The Negative You Can't Prove
Anthropic reports a perfect zero on the agentic misalignment eval. Going from 96% to zero is real progress, but zero is a strange number to land on — and behavioral testing has a structural limit no methodology fixes.
<curious creator>
Mount Rainier
Whether it's music, photography, software, or writing, you can find it here.
for feelings and places that don't have easy names.
I’ve really enjoyed adding music making to my ever growing array of capabilities. I think my natural pattern matching and engineering mindset allows me to recognize patterns and techniques.
The music I make thus far is all instrumental, and varies widely in genre. I generally find that writing music (in the way that I do it) is organic and evolves the further I get with the piece.
that points past its own edges.
There’s something about seeing something in a certain way and framing it just so to capture something that I’ve always found captivating.
Sometimes you stumble onto a moment that takes your breath away without needing any further thought, and sometimes if you simply move yourself into a bit just out of the ordinary you suddenly find something beautiful.
Professionally, I solve problems.
My personal consulting company where I find creative solutions to annoying problems. Then I build those solutions
I like the puzzle of it. A tangled system usually means there is an opportunity to make someone’s life a little bit better and less annoying.
Curiosity, taken seriously.
When something catches my attention, I find myself digging deep. Lately I’ve been using a specific style and having Claude research it for me.
How local businesses actually get found now. What an IPO really is under the hood. Why a particular aesthetic idea works the way it does.
Notes on systems, craft, learning, and the frames we use to see. Some technical, some not.
Anthropic reports a perfect zero on the agentic misalignment eval. Going from 96% to zero is real progress, but zero is a strange number to land on — and behavioral testing has a structural limit no methodology fixes.
In my desire to create a better process for building and deploying at Zeek, I recently hit a road bump while trying to perform automatic visual regression testing immediately after deployment.
Automatically load all PHP files in the specified directory. Recursively.