Google is widely considered to be a very smart machine. Whenever I’m speaking to other SEOs and inbound marketers, they will often give a lot of credit to the search giant.
I was at a Distilled presentation and networking event in London a few months back; one of the hot topics (and still is) was Google’s authorship. After the speaker finished, a bunch of us stood by the bar and had a chat about where we thought Google authorship was going. I brought up the possibility that it might become totally hijacked by tons of fake profiles – profiles that were there purely for click through rates and ranking speculation. One of the responses I got was that Google would be able to see if a user was logging in and out of multiple profiles; as a result Google would simply discount their profiles. As a human, it sounds totally simple right? If I was watching someone logging in and out of fake profiles I’d be all like “NO! Stop that!”… but if I was a machine, how the hell am I supposed to judge the this person’s intent to be manipulative or not?
It’s hard to debate as no one knows for sure, but the common response from those who rate Google’s intelligence is that Google has a shit load of data. This is very true, Google is indeed frolicking around in fields of data, but data on its own is not worth a penny if there’s nothing there to understand and process it. We don’t just throw numbers at a calculator and expect it to know what to do with them.
So now I’ve resorted to this; a crudely titled blog post about some of Google’s biggest screw-ups.
Google+ Authorship and Francois Hollande
So when it comes to fake authorship there’s one person that takes le petit gâteau. It’s a guy in the binary options market, but not just any guy – it’s a familiar face I’m sure you’ll agree.
But despite what you may have first thought, the French president, Francois Hollande does not own a binary options affiliate website. This isn’t Mr Hollande at all; it’s a guy that just looks a lot like him. In fact, he looks so much like him that Google will tell you just who it is…
… and this is where I struggle. Google is able to feedback, correctly, in a few milliseconds, the person that is shown in the image. What we’re seeing here is not smart or semantic, it’s this idea of throwing numbers at a calculator; Google is able to tell me what those numbers are, but it’s certainly not able to say “hold on a minute, this author is clearly fake – we should remove their profile from our search results”.
Google Tip! If you weren’t aware of how to search for similar images, go to Google image search and simply drag your file into the search box. Easy!
“LAC is the secretion of a number of species of insects…” Oh really?
In February I will be attending the London Affiliate Conference, or as it’s widely known – the LAC. I’ve been on the LAC website a whole bunch of times in the last few months, in fact my Google history is probably full of affiliate related stuff. But nope! Google decides that my search for LAC warrants a knowledge graph result about insect secretion…
This isn’t exactly a poor user experience and it’s not going to cause me any trouble… But this is a simple example of something that many consider commonplace in search – Google’s ability to personalise search results based on browser history and social.
Blue is the new Yellow
And finally, when a search query couldn’t be an easier to interpret (surely!) Google messes up enormously. Credit to Google, this result stopped after a few weeks, but it’s incredible that this ever happened:
I just want to finish by saying that I’m by no means anti-Google. But in my opinion, SEOs shouldn’t be so quick to assume Google knows what it’s doing. Google makes mistakes constantly – these are just a few that I’ve seen. Don’t use this sort of information to try and game the search engine, but I recommend looking at big G in a different light and not overthinking your strategy into a strategy that doesn’t work.