The secret truth about The Algorithm

There's an eye-catching flavor of news that keeps coming up from time to time. According to the stories, powerful Silicon Valley execs are fine-tuning the algorithms in their products and platforms to manipulate users.

It reared its head again last month for the election. Are the brass in Big Tech giving special treatment to certain kinds of content from users?

Now trust me, I really don't want to dive into the politics around that question here. And yes, tech companies do control what's shown on their platforms and when it's shown. But I do think it's a good opportunity to dive into how algorithms work, why there's a chance it might not be as nefarious as it seems in headlines, and what we can do to make algorithms better.

In this issue, we'll talk about how:

• Deciding how an algorithm works can be a make-or-break decision for a lot of tech companies.
• The human part of algorithms is really important (as is the case in so many parts of tech).
• It's really important to create algorithms people can trust.

Writing the recipe for an algorithm

I talk with a lot of tech entrepreneurs and people who work at tech companies, and the truth is that how an algorithm works is rarely at the top of their priorities. (If you need a refresher about what algorithms are, check out an earlier edition of Tech Translated called "A beginner's guide to algorithms".)

Let's imagine an innovator named Julie who wants to start a dating app as a side hustle and grow it into her full-time business. Julie's head is likely full of ideas, including:

• How should people interact with each other? Who should be able to message whom, and when?
• How should profiles look? What information should be on them?
• Should the platform focus on people looking for serious relationships, flings, or every kind of connection?

But there's a key issue that's just as important –?and often overlooked.

Imagine a new user signs up, and there are 100 potential matches on the platform. How do you decide which matches they should see first?

That question is what's at the core of almost every algorithm-based platform. For a search engine, it's which pages should show at the top of the list of results. For a social network, it's which posts should appear first when a person loads the app.

But here's where things get tricky: there are competing interests for what shows up in those top spots. If you're really focused on revenue, for example, maybe you'll choose to throw some ads in the list.

And for many platforms, two goals rise to the top pretty quickly: keep people on the platform and make people want to come back.

Algorithms meet humanity

Algorithms are really good at doing what you tell them to do –?but if left unchecked, they can turn into a bit of a literal genie when you factor in how humans tend to behave.

Let's think about a social network. These kinds of platforms (generally) run on advertising. The longer people stick around looking at content, the more ads they see, and the more money the platform makes. With that in mind, some enterprising director of revenue might dictate a simple, company-wide goal: keep people on the platform.

There's a slight problem there, though: people tend to engage with content that evokes emotion. That can be great if you're liking a photo of someone's wedding or newborn. But it can also lead to echo chambers or even vicious cycles of ragebait.

In other words, it's not always executives putting fingers on the scale that leads to issues with algorithms. Sometimes it's just a matter of letting them run wild on their own without guardrails.

Forging the future in an algorithm-heavy world

As more and more of what we do and see is guided or controlled by algorithms, we need to dedicate time to making sure they're working as intended –?for business, for users, and for society.

And believe me: I know that entrepreneurs have enough on their minds already without figuring out what goes into making the algorithmic sausage.

Fortunately, the best solutions rely on something you (hopefully!) already have for your company: a strong set of values. If you make sure your values are the top ingredient in your algorithm's recipe, you can set up guidelines that will help direct the experience for your customers.

If you have a local reviews app, for example, think about how you want that recipe to look. If you want to focus on trusted, quality experiences, maybe you'd show spots with lots of high ratings first. If you want to encourage new businesses, you could boost posts from places that don't have many ratings at all.

And even better, there's a ton of data you can collect from how customers interact with your product that can help you decide if you should make any changes along the way.

One last thing: it's important to be as transparent as possible about how your algorithm works. That said, algorithms are like the secret sauce of digital platforms –?and they're often literally a trade secret. But even sharing what kinds of factors go into what shows up can earn you a lot of trust from your users.

And as we move forward into a more digital and automated world, that trust may be the real make or break factor for the success of a platform.