So “the algorithm” has become a topic lately. This is what Steam refers to when talking about trying to direct you to the best games for you, and to bury the garbage. This is what Google talks about when trying to optimize search results. Some people complain that more human oversight is required. Sometimes these failures can be rather striking, like with old “Google Bombs” that hijacked the algorithm to usually amusing effect.

Clearly from several UI overhauls ago. The “I’m feeling lucky” button isn’t around there anymore any, so this doesn’t exactly work anymore.

What is the Algorithm?

In case it’s not clear, there isn’t just one algorithm. The use of the term “The Algorithm” is just a more comfortable way to refer to trying to automate management of certain dynamic elements algorithmically. This has also been referred to as “intelligent agents.” For the most part, this only affects people’s lives when it comes to online activities. There are some exceptions, but they aren’t very significant for this discussion.

This approach of finding an effective algorithm for catering to users has morphed a bit over the years. Things like having locally stored browser cookies that remember your actions on a website are one example. However, as time went on, the methods became more and more sophisticated, as well as what could be done with them. This is how Google or Yahoo news can work as a customized news aggregation site. They see what you read, and offer more similar stories based on those interests.

Why no human oversight?

The reason that everyone is on a search for the best algorithm possible is because it is automatic and it scales up easily. Online you’ll have millions of users each making dozens or hundreds of requests every day. Each one of those users will have different interests and preferences. Trying to curate for them manually will require either an astronomical amount of effort, or only very generic curated and shallow pools.

Manually, we get a national sports page. Algorithmically, I get news on Sumo, while you get news on the Seattle Mariners, even though you live in Michigan.

This is also why Steam wants to solve their game discovery issues with an algorithm. They’ve been open that they want to be surprised. With a manual process, they never would have had “Hatoful Boyfriend”. With good algorithms, more players will be able to find games they like more, making everything more profitable. Less would have to be spent on advertisements, or other forms of user acquisition. All kinds of good stuff would follow.

This all doesn’t even get to the legal implications. Don’t underestimate the significance of that aspect. If an algorithm processes everything then there can’t be any accusation of bias on the part of the platform. The function is inherently neutral. That’ll reduce the exposure to legal risk significantly… if done well.

What’s going wrong?

The best laid plans of mice and men…

People are unpredictable, clever, and sometimes malicious, even if unintentionally. Recently it was discovered that Facebook sold advertisements that were targeted to Anti-Semitic users. This option was only even available because Facebook had an algorithm to create advertising categories. No human was involved from Facebook’s side in the entire process of getting hateful adds to hateful people to stoke their hate. This is, in turn, how gaming the algorithm can in turn serve hate speech.

Another problem that arises is a sort of a self-selection bias. Going back to news aggregation, if I show interest in one side of the news, it will serve me more news along those lines, or from that source. Eventually this selection will drown out opposing views, which will then create an echo chamber effect. Whatever your beliefs going in, you will be given all of the support and affirmation you could want. You may not even notice the confirmation bias.

This extends to communications as well. How does Twitter prioritize what messages you see first? If there is an algorithm to calculate the terms you use, the people you follow, the messages you like, and so forth, then the same echo chamber will creep in around you. This has two big differences, though.

One – This can manipulated. Bots can put stuff out there to “twitter bomb” via affirmation of biases.

Two – this gives the impression of poplar support, rather than a preponderance of news coverage. This, in turn causes “The Mold” to come into play. You feel that your view is correct, and anything that does not fit is “the other”.

So what do we do now?

With the social elements? Always challenge your point of view. Resist your own confirmation bias. Try to understand the other perspective rather than just write it off as stupid, or someone being lazy.

With things like the Steam store? Take the time to understand what they are trying to do, and why. If you’ve read all of this, you’re well on the way. That doesn’t mean don’t hold them accountable. If Steam is allowing people to take advantage of their platform with shovel ware and card trading schemes, they do need to step in and take a more active role.

Until then, the Steam Cleaner still has work to do.

With so many working on them, some day algorithmic solutions will be robust to the point where exploiting them is more difficult than it’s worth. That day is not today. In the mean time, it would likely serve everyone best to allow for a manual release valve of some sort. News sites blacklist bad sources. Twitter suspends bots. Steam removes games with stolen assets. These would all be difficult things to do, would increase legal exposure, and would have to be handled on a case-by-case basis. However, it would all be easier if there is consideration given to this kind of manual override scenario early in the design process.

Kynetyk is a veteran of the games industry.  Behind the Line is to help improve understanding of what goes on in the game development process and the business behind it.  From “What’s taking this game so long to release”, to “why are there bugs”, to “Why is this free to play” or anything else, if there is a topic that you would like to see covered, please write in to kynetyk@enthusiacs.com or follow on twitter @kynetykknows