It is generally agreed that when we potter over to Google, hunting for enormously exciting topics to read about, we want it to return the best results possible for a search. But who or what should decide what those “best” results are? Option one – is a machine. Since its conception in 1998, Google has based results on its Page Rank™ algorithm. A method of measuring a page’s importance based on its incoming links, which – it is generally accepted – was Google’s major unique selling point. One of the keys to Page Rank’s success was its pure objectivity. All pages were born equal in Google’s eyes, and had to “earn” their rank by other pages “voting” for them via an incoming link. Google’s – or anyone else’s – personal opinion of a given page was effectively irrelevant.

However there are problems with this which first started to surface a few years ago. Google appeared to be struggling. With the Internet growing at a rate of 10 million pages per day, they appeared to be struggling to find a relevant page for all those obscure topics out there. The tiny amount of relevant content was slowly getting lost in all of the rubbish (and porn).

Option two – is a human. The best current example of a large human created information resource is Wikipedia. Written and reviewed by humans, Wikipedia is a collection of over 9.1 million pages covering almost all topics of interest. Because Wikipedia is peer reviewed so closely, you can almost guarantee a relevant page is returned, whether you’re searching for “Aabenraa”, “Zaafaraniyeh” or “Basil Brush” – the lovable anthropomorphic fox. But there are obvious problems with this. A human created resource is subject to bias. Even when reviewed by large numbers of people, cultural bias still exists.

So why is this relevant? Well – in my opinion – Google has hit a wall (again) when it comes to returning relevant content using purely machine learning. With the Internet now containing more than 1 trillion (that’s 1,000,000,000,000) pages, Google’s job in finding those top 10 relevant pages for a given term is now harder than ever. At the end of the day – for now at least – a computer cannot fully comprehend all of the possible variations a given search term could mean, and the relevance (if any) that those variations can have with each other.

One option Google has is to start applying human editorial judgement to their search results. This however would surely go against Google’s founding principle. Then again, there are signs that Google are starting to consider just that. Their Terms and Conditions used to state that “Google’s indices are indexed by Google’s automated machinery and computers”, but this disclaimer has now been removed. Perhaps the most interesting development to this effect is Google’s recent interface change. When you now perform a search on Google whilst signed in, you are graciously supplied with additional controls relating to their “SearchWiki”. As you can see from the image below, we are now given the options to “Promote” or “Remove” a URL (using the icons to the right of the page’s title).

Now initially, using these options only affects what you (you being the person currently signed into Google) see. However Marissa Meyer (Vice President of Search Product and User Experience at Google) has stated that “in the future it’s likely Google will use the data to at least make obvious changes”. Now this comment, made at Le Web conference in Paris last December, is enormously significant. This is – for the first time – Google admitting that they are going to use human generated information to affect their search results.

So what are the ramifications of this? Well I could go on for quite a while about these, but let’s just stick to the key factors. First, this is going to affect the Search Engine Optimisation (SEO) industry. At present SEO “experts” can go about their task with the sure knowledge that Google will treat the site they’re working on with the pure, unemotional objectivity as every other site online. However once human subjectivity is added into the mixing bowl, all this changes. At this point it doesn’t matter if you have the perfect keywords in just the right positions around your site, if enough users don’t agree that your site matches the search term, then that alone could send it tumbling down the search results.

The other key factor is trust from the perspective of objectivity. One of the reasons for Google’s mass popularity is that when you enter a term to search for, you can be sure that the results returned are not there due to any bias, be it political, religious, cultural etc. However if Google do decide to go down this path, then it’s quite likely that we will start to see a bias on the side of those who choose to use tools such as the SearchWiki. This might not only affect our faith in Google, but more importantly – from Google’s perspective at least – advertisers’ trust in Google. After all, nobody is going to want to advertise with a company who appears to be favouring a competitor.

So what would you rather? An objective search engine run by a machine that returns ok results, or a subjective search engine run by humans that gives “better” results (assuming you side with the bias)?