A lot of people have lately been complaining that Substack has only narrow restrictions on the content it allows. Some aren’t just complaining but are leaving the platform. The issue is “racist or bigoted speech”, and in some cases, “explicitly Nazi” material.

It wouldn’t bother me if Substack had somewhat broader restrictions on content, but it’s a dangerous path to go down, and I’m sure they know it. Deciding whether something violates content restrictions is often tricky, and sites with lots of user-generated content rely on software and people under time pressure to decide. They generate a lot of false positives. On pre-Musk Twitter, I was suspended for making a joke about the health hazards of Krispy Kreme, for recommending a sharp blade to separate uncut book pages, and for objecting to an endorsement of mass murder. The innocent were punished more severely than the guilty, since reinstatement required admitting to a violation of the rules.

Sites like Twitter and Facebook have to be careful because they shove unsolicited content at users. Substack is different; you get only what you subscribe to. (Likewise with Bluesky, which I’m enjoying a lot.) If you don’t like a newsletter, you don’t have to see it. There’s a cost to content moderation which Substack has decided to limit.

Think about it this way: You start by banning explicitly Nazi content. Fine, but the people who produce it can make a few cosmetic changes and stay within the rules. To have real enforcement, you need to ban content which is Nazi-like in character. This could include some pro-Palestinian material. The cry of “Gas the Jews!” was heard at a rally in Australia, and swastikas have been seen at at least one protest. “From the river to the sea,” which can be interpreted as calling for the abolition of Israel, is popular. Once you start banning them, though, it’s easy for people who are simply upset by reports of Israeli war crimes to get kicked off.

If you’re going to ban Nazis, you should also ban supporters of Communist dictators such as Stalin, Mao, and Castro. That could mean banning Bernie Sanders, who’s said some favorable things about the Castro regime. The limit on what to ban is always going to be ill-defined.

Automated reviews sometimes fail to distinguish between advocacy and reporting or education. YouTube has demonetized or banned numerous channels because of inability to recognize context.

Would banning Nazi-like material even limit the spread of the ideas? It’s more likely that it would let the authors pose as martyrs, and the people who could no longer read them through Substack wouldn’t be any less inclined to support them. A broad ban would carry a lot of risks with little benefit.

See also “Substackers Battle over Banning Nazis” by Elizabeth Nolan Brown.

My complaint with Substack is a different one. When you get their newsletters by email, every link in them is a tracking link. When you click on one, they know you clicked it and where you went. Several email services, such as Mailchimp, do the same thing. People who complain vociferously about Internet privacy are oddly comfortable with this issue. You hardly ever see complaints about it, but it strikes me as very intrusive. I dropped a couple of my Substack subscriptions for this reason and currently have only one subscription, to a newsletter by a personal friend.

That’s a different issue, but I don’t want anyone to think I’m a big fan of Substack.