Down the Rabbit Hole with Nextdoor’s AI Censors

By Paul Krueger

When I tell friends that Nextdoor “temporarily suspended” my account for no good reason, they just roll their eyes and ask why I waste my time on that social media network.

They’re correct that Nextdoor has its share of trolls, extremists, and dim-wits, but those aggravations are no worse than what I encounter on “X.” (I post almost exclusively about news and politics, and find Facebook users to be the most engaging and civil of those three sites.)

Nextdoor’s neighborhood focus also has distinct advantages: I can post information about issues and problems in my Talmadge/ Kensington/ College Area, generate interest and attendance at meetings, answer other resident’s questions, and get new and helpful information from readers who share a common interest.

When one of my Nextdoor posts generates significant interest and discussion, it really can go “viral.” I’ve had seven thousand or more local views on an issue, and my local viewership routinely exceeds what I get on X or Facebook. If someone disputes a post, I respond politely with the facts as I see them. And if my attempts at a meaningful conversation go south, I just ignore it and move on.

That’s why it mattered when Nextdoor recently notified me that my account had been temporarily suspended for posting “spam.” I was puzzled, because the post in question could in no way be classified as spam, which Nextdoor defines as “unwanted, unsolicited, and/or repeated actions that negatively affect neighbors and the Nextdoor community.”

Nextdoor’s AI-sounding notice informed me I had violated its spam prohibition by reposting a legitimate news story from a legitimate news source: a Times of San Diego report about how the animal rights group PETA had criticized San Diego Police for using a K9 to subdue a suspect armed with a torch, which burned the dog.

I’ve reposted numerous stories from local and national media — the UT, Voice of San Diego, Times of San Diego, Axios, New York Times, and Wall Street Journal, etc — and never been warned, much less suspended — for any alleged violation of Nextdoor’s guidelines.

This recent  suspension rendered my account inaccessible to my neighbors (and myself) for four days. It came without warning, and it was patently incorrect and unjustified. So I immediately appealed, explaining in polite, well-reasoned language that my repost could not, by any definition of the word, be considered “spam.”

I was supremely confident that no matter who, or what, read my appeal, logic would prevail, and my account would be immediately reinstated, maybe even with an apology.

What I got instead was a generic, unsigned response titled “Appeal not approved,” informing me that “on second review, your account was determined to have violated the community guideline: No spam.”

An internet search yielded the suggestion that I escalate my appeal and specifically ask for a “human review.” I did that, and when no response was forthcoming, I tried to contact my Nextdoor neighborhood “content moderator.” That search also yielded no information, but another internet query told me Nextdoor does not reveal the identities of its volunteer moderators, who, more to the point, reportedly have no power to suspend or otherwise discipline Nextdoor users.

By then, I’d  served my four days in social media purgatory, and attempted to access my Nextdoor account. No such luck: “Access Denied” read the message.

That request for help did yield a response from what might have been an actual human being. Though the reassuring email that told me to try logging in again was sent from support@nextdoor.com it was signed by “Ron Michael.” (Yes, the use of two first names is suspicious, especially under the circumstances, but in good faith, I reached out to Mr. “Michael,” asking, again, for a human, if not humane, explanation for my unwarranted suspension.

I waited… and waited… and am still waiting for a response.

So I’m left to figure out why Nextdoor targeted me for that unwarranted discipline. Though I’m definitely not the paranoid type and tend to understate my influence on local issues, I suppose it’s possible that that someone who takes offense at my frequent criticism of city government, the Mayor, and my city council member (Sean Elo-Rivera), gamed the Nextdoor discipline system to temporarily silence me.

But it’s much more likely that my suspension was probably a random surveillance error compounded by a cost-saving AI workforce that lacks human oversight. (Others who have been disciplined are less charitable about Nextdoor’s review process, including this former Connecticut state legislator.)

I’ve also since heard from other Nextdoor users who’ve been disciplined or banned for no apparent good cause, and whose questions and appeals yielded no information. Some of them didn’t bother to appeal their discipline, and now think they’re wiser than I am.

I spent my working life as a journalist committed to righting wrongs, big and small. But my experience with Nextdoor taught me you can’t persuade a bot to answer an unscripted question, admit a mistake, or even give you a “no comment.”

As my wife frequently tells me, I’ve got to “pick my battles.” She was definitely right on this one. If you find yourself in AI hell on an issue of questionable importance, don’t waste your time trying to clear your name.

The bots don’t care.