Don’t be slaves to the algorithm. Understand them.

Matt Rhodes, Head of Digital Strategy at WCRS on our over reliance on algorithms and why we need to understand more about them

“We are slaves to the algorithm.” So said Tanarra Schneider of Fjord Trends on day 3 of this year’s SXSW festival in Austin, Texas. Neatly summarising a topic that surfaced in sessions across the whole week on topics from Fake News to bots, marketing and autonomous cars.

Algorithms increasingly inform and are part of many tasks we do online and many parts of our lives. We are willing to put our trust in them to find information on Google, to find news via trending topics on Twitter or to recommend products for us to buy. They help to drive automatic cars or to target us with advertising that is thought to be right for us.

“But we don’t really know how they work,” stated Mark Hansen, Professor of Journalism at Colombia University, during a session about detecting Fake News. There can be a temptation to think that things that are programmed, that are powered by the machine, lack the bias that human analysis might bring. And whilst, when we think about it, it might be patently clear that this is not true, we continue to put our faith in algorithms without interrogating how they work.

How do we tell if a news story we read online is true or not, for example? When people a common technique is to go to Twitter to see if that topic is trending. If it’s in the trending topics list, if that many people are discussing it, then it must be true we reason. But who knows how the trending topic list is generated on Twitter, Hansen asked the room. The answer is that nobody there does, and indeed none of us really do as Twitter don’t tell us.

Or what do we expect to get when we search for things on Google — as Schneider described, the search term “Where can I get an abortion” will take you to pro-life abortion counselling clinics in certain states in the US, for example. Not necessarily the information the user was looking for or expecting the Google search algorithm to deliver.

So as we increasingly rely on algorithms to help us in our lives, or to sort through the wealth of information that is available to us online, we need to more actively interrogate what these algorithms are and what biases they might have. Indeed in many cases we need to actively understand that these decisions are being made by algorithms at all.

We need to understand the data we are looking at, the algorithms we are using and how these algorithms are designed and programmed. And this is becoming increasingly important as trust in technology and digital tools is being questioned more and more.

We are now more aware and conscious of the technology we are using from ever before. In the early days of Facebook we may not have considered that things we shared, seemingly with our friends, was also available for others to see or for advertisers (and others) to access and use. As technology has become a more prevalent part of our lives, and as the role of technology is increasingly discussed in the media, we are now more aware of this. We have become less trusting of technology, more skeptical of what firms are doing with our data and how they are acting.

And as we become more conscious of technology so we become more skeptical about it, more questioning of what it does and how it works and more demanding of the platforms and services that provide it. We are more aware that we are being targeted with advertising, for example, and we want to know how that works. We are more aware of clickbait headlines and know that clicks don’t mean quality.

So we should be questioning the algorithm more. We should expect people to question the algorithm more. And we should expect that the full benefits of the many new technological advances on display at SXSW this year will only be realised at large once we understand not just what they do but how they do it.

And, as an industry where algorithms play a role in much of what we do, we need to prepare ourselves and our clients for this more technologically savvy, and technologically curious, audience. We need to take the lead on understanding the what as well as the how. And there are two clear things we can do:

  1. By educating ourselves. Across disciplines in advertising we can create better work if we better understand the algorithms that will be used both to help people discover the work that we create and also to help people discover content that might compete with what we are creating. To do this we need to be better at asking questions, being inquisitive and pushing to find out what the algorithms are and how they work.
  2. By ensuring that where we use algorithms to create or deliver advertising for brands, what we are doing is fair and responsible. That we are clear why we are using algorithms and what the benefit is. And that if the people we are reaching query this then our answer can be genuine and honest. This isn’t about compromising what we do, or explicitly saying how we are doing it. It is about being responsible about what we are doing and why.

For more from the IPA on SXSWi 2018, read Susan Poole’s blog Let’s create a future that is less techy and more magical.