Tony Stubblebine
3 min readOct 6, 2022


Have mostly been replying to Jessica privately in case I'm misunderstandng her. But this response deserves more public visibilitity. A lot of people believe this but it's basically wrong

I have people telling me we changed the distribution model and it's been great for their page views. And I have other people telling me we changed the distribution model and it tanked their page views.

But mostly I think people are wildly underestimating how much traffic patterns can change for reasons of taste and competing articles.

The period that I think people are referring to had zero changes to the recommendation and curation part of distribution that involve Medium employee opinion or heuristics about what is worth distributing. I know that's very specific wording, but I'll explain the caveats below. I can't remember when the last major change was--probably six months ago at least.

And then the first change that I oversaw started on Monday and is minor as in less than 5% difference in results. It'll announce on our blog once it gets a bit more data behind it. I think most people are referencing page views swings of 50% or more and are referencing a time period prior to Monday. So I don't think this change is behind any of the opinions.

We did change three things.

1. We removed a notification about what was curated or not. The message had become severely outdated and was wildly misunderstood. But the changes that made the message outdated, namely a swing toward algorithmic curation decisions, happened well before my time. Nothing in this change of messaging should affect your page views or earnings.

2. We introduced a true following feed and reads by followers is up (despite what you might think based on your personal experience of this change). This should be good for high affinity authors. But for some authors and publications, readers reacted by realizing they don't want to read much of what that author/pub publishes and they unfollowed. Previously that feeling had been hidden by our recommendation algorithm because it could figure out which things an author published that you wanted to read.

3. We introduced a "show less like this" button to readers and are factoring that into recommendations. This is a change in the recommendation algorithm but it is reader driven. Is Jessica a divisive writer who generates a lot of these? I don't know the answer. Our current data is that this isn't a massive change in anyone's traffic and is fairly accurate in representing the interests of the readers.

There's a lot of detail in there. But #2 and #3 are the opposite of us "gripping the steering wheel." Those are changes that put the reader in control. That's why I don't think anyone should currently be thinking that their page views changed based on some whimsical changes that Medium made to the algorithm.

Now, that said, I do want to make changes to what we recommend, especially in advice categories. We have a lot of very credible writers who get swamped by under-informed, but more virally crafted content. The <5% change above is human driven. But it's too early to go in depth on that and you can count on me making an announcement on our blog when it's ready.