In August, Medium, the publishing platform used by apologetic CEOs, PR departments, and freelance writers, announced it was introducing a new way to react to posts on its site, as part of a larger brand redesign. Medium started the year with sizable layoffs in an attempt to monetize the site beyond the traditional advertising models employed by most online publications, which founder Evan Williams (having previously cofounded Twitter and Blogger) believed were “broken.”
Medium put up a monthly $5 paywall earlier this year, which publishers could choose to put content behind, and opened up the pay model to its individual writers on Oct. 10. Quartz recently spoke with Williams about Medium’s new focus, and why it had chosen to design its new method for appreciating content around a clapping emoji button that it calls “Claps.” The company is using how many times readers press this button, among other metrics, to figure out how much to pay individual writers.
This interview has been edited and condensed for clarity.
Quartz: Can you explain the logic behind the claps mechanism?
Williams: We put a tremendous amount of thought into how do we actually reward the right stuff and really get that feedback loop so that the better something is, according to the subscribing reader, the more it gets rewarded. And for example, the worst possible version would be to pay on views, because then you’d have the exact same incentive structures you would have with an ad model.
So what are the other signals we can look at that would identify something is not only high quality, but highly valued by those who are actually paying for it? There’s lots of metrics we can look at: We can look at views; we can look at the read ratio; we can look at time spent. We can look at what was our one explicit signal before, which was the recommend [button]. And we thought, well, what would the dream scenario be? The dream scenario would be if every time you read an article we could ask you, “Well, how good was that relative to other things?” Which is essentially a rating system.
Because anything, a “like” or a “recommend”, a binary mechanism, is useful and is simple. But anything, whether it’s a restaurant or a book, like, saying, “good” or “not good”, that doesn’t really give you the depth of meaning that is important. And when it comes to information value or story value, longer things maybe you should be paid more because, you know, they take more work. Often they can be more value. If someone actually read it, that’s sort of a proxy for value. But rewarding length in and of itself didn’t seem quite right.
So the claps mechanism is the closest thing to a real rating system without making it feel like a rating system. Making it feel organic and something you didn’t have to think about too much that really captured that degree and variability of sentiment. It’s per user, so it’s relative to if you give one to three claps to everything, then we parse that out. We parse out the [subscription] money that is earmarked for your subscription based on that. And if I give, like, 30 claps to everything, we parse it out relative to my claps.
If you just view something and spend no time on it, it won’t get rewarded in our system. And if you think, like, “This thing blew my mind,” and you, like, really want to just go nuts on that, it will get a disproportionate reward, which I think is the closest thing to actually rewarding what people are valuing. It gives a lot of agency to the reader, which is important to them and, I think, important to the writer in order to say, “I may not write as frequently, or the longest things, or the most popular things, but if I’m able to build an audience who really loves what I do, that will be reflected more in the system more than it would in another system.”
I see. What was the decision behind the actual name, “clap”?
There was lots of debate, as you can imagine. There’s this little experimental forum in Medium, in the app, called Series, which is a mobile centric storytelling format that is only available in the Medium app. It’s serializable, so you can add to it over time. You get to a tappable story thing. Instead of the “recommend”, the team came up with this thing that was like a claps emoji. And you could hit it multiple times. You could even hold it down and it became a sort of exploded emoji. It was just fun.
Because they’re serialized, the binary “recommend” actually didn’t make sense, either, because you can come back to the story multiple times. We found it actually did sort of capture the sentiment or level of appreciation you had for these stories. Like, if it was awesome, you would hit it more.
We went through a bunch of ideas—we were, like, “Is it multiple hearts and is it stars or is it a rating system?” And we decided on the claps because, it’s the most natural metaphor there is to what happens in real life. While it seems kind of goofy, it’s not goofy in the real world. It applies to art. It applies to lectures. It applies to conference speeches. It’s a natural thing people do to show appreciation and reflect how they feel about something. And it’s kind of fun.
Well, we give applause, we don’t give claps.
Right. So in certain parts of the product, we say, “applauded.” But what is it when you get applauded? The language is kind of funny. The picture works better than the language. And yeah, we’re hoping people get used to it.
It’s like moving from stars to hearts on Twitter.
It is. Although, the opposite, because they were, like, “Ah, hell, let’s just do what everybody else does. Let’s call them ‘likes.’” You know, we had the heart forever [on Medium]. And I never loved that, frankly. Like, we called it a recommend, which I’m still a fan of. And there’s a few other things that use “recommend” now. At the time, I had never seen anything that used “recommend.” It was a very intentional word. It was a big word. We wanted to make it more meaningful than a “like.”
But then we were trying to iconify it and we just couldn’t come up with something that meant “recommend”, like, in a picture. And because it is essentially a “like”, it was, like, people will get what the heart means, even though it doesn’t feel like quite the right thing for an article.
It’s, like, heart? It was exacerbated when we said, “Well, there’s multiple,” because multiple hearts feels even worse. We had a million different mockups. We had, like, floaty hearts, kind of like Periscope. And it’s, like, “No, this is, like, thoughtful, brainy stuff.” It’s not multiple hearts. I mean, it can be tear-jerking, sentimental stuff. But it just didn’t feel right.
People have not had the best reaction to claps.
We’ll see if people get over it. The adverse reaction, I have to admit, was much stronger than I expected.
People have a hard time with change. It was not just change, it didn’t mirror how other systems work. And the idea you can do it more than once. Which I guess is not intuitive, but to me, you know, we spent a long time thinking about it. We get a much higher fidelity of information by capturing zero to 50 on a scale of response than we do with one. For the system itself, it’s meaningful. And so for all these different reasons, even information discovery, we thought it would be useful. Some people said it was like Facebook’s different types of responses, but that’s actually completely different. Different. It’s not degree, it’s…
People really had trouble wrapping their head around why the change. What I saw when we announced how it was tied to money, that was also misunderstood because it was, like, “Well, why couldn’t your friend just clap 50 times and you get all the money?” Well, because that’s not how it works. But then, I did see many people, especially the writers on Medium, being like, “I didn’t get the claps thing. But now that it’s tied to money, now I get why this makes sense.” But we’re still in a learning process. And it’s a “v. 1” of a new system that hasn’t existed before. We’re still learning about this and about is it the right UI? Is it the right metaphor? Are we showing the right information?
And there is a big debate about do we show the total number of claps to the people that clap or the combination or the average or, like, some graphic? And all of that is very “v. 1.” And I think there’s probably things that we might tweak on that. Whether we show, we don’t show. There’s so many nuances to it. Like, if you click on the number of claps, you can see actually how many people there are. If you’re the author, you can see how many each person did. But if you’re a reader, you can’t see how many each person did. It is definitely a signal back to the author. And would the authors be able to see, like, “Oh, super fan”? And what the limit is. We felt like it needed a limit, but what is the right limit? We needed to see a lot of data before we really knew that.
But it was funny because we obviously thought about this a lot. It’s an experimental, new thing. And it’s, like, “No, wrong, stupid. Worst idea I ever heard. Like, this is retarded. You guys didn’t even think about this.” Oh, thanks internet. Great.