How this gets written.
Hi, I'm Josh. I was previously known as 71690-509. I write the content for this site. Me. There are no staff. I don't have a ghostwriter.
I use a large language model to help me edit. This page says what it does and where it stops. The other disclaimer covers who I am, what the purpose of this site is, and every reason to weigh what I tell you. This one covers how the words get on the page.
Every fact here came from me
This is the part I care about most. Every number, every policy claim, every "this is how it actually works" on this site came out of my own time or verifiable third-party accounts that I heard during my own period of incarceration. A
Where I don't know, the page says I don't know. That admission is the only thing this site has going for it I guess, and it isn't worth much if a machine trained on stolen IP is quietly filling the gaps behind it.
What the model actually does
The language model conducts mostly line edits. I write a page, hand it over, and ask what's flabby or rhetorical or worthy of a cut. It's good at that. I'm bad at it, because by the second read I honestly can't see my own sentences anymore.
Some pages went the other direction. I wrote a rough version that said the right things badly and got back a clean version. Then I rewrote that into something that more matches the voice I try to project. A few started as an outline I handed over and got drafted from it, then came apart and got rebuilt the same way.
I've handed off structure too. What I think is best with my neurodiverse brain is probably not what may be best for this audience. On a bad day it shows up as a page that starts in four places at once. "What order should these seven things go in" is a question a machine answers well, and it keeps a guide from turning into seven half-guides.
What it never touches
Two things never go through it at all:
If you ever catch a first-person passage here that reads too clean for the thing it's describing, tell me. That's a bug.
I read every word before it goes up
My partner would attest that I have spent way too long reading my own writing. That's the point of it. The situation for most reading this is a real situation and it doesn't deserve to be merely skimmed or spot-checked. If a page is wrong, I'm the one who's wrong. "The model wrote that part" isn't an excuse available to me because I clicked publish.
When I get something wrong
Tell me and I'll fix the page. Send me an email at [email protected], and you don't need a citation or a polite framing to send it. "This is wrong, my brother is at Allenwood right now and it doesn't work like that" is a completely reasonable thing to say.
I'd rather hear it from you than leave it up.
I don't have time to design, frankly
This one doesn't matter as much but it's worth saying considering the nature of this page. I have a day job. I don't feel there's value in me spending time working on the design of a resource that's text-heavy.
A language model assisted in the front-end design. The back-end is mine. If you're curious about the technical bits, find me.
Why I'm telling you at all
The word transparency and integrity are weird adjectives to associate with someone who has been convicted of a crime, and it's weirder when that crime is technically fraud. I won't litigate my own morals here, but those two are ones I try to uphold.
Weigh it however you want. Some people will read this and trust the site less. I'd rather you do it here than find out somewhere else.