Search Engines are not foolproof

They give you one box, you type in your question or query and press Enter. You might think that there is absolutely no way to use that incorrectly, or that something bad might happen as a result of that search. Yet, you would be wrong.

The thing about search engines is that they use algorithms that were designed to try and find results that are most relevant to your query. Algorithm has a number of factors it considers when deciding how relevant something is. That all sounds good, doesn’t it? Well, for starters, those factors are not whether something was proven true or is something based on evidence, but they are things that search engine manufacturers have decided that are likely to produce quality results. Of course, theoretically those things can lineup, so that search engine really presents you with credible information, but that is not guaranteed in any way.

To put it simply, google doesn’t care what’s true. It cannot possibly care about what’s true.

Yes, you might say – of course, google doesn’t have all the facts of the world programmed into it, so it cannot answer those types of questions by itself. Google doesn’t want to answer those questions themselves, they want to point you in a right direction. That is all fine – but what this post will try to argue is that search engines are not equipped to correct any initial bias in your query – if you start from a premise you want to prove correct – google will only basically support what you already think you know. Take a look at the following example, where the only difference is singular or plural.

Of course, using such terms which are more or less close to each other, you won't get a huge discrepancy. Still, using singular or plural search terms can cause google to not show you the answer snippet from Harvard, but from a fitness magazine. This is not to say fitness magazine doesn’t know what they are talking about, but to show how tiny, miniscule changes in your query can result in differences.

This is what can happen even if you start honestly.

Now, to show a more sinister example:

If you Google "what happened to the dinosaurs," this is what you'll see: http://t.co/FyikF8PUmL pic.twitter.com/61IxVUePgn

— Science Friday (@scifri) June 5, 2015

When I googled it today, the term was on second place:

Clearly, this is more or less a case of gaming the system. Two religious arguments on what is clearly a scientific query. So, clearly it is possible to game the system. Either that, or the system is not perfect in the first place. Take your pick. The most likely truth is that it’s actually both. Search engines have not yet attained even the ability to consistently show more or less same results if you change singular to plural. The point is, you simply cannot trust it to reach the conclusion for you. You can use google to find more information, but it is not a fact machine. So, even if you start with NO BIAS whatsoever and ask a honest question, you are still likely to be served with nonsense.

Just a quick point before we continue – nobody is talking against google specifically. Google’s chief competitor Bing is not doing too well either. They have religious answers as both first and second one.

For more examples, visit - When Google Gets It Wrong: Direct Answers With Debatable, Incorrect & Weird Content.

Also, a visit to the “Google’s featured snippets are worse than fake news” can be VERY enlightening. There you can clearly see how google messes up even the most basic searches – a search about a Law propositions has a conspiracy website as a first result.

Clearly, search engines alone are not the path to discovery what is really true. Not only that, but getting to the truth is not even the goal of search engines. The goal of search engines, is actually, believe it or not is to make money – show advertising. Not suggesting that there’s a conspiracy or anything, but the simple fact is that search engines are not designed to look for the truth. They look for pages that are relevant to what you typed in. And factors that increase perceived relevance are page load speed, number of links around the web to these pages, how people use programming language of the web (HTML), whether site uses encryption, whether page is optimized for mobile devices, there’s a suggestion that even each user’s search history is included into consideration, so someone is more likely to continue being presented with dubious sites … it’s clear that they do not care simply about what’s true.

Those algorithms can even be influenced by various techniques. Just google for technical term “Search Engine Optimization”. You will notice it also has a so-called black hat variety, which is clearly a technique to game the system.

How dinosaurs died or what happened to them is not VERY likely to incite anger or hatred towards anyone, but I am sure you can think of such examples. You might argue that it’s still people’s decision what they enter into search engine and Google and Bing and others should not care about that is true, but – as can be frequently pointed out, search engines also offer suggestions about what you should search for – sometimes things you would not even think about. Take for example this article from The Guardian -  Google, democracy and the truth about internet search by Carole Cadwalladr – where she argues perfectly how Google inspires people to look at results of searches they would never even consider. Read the article, it’s really awesome.

