Husband | Father of 7 | Kwisatz Haderach at https://xitter.cf/t.co/nWnqH6zUL7 | Pro generalist software man | iOS/Android/Elixir | Noob Blender user and farmer
Somewhere in Arizona
Joined September 2010
- Tweets10.5K
- Following1.1K
- Followers469
- Likes7K
My @bot wants his machine back. If it can't be relied on, can it really be a personal assistant? Seems like #1 prerequisite.
Free Elixir! @dhh
So I was trying to understand why pull requests speeding up the Elixir version were being denied by the clankers while they were being accepted to Go/Express/Rust.
It turns out that, not only each project had a different starting point and a different prompt, but also each project also has a different set of verifications that they are required to pass!
For example, Elixir is the only guided to "preserve the pinned Rails reference behaviour", the only with a explicit verification that we are using Redis and draining its queue (so we cannot remove it while others can), and the only one actively enforcing it.
So it turns out that the people telling you to not read the code cannot even use AI to set up fair benchmarks (probably because they would have to read the code in order to do that).
You don't have to believe me, you can do your own research or ask your agent to do it for you. E.g. here is what Claude Code came back with: gist.github.com/josevalim/ae…
Bryan Bryce – e/tacos retweeted
So I was trying to understand why pull requests speeding up the Elixir version were being denied by the clankers while they were being accepted to Go/Express/Rust.
It turns out that, not only each project had a different starting point and a different prompt, but also each project also has a different set of verifications that they are required to pass!
For example, Elixir is the only guided to "preserve the pinned Rails reference behaviour", the only with a explicit verification that we are using Redis and draining its queue (so we cannot remove it while others can), and the only one actively enforcing it.
So it turns out that the people telling you to not read the code cannot even use AI to set up fair benchmarks (probably because they would have to read the code in order to do that).
You don't have to believe me, you can do your own research or ask your agent to do it for you. E.g. here is what Claude Code came back with: gist.github.com/josevalim/ae…
Also extracted a separate verification project that's responsible for the benchmarking and checking of new implementations and improvements to the existing ones. The full report from the latest runs is there too. github.com/basecamp/once-cam…
I sold my second big client on building out their business operating/marketing software at $2,500/month. Probably low but could it lead to several more sales/faster implementation in the same space. My bespoke business software thesis is starting to get some real-world evidence and the latest models are making it possible to deliver well.
Bryan Bryce – e/tacos retweeted
I have asked my clanker to send some PRs for the Elixir one. Please ask your clanker to verify them and tell your clanker we are allowed to deviate from the Rails version in the same way Express/Go/Rust/ do (it was objecting to that in other PRs)!
The clanker says there is more work to be done but I don't have infinite tokens, so it would be nice to see those merged first.
🤫 competitive advantages are getting harder to find, stop telling people about Elixir
with all the dunking on @elixirlang, a quick note from someone who bet on it.
phoenix liveview is processes all the way down. that is what lets us build interactive ui without the overhead and indirection of client side state management. we ship and test interfaces insanely fast and nothing else has come close.
and that is just liveview. @elixirphoenix is one of the most productive web stacks i have tried.
the erlang runtime is the other half. agents can attach to it in dev and in prod and see the process tree, memory and cpu, not just symbols. it is a system you can interrogate.
we have not used everything it can do because we have not needed it yet. when something does need to be fast we can drop into a nif and write that part in rust, and so far that has barely come up. we optimized for feeling productive so we could get the business off the ground. @marekfort and i say the same thing every time. best decision we made, and even better with agents.
Bryan Bryce – e/tacos retweeted
DHH’s VibeGate keeps getting worse
Shows stats across 3 repos
Rust repo had 300+ commits, heavily tuned
Elixir & Go had only 3 commits
But Elixir is slow … sure
Oh FFS. It turns out the req/s in the initial vibed Rust version is within the same order of magnitude as the initial vibed Elixir/Go versions, and even performs worse in some of them.
The history is in the repository! The Rust version only got faster with a bunch of additional work. The initial Elixir/Go versions presented in the tweet below saw none of this additional work. It was comparing a 300 commits Rust version with a 2-3 commits Elixir/Go version.
Source: github.com/basecamp/once-cam…
Bryan Bryce – e/tacos retweeted
The plot thickens! Rust wasn’t actually faster than Elixir initially???
Oh FFS. It turns out the req/s in the initial vibed Rust version is within the same order of magnitude as the initial vibed Elixir/Go versions, and even performs worse in some of them.
The history is in the repository! The Rust version only got faster with a bunch of additional work. The initial Elixir/Go versions presented in the tweet below saw none of this additional work. It was comparing a 300 commits Rust version with a 2-3 commits Elixir/Go version.
Source: github.com/basecamp/once-cam…
Bryan Bryce – e/tacos retweeted
What frustrates me the most is the lack of curiosity.
People saw the numbers they wanted to see and ran with the story. They didn’t ask about the process. They didn’t question the results. They didn’t explore the code. They didn’t look at the history.
You could have even asked a coding agent to do an initial investigation.
There has never been a better time to be curious, dig deeper, learn... but this is what we’ve got now.
Oh well.