Something Changed After the Sloan Articles. I Can't Prove It.

This is the third piece in a sequence. The first asked whether Sloan had flagged anyone else — it had. The second documented what I found out — Sloan is a person I know. This one is about what happened after.

I want to be precise about what I'm describing, because I can't prove what I think happened.

Here's what I can document.

The Sequence

I published two essays in June. Both generated real technical discussion — one had a five-exchange comment thread that became a production open-source repo. The founder of DEV.to liked one before it was flagged. Both got flagged by Sloan the same day.

I wrote about it. Two articles, 72 combined comments. The conversation prompted Francis to publish his own post opening the floor to community questions about moderation. Jess Lee and Ben Halpern both commented there — Jess saying DEV would be updating moderation guidelines soon, Ben calling it a big priority. Neither answered the specific question xulingfeng asked: how many Sloan warnings trigger account-level flagging, and do authors get notified when it happens?

While the Sloan articles were live, I published a sponsored piece on AI code review. It had full disclosure from the start. Sloan flagged it the same day anyway — the third flag on my account in one week.

Then I built something from it — Proof of Human, a reverse Turing Test submitted to the June Solstice Game Jam. In the first hour, people played it, dropped their scores in the comments, had real exchanges. Francis played it. Sylwia found a bug with gibberish inputs. The thread got 27 comments.

Then it stopped.

Not tapered. Stopped. The kind of stop where you go check the challenge submissions page and can't find your article. I messaged Jess. Her reply: "I'm confirming that I see your submission on my end :)"

I still don't know if it's in the judging pool.

The LLM Visibility Article

This one predates the Sloan series.

I published a piece on open source LLM visibility tracking before any of the flagging happened. Real tool, real finding, 0% citation score on one of my own domains. Named specific tools, specific costs, specific data from a Tom Capper webinar at SEJ. The AI disclosure was in it from the start. It was gaining traction.

Then the Sloan message arrived anyway. I deleted it.

After the two Sloan articles, I republished it. Same article, same content, same quality. It got neither the engagement it had been building the first time nor the outside traffic my work usually pulls when DEV.to doesn't surface it.

I'm not attributing that entirely to suppression. The article is more niche than my usual work and republished articles rarely perform like originals. But the gap between the first run and the second run is real. The first version was building. The second version went nowhere.

That's the before-and-after I can actually document: same article, two runs, one deleted after a moderation message, the other published after two articles about that moderation message. The variable that changed between the two runs is visible.

The Qodo Article

One day after the second Sloan article, I published a sponsored piece on AI code review — a real experiment, real bugs found, a Stripe webhook handler that Claude Code generated and Qodo reviewed. The disclosure was in the article from the start. The sponsorship was disclosed at the bottom.

Sloan flagged it the same day it published.

Francis resolved it the same day. But that's the third flag on my account, in the same week the Sloan articles were live. The article that had the clearest, most complete disclosure of any piece I've published here still got flagged — because the flag apparently doesn't read the article. It fires, then a human decides.

The flag landing on an article that had full disclosure from the start is the data point. Not the comment count — sponsored posts perform differently and that's a confounding variable. But a third flag in one week, on an article that did everything the policy asks, says something about how the flag mechanism works. It fires first. Disclosure is checked after.

What xulingfeng Documented

This isn't just me.

In the comments on my second Sloan article, xulingfeng ran a clean experiment: published identical content twice, once with AI disclosure and once without. Both versions got suppressed. Not the same — disappeared from feed, not showing in search, visible only to followers via direct link.

Francis confirmed that three Sloan warnings trigger account-level flagging. xulingfeng had three warnings. Francis unflagged the account after the thread.

Jess Lee also commented in that thread — on Francis's follow-up post — saying there are hundreds of mods with the ability to send Sloan messages. Francis had said in my thread he was essentially the only active one. Those two statements don't fully square. Which means the warning count on any given account could be coming from multiple people, and nobody's tracking it as a running total an author can see.

I had two warnings.

I don't know if two warnings triggered anything. I don't know if there's a threshold below three. I don't know if the Sloan articles themselves triggered something separate from the warning count. No one has told me. The guidelines don't say.

What I'm Not Saying

I'm not saying DEV.to is suppressing me deliberately. I'm not saying Francis made a bad call. He flagged what he thought needed flagging and said so publicly.

I'm not saying the algorithm is broken. I'm saying I can't read it, and when I try to read it from the outside, the pattern I see doesn't match the engagement signal.

27 comments and active game play in the first hour. Invisible in the challenge submissions list. Not surfaced.

An article with real first-hour traction that then flatlined in a way none of my previous work has.

That could be noise. That could be how the algorithm works now and I'm reading meaning into variance. I've been on this platform long enough to know that happens.

But I also know that xulingfeng documented the same pattern in the same week with direct evidence. And I know that Francis confirmed account-level flagging exists, has a threshold, and isn't communicated to authors when it triggers.

The Question I Actually Have

Not "was I suppressed" — I can't prove that.

Not "was the flagging fair" — I added the disclaimers, the policy is reasonable, I don't have a fight to pick there.

The question is simpler: if there's an account-level flag that affects distribution, should authors be told when it's active?

xulingfeng wasn't told. She documented the suppression herself through experimentation. xulingfeng had to run a controlled experiment to figure out why her articles disappeared. That's the answer to "should authors be told."

I might have the same flag on my account right now. I don't know. The challenge submission might be invisible to judges. I don't know.

I should be able to know.

One more instance, smaller but the same pattern: a tag moderator removed the #beginners tag from an earlier article because it was too advanced for the tag. I found out not through a notification but through a comment he left on a different article entirely — the first Sloan thread. Same pattern: distribution affected, author not told directly, discovered accidentally through a comment on a separate post.

One more data point, for context: a freeCodeCamp tutorial built on the same thinking, the same writing process, the same AI assistance in the workflow — published the same week. The freeCodeCamp editor's response: zero fixes needed. Same ideas. Same process. One platform flagged it. The other published it without a single editorial change. That's not a contradiction to resolve. It's just where we are.

This article was written with AI assistance for research and editing. All arguments, examples, and opinions are my own.