complex of rectangular, somewhat boring looking buildings in the San Francisco Bay area of California that hums constantly. It isn’t an industrial plant, it isn’t part of the public utility infrastructure and it isn’t a government installation. The hum is caused by over 45 thousand computer servers and the system that powers them and keeps them cool. It is the physical manifestation of information and its manipulation by millions of web users as they try to answer questions, plan holidays, find products and acquire new knowledge using the world’s largest search engine. This is the hum of Google.
The company itself keeps quiet about exactly how many of these data centres it has, protecting its commercial information behind anonymous limited liability companies and secretive business entities. The general consensus however is that there are at least a dozen in the United States and at least a dozen more dotted around the rest of the world, each of which uses between 50 and 103 megawatts of electricity.
While the prospect of visualising all this data is terrifying enough at present, the scale of this operation is only set to grow as Google continues to strive towards its version of the most ambitious, revolutionary (and some say inevitable) development in search engine technology: semantic search.
What is Semantic Search?
Up until now when a user inputs a search term into Google (or most other search engines) the engine has scoured its databases for all the documents it considers relevant to the particular words in the query and displayed the results in order of their quality and relevancy.
The problems with this method are many and obvious to any user. When you use a term with an ambiguous meaning, the search engine doesn’t know which meaning you intend. For example, the word ‘Queens’ could refer to female monarchs, a borough of New York, the reproductive females in bee colonies or drag acts. The engine has no choice but to display all the results it thinks might be relevant, leading to a collection of links of which many may be useless to the user.
Similarly, even when a word has no ambiguous meaning, the intention of the user, and the information they need, can vary dramatically. A user searching for ‘golf’ could be looking for an overview of the game, current scores in a major tournament, a course in their area, a shop where they can buy clubs etc.
Semantic search attempts to overcome these problems by removing the ambiguity from search. Instead of just ranking each page for relevancy to a particular ‘keyword’, a semantic search engine will attempt to store information on the various meanings of each word and try to work out the users intended meaning based on a number of factors: how they use the word in combination with other words, how they’ve used the word before, how others commonly use the word, their location, current events and many more.
Of course gathering, storing and sorting all this information – all the many connections between word and their various meanings – requires huge amounts of computing power, so one way in which we can expect semantic search to change the web is in the amount of energy required to run it. But there are many more ramifications of semantic search for the web and how we use it. What are they? I can, at this stage only speculate – which is exactly what I intend to do…
1. The web will get more efficient and we’ll spend more time on it
If Google spits out exactly the information you need every time then there’ll be no need to trawl through loads of web pages looking for it. As a result we’ll spend a lot less time on irrelevant web pages. Will this time we’ve saved be spent reading books under trees and gambolling in dewy meadows? Probably not. We’ll end up spending more time doing what we actually want online – which is exactly what Google wants. In order for their semantic model to work, and outstrip any potential competitors, they need to gather as much information about your internet behaviour as possible. This information strengthens their semantic search database (currently called the Knowledge Graph) and makes it more likely that you’ll stick with Google products.
2. Ads will ‘know you’
Google isn’t going to keep all that juicy information about user behaviour and intention to itself of course – the plan is to use it to deliver semantically targeted ads for large monetary gains. Anyone who’s been followed from site to site by ‘retargeted’ ads which remember which products you’ve been looking at recently will know that some of this information is being utilised by advertisers already. Keeping this kind of ad delivery on the right side of creepy will be tough – nobody likes to feel that advertisers are getting inside their heads.
3. SEO will die (again)
Search engine optimisation – the act of altering the structure and content of websites to make them appear higher in search results – has up until now been entirely keyword focused. That has to change when the large search engines move to semantic search – the task will then be make your web pages appear more relevant to the user’s intention than the words themselves. SEO will therefore have to reincarnate itself, once again, to adapt to the new search environment,
4. Google will face serious competition
Google is in pole position on the semantic search starting grid: their huge resources, massive amounts of data already collected and their tremendous experience of both building and marketing the world’s preeminent search engine testify to this. But semantic search is a whole new race, and while Google have to be applauded for leading the way, what got them to where they are today was superior text-based search. This is a great opportunity for competitors to find a better way to do semantic search than Google and start to eat into their market share.
5. Google’s replacement won’t be a Western company
This prediction is based more on economics than the mechanics of search itself. In order to provide a rival to Google you need huge computing power which itself requires huge investment. Huge investment in start-up companies isn’t going to happen the West for some time – the global financial crisis has seen to that. Looking to China and India in particular we see economies which are still growing (albeit at a reduced rate since 2007), still attracting outside investment and still under the control of governments which see technological innovation as a hallmark of prosperity. If semantic search throws the game open to new competitors (and new languages – since semantic search no longer relies on words themselves but their underlying meanings) it’s tempting to predict that the next global player in search will come from the East.
Jamie Griffiths is a writer, editor and thinker for Approved Index – a leading UK B2B directory and marketplace (and a source for quotes on translation).