What’s Waverly and Why I Dreamt of It

Waverly is a discovery engine for your better self. When you install our iPhone app you gain a superpower: the ability to craft your ideal content platform. A place you’ll love to go to, that will be free of doomscrolling, that will promote the stuff you really care about and which, ultimately, will make you feel good about how you spend your time on the Internet.

You’ll need to work a little to get that superpower, though, but it’s worth it. More precisely, you’ll need to tell Waverly how your ideal content platform should look. You do that by writing waves. Short paragraphs that describe the things you love, using words that feel right to you.

Here’s an example of one of my waves:

You can steal this wave and tweak it to make it your own. If you prefer, you can write something from scratch. Waverly understands what you’ve written and uses it to guide your discovery experience.

You can write as many waves as you want. You can augment them or change them as your interests evolve. You can turn them off when they’ve run their course. Waverly’s promise is to abide by your waves. If you change something, the effect is instantaneous.

Why did I dream of Waverly?

Angry, divided, polarized, addicted… These are some of the words we hear more and more frequently to describe the people and communities we find online. Are we really that bad? I don’t think so. I believe people are better than what we see on the Internet. We make our tools and then our tools make us, and, right now, I believe our algorithms are making us.

Let’s rewind a little…

In 2015 I was working at Google on Chrome’s recommendation engine — the 10 links you see when you open a new tab. I felt the way in which we were building it was wrong. We were recording what users clicked on, how much time they spent on different web sites, the time of day at which visits occured… And from the accumulation of this data we computed a score for all web sites in a user’s browsing history. A high score meant that the user was likely to click on this site, therefore Chrome recommended it.

People who knew I was working on this feature often asked me for more diversity. I’d love it if Chrome would throw something unexpected in there every now and then, they would say. Yet anything we tried to increase diversity resulted in a lower click rate, and these experiments we scrapped. The saying internally was simple: Look at what people do, not at what they say.

I felt something in that was wrong. In fact, when I looked at my own data, I found a lot of clicks I came to regret. Addictive videos I had spent too much time on, angry clickbait I should never have engaged with… my own history was full of crap, and that crap was feeding the algorithm. Even worse, I knew that if another piece of angry clickbait came up at just the right time, I might very well end up clicking on it again — even though I’d regret it a few hours later.

Chrome is not an exception. All the recommendation engines out there work in a similar way. Facebook, Twitter, YouTube, Apple News, Tik Tok, Instagram… They all abide by this same rule: look at what people do, not at what they say.

So if we were to test my hypothesis that people are better than what we see on the Internet, we had to flip the script on our algorithms. We had to create an algorithm that would listen to what people said.

That’s why I dreamt of Waverly and that’s why our team built it. So we could see what kind of world we would create if the Internet was populated by our better selves.