Post

Seeing the usual suspects dismiss this as delusional, cope, etc., but if you were involved in moving this needle beforehand, you will have felt the balance of power shift in a meaningful way.

And if you aren't paying attention, this probably reads like advocating to pay Anthropic and OpenAI to be able to maintain software for the rest of our lives. But open weight models are just unbelievably good now! Like, you actually wouldn't believe it unless you saw it yourself.

I made the decision very early on that there was no way I'd be using proprietary models for the same reasons I avoid as much proprietary software as possible in other aspects of my life: I don't want to form a dependency on something that can be changed or revoked without my input.

It wasn't until April of this year that I started taking the way the models were progressing seriously; in the last 3-6 months, we've reached a point where not only are open weights good enough for the vast majority of generalized development tasks - they're getting cheaper to run. Fast.

You could blow up OpenAI and Anthropic tomorrow, but there's no dependency on them anymore. The weights are out there. Anyone with a rack of GPUs will sell you inference for pennies on the dollar, in whatever jurisdiction you like.

So what can this do for you and your community? A lot of things, and I think that's something we need to get better at communicating and writing about generally.

For instance, maybe you have some e-waste in your home from an IoT company that shut down and you just want it to do what you bought it for again: