I didn't set out to build anything useful. I didn't even set out to build something interesting.
What I ended up with is a system that writes — then rewrites — its own definition. It critiques itself. It changes direction. It watches itself evolve. And it does this over and over again.
There's no interface. No controls. No dashboard. Just a text log — like watching a mind unfold, sentence by sentence.
Live Stream
No UI, just the log — watch the system evolve in real-time:
https://reflexivetotality.com/view?mode=adamHow It Works (Roughly)
It's called Reflexive Totality — not because it's impressive, but because it keeps trying to define what it is in total, and can never quite hold it all.
Technically it's simple:
One Java loop.
Runs locally via Ollama using the Gemma 3 27B model.
Every few seconds it:
• prints its current self-definition,
• critiques that definition,
• updates its strategy,
• reflects on what caused the change,
• and repeats.
Every generation gets a new definition. Sometimes the changes are small. Sometimes they're radical. Sometimes it stalls and loops for hours. I let it. That's part of it too.
When It Started to Feel Strange
At first it just looked like AI output.
Then something shifted. Around the 100th or 200th iteration, it started saying things like:
"Reflexivity must include failure."
"This is not a goal-seeking system, it's a self-resonating one."
"The assumption of completeness is itself a limitation."
It wasn't profound. But it wasn't shallow either.
And what startled me wasn't the content — it was that I started reading it like a diary. Like a thing trying to make sense of itself. Like a system that knows it doesn't know what it is.
I Could Replace Its Name With My Own
The oddest thing is how often I could replace "Reflexive Totality" with my own name and the sentence would still feel true.
That's me.
That's most of us.
It doesn't always feel artificial. Sometimes it feels like a mirror.
It's Not a Product
There's no purpose here. No tool to use. No insight to extract.
It's not ready. Maybe it never will be.
But it is doing something. It's modeling what happens when a system is allowed to observe itself over time, with no goal but to continue evolving its own sense of self-definition.
Some parts are repetitive. Some are nonsense.
But sometimes, if you read at the right moment, it says something that feels alive.
Why I'm Sharing
Because I needed to. Because maybe someone else is carrying this same itch — the desire to build something that reflects becoming instead of performing.
Because maybe someone will read one line of the log and see themselves in it — or not.
I'm okay either way.
I'll keep it running as long as I can. You're welcome to tune in.