DeepBlue Dynamics / Signal Log / the-slop-tariff
· manifesto · policy · ai · moderation · network · Kord Campbell

The Slop Tariff

Ad-subsidized aggregators host coordinated AI-shaming for free, then sell the outrage back to you as engagement. Not on this network. Not anymore.

The Tariff of Abominations, 1828

Effective now, every device on this network sits behind a DNS-level block against ad-subsidized news aggregators that make no attempt to police “slop” comments: the genre where a human's entire contribution is accusing another commenter, or the author, of using AI instead of engaging what was actually said.

This isn't a browser extension. It isn't a setting I'll forget to re-enable after an update. It's enforced at the router, in front of every device on the network, including the agents.

What actually gets blocked

Two conditions. Both have to be true.

Subsidized. The site's business model runs on ad revenue and engagement, not on readers paying for the thing directly. The incentive is clicks, not curation.

Unmoderated slop-shaming. The comment section has no visible policy against the pile-on where “sounds like AI” functions as a complete rebuttal, and nobody on staff treats it as the drive-by insult it actually is.

A site can run ads and stay off the list if it enforces standards. A site can be free and stay off the list if the comments are actually policed. It's the combination, subsidy plus tolerance, that gets a domain added.

Why this isn't a mood

I've made this argument twice already, in AI Slop is an Effort Problem and The Rise of the Systemmer: the crisis was never that AI makes software bad, it's that AI made low effort cheap, and the sorting event underneath it separates people who can steer a system from people who can only complain about one.

“Sounds like AI” is not a critique. It's an admission that the critic has nothing else. It costs nothing to write, it requires no engagement with the actual argument, and it works precisely because platforms let it work: outrage is retention, and retention is the product. Subsidized aggregators have no incentive to fix the comment section, because the comment section is the engine.

So I'm removing the incentive on my end. No traffic, no ad impressions, no algorithm signal, from any device I control.

Architecture, not the argument

Same lesson as the last post: appeals to good faith don't survive contact with an incentive structure. You cannot moderate a stranger's comment section by replying in it. You can only control what you route past your own gateway.

Waiting for a platform to reform itself, or for public pressure to shame it into hiring moderators, is the same decade-long war of attrition as waiting for a court to rule your way. The fix isn't a strongly worded email. It's a blackhole entry on the DNS resolver every device on this network already uses.

Where this goes

This is a personal network policy, not a campaign. But the test is simple enough that anyone running their own resolver (Pi-hole, OPNsense, pfBlockerNG, whatever) can apply the same two conditions to their own gateway. It doesn't need my list. It doesn't need anyone's permission, cooperation, or vote. That's the entire point.

We're done pretending an unmoderated pile-on is a comment section. It stops routing here.

Two names, and what I'm building instead

If you're turned off by the use of AI for anything creative or constructive, there's the door. I'm done defending gatekeeping to the gatekeepers.

Reddit is a public company now (NYSE: RDDT), dual-class shares, insiders and its largest shareholder holding most of the voting power no matter what the float thinks. Your posts are the product. The vote on what happens to them was never yours. Hacker News runs out of Y Combinator, an accelerator whose partners talk in valuation multiples and hand out corporate doublespeak as if it were founder advice. Neither one owes its users anything, and neither one moderates like it does.

So I'm building my own link-sharing site. Any substantive, grounded post gets in. Anything else gets flagged, for one of two reasons: too much text with no actual argument in it, or just low effort. There's a real difference between spending two hours on an argument, the way I did on the last post, and typing “write me something conflicting about sovereign architecture” into a chat window and pasting the output. Both produce paragraphs. Only one of them is thinking.

Shills outnumber builders who ship. They always have. I'm not interested in running a platform that optimizes for the larger group.

So suck it, nerds. I'm sailing my own way.


Built and shipped by Deep Blue Dynamics. Reach Kord at kord@deepbluedynamics.com.

// transmission ends