The thing is – search engines are not only making information available – they create it. Autosuggestion like The Guardian author had – “Are Jews Evil?” leads to a search about it – and there, thanks to the imperfect results or gaming the system or whatever – person is presented with supposed evidence that they are indeed evil. In this manner someone who is not even thinking about those things can start to believe that crap – and even believe it to the point where he will start to create such content himself.

I could see a situation where this could literally happen – a person who possibly does not even have any earlier bias is interested in a topic. They search for it and find a huge number of conspiracy sites which gamed the system successfully or the system didn’t do good enough job in the first place. They start reading it and decide that these sites do have a point. Of course, google cannot be blamed for this directly, as a person has still decided to trust what he read online, but if a search engine had a better algorithm which puts more credible sources closer to the top, it is possible such thing would never happen.

People do want quick answers and search engines do want money, but there should be a sense of corporate responsibility, or even some kind of legislation which would force search engine manufacturers to have this responsibility. Clearly, it is not the same thing as forcing companies not to dump waste, as search engines do not produce waste themselves, but maybe there should be some kind of inquiry by independent experts to ensure that search engines are doing something to combat fake news and fake facts. Obviously, this could arguably impact the business and it is even a question of slippery slope to censorship, but it is not something to completely discard. I believe that humans have the capability to fine-tune such measures in order to ensure a bit more responsibility without having an impact on freedom of speech or the like.

I am just going to say this – I believe that a company as big and as important as Google or Microsoft or Facebook – mentioning Facebook because they have a HUGE problem with so-called fake news – can and MUST do better. This is simply not good enough when you consider the amount of resources these companies have. And to drive the point even deeper – some of these companies are making Artificial Intelligence, self driving cars, etc. Will those technologies also be so sensitive to an omission of “THE” from your question? In these areas, there can be no mistake whatsoever. And today it’s clearer than ever – the goal of getting to the AI is not to get it RIGHT, but to be there FIRST. Do better. You have huge amount of resources.

So, what should we do? What are the Conclusions?

The conclusions are – take everything search engines give you with a grain of salt. The results vary greatly even on how you arrange your sentence, not to mention which angle you come from. The conclusion is most certainly not that search engines are evil, or are a conspiracy, or even that they are not doing a good job. The conclusion might be that search engines have their flaws, like everything else – and even if you use them with good intentions and completely honestly, you can still end up with a bunch of nonsense. So, the first step in looking for answers using a search engine would be that if you are unsure of something, try to correct for bias in your query.

Take food for example. You are much, much more likely to find something useful if you search for “what should we eat”, than if you have an initial bias which you might want to prove – you may believe that milk is bad, so you are likely to search for “is milk bad” or “is meat bad”. When you do that a search engine takes the words "BAD” and “MILK” and gives you the results it considers most relevant for these two terms. So, if you search for “is milk healthy”, you are likely to get different results, yet not completely without bias. This is mostly because legitimate studies and sites possibly do not even bother with such questions, as they do not seem relevant. Or even if they do, conspiracy sites are bound to rank high, because of other conspiracy sites linking to them. Instead, the legitimate and science-based sites are worried more on answering the general questions, such as what to eat. The correct approach seems to be to ask what should we eat, or what should a pregnant woman eat, etc. The search engines themselves are not yet smart enough – and it’s questionable if they ever be that smart – to understand that you really wanted to know should humans drink milk.

The time-proven advice is to always:

• Proportion your beliefs to evidence
• Think critically
• Remain skeptical
• Look for legitimate sources
• Look for possible hidden motives – remember, large corporations might have hidden motives, but so do individuals who are trying to sell you healing crystals or chakra therapy, or organically grown potatoes. Just because a corporation might have a motive, that does not mean that the other side does not.
• Combine multiple of the above
• Rinse and repeat