The Fuckery Files

A blast from the distant past.

The reasons are not entirely clear to me, but some of those savants in that one chat room I inhabit have developed a great interest in relics of the first age. Apparently the segue from Arab Spring to Occupy Wall Street to Gamergate contains clues to what might happen next.

I do not have time to wallow in their historiography, but I have been fitfully working at putting Disinfodrome back into limited operation … and I needed a nasty document test set that would fully exercise the system, rather than some neat FOIA archive. I saw that folder named “TFF” and I managed to smile, grimace, laugh, and groan all at once. Then it was on …

This is an instance of Open Semantic Search, the software used by Disinfodrome. The disinfodro.me domain remains down - this one is accessible via Tailscale. That’s enough for a small group.

The Fuckery Files are, for those of you who were not around fifteen years ago, are my early attempts at curating a complex conflict. Have you ever encountered one of the Conspiracy Brokers? Did you … Barely escape the Cauldron of Stupid? Have an account on the Computational Troll Dynamics Supercomputer? We’re talking the Anonymous class of 2011-2012, the Kookpocalypse, all that early weaponized ARG stuff, back when you didn’t need a grand a month in disposable income to drive one.

There are some VERY worrying signs here. I sorta cobbled together Solr7-mcp, a simple MCP server that works with the old version of Solr used in Open Semantic Search. And then I offered it to the long suffering OSS regulars.

I spent quite a bit of time, maybe on the order of five to seven hundred hours, back in 2019/2020, learning to hand build all the components that go into OSS. And today I forked the main Open Semantic Search repo. Then I used Claude Code and rigged up a way to reset an OSS appliance.

It would appear that SOMETHING is happening here … but I’m afraid to even give voice to WHAT that might be ..

The Xeon W-22335 upgrade moved my workstation from four cores to six. This is hyperthreading in action - that’s four cores that can each do two threads. It’ll be like this for many hours, perhaps a whole day to digest the total corpus of 27,000 files.

Existential Threat:

Having survived Model Mayhem earlier this week, it’s put some sharp edges on my moves. Formerly, I would have played ping pong with ChatGPT because it’s just right there in the browser, and I haven’t done much with Claude Code in the last while, I’ve been on integration duty. Now that I know the character of Gemini3, Claude Opus4.5, and ChatGPT5, I had a couple volleys with ChatGPT, then decided to pull out the big guns. Opus took what would have been a solid month of reading source code, taking notes, and testing theories, and gave me that reset procedure in under an hour.

So here’s the problem - as I’ve climbed the learning curve on this stuff, in particular getting into Antigravity, I have the deadly combination of an old man’s judgment and a young man’s energy. And this is before I even properly exercise the Claude Skills I picked up from Nate Jones last week.

Let me express this in automative terms. I’ve always had enough intellectual horsepower to “break traction” - I mostly work alone, because in groups I will get ahead of the rest, or suddenly go in some direction that they don’t immediately understand. Having suddenly bolted a supercharger to that … it’s welcome in the startup environment, but there are SO MANY TEMPTATIONS.

Wading into OSS is defensible - I used to make money with it, and may do so again one day soon. Even if that doesn’t come to pass, it’s going to be a fine framework for OCR testing.

The concern is that I’m having those “college age Neal deep hack” experiences. I got up at 0700 the day I wrote this … and were it not for a doctor’s appointment at 1600, I’d have gone twelve to fourteen hours, staying on my chosen task.

Conclusion:

I will relax a whole lot when we close on financing. Right now I’m even more of a hermit than normal