It's insane how this book was published in 1995, and there hasn't been better material on NURBs modelling since.
Thought experiment:
If codex predicts my follow-up messages accurately, am I rly needed at that point?
If labs can perfect prompt prediction, won't the models just endlessly post-train on auto-generated threads to become intelligent beyond humancomprehension?
Internet censorship is getting worse (better, technically). I barely see any posts around the protest. Even the railway closure is not on my feed.
Was not expecting this.
'Ghosts of Departed Proofs' is my all-time favorite typed FP paper. Highly recommend going through it if you haven't.
If gdp-ts is as good as the tweet says, I'm moving most compile-time invariants to this.
Introducing gdp-ts: Ghosts of Departed Proofs for TypeScript.
gdp-ts is a library, linter and AI skill for safer API design. Under this contract, sensitive functions require 'proofs' that the caller performed an authorization check.
The typechecker verifies these proofs at compile time, preventing your team and agents from shipping catastrophic security (and other kinds of) bugs.
While these patterns have existed for quite some time, especially in ecosystems like Haskell, ① human code review and ② cognitive and syntactic overhead made these solutions niche.
The situation is now inverted. Agents are writing more code than we can review, and they *thrive* in tight loops with hard constraints that would frustrate us. We see this with the rise of Rust, borrow checker, code aesthetics debate and all.
The README and examples model a real-world Vercel API product constraint: changing the password on a Project requires a proof of a certain role + a certain entitlement. Thanks to Matt Noonan and Ollie Charles for their research in this space.
github.com/rauchg/gdp-ts
Agents are can now write ~100% of most code. And yet, knowing the implementation details in-depth *is* important.
Apps where devs don't care about code have a "smell". Your users can feel it. I've felt it myself while looking to switch text editors.
[1/n]
This is really simple and clever. Makes you wonder why culling didn't occur to competitive FPS/Mobas in all these years!
injuly retweeted
you know what else Jev helps classify?
people who actually understand ML, and people who don't.
injuly retweeted
The "whiteboard defense:" I should be able to pull you aside at any moment and ask you to explain any customer-facing system you've shipped. You should be able to clearly explain how it works and defend the decisions you made. This is my benchmark for responsible AI usage.
I don't expect line-level familiarity with the code. I don't care if you remember the exact function name or implementation detail. You may not even know it. I don't care.
But if I ask "why did you do X instead of Y?", "what happens if this actor behaves maliciously?", "what data structure did you use here and why?", or "where does this fail?" you should be able to answer confidently.
For PoCs, demos, experiments, whatever: I don't care. Generate 100% of it and understand none of it. Speed over quality every time in those specific scenarios.
But if you're shipping customer-facing work, you can't be shipping things you don't understand at a high level.
GG
DeepSource has been acquired by Harness.
Sanket and I started DeepSource on one bet: every line of code shipped should meet a high bar for quality and security, without a human having to review it.
Over the last eight years, we built DeepSource from the ground up: static analysis across 20+ languages, Autofix in 2020, before automated remediation was a category, code formatting, coverage, SCA and security, and eventually AI-powered code review and remediation.
The idea never changed much. Find problems before code ships, and fix as many of them automatically as we can. With agents producing code faster than humans can review it, there is a lot more of that problem to solve now.
I’m proud that everyone from open-source maintainers to enterprises relied on DeepSource to ship better code. More than anything, I’m proud of the team that built it.
Harness is building the Autonomous SDLC Platform to catch risks in software before it ships, and our vision aligns perfectly with theirs. I'm excited that DeepSource is joining forces with them and continuing our work.
deepsource.com/blog/deepsour…
injuly retweeted
DeepSource has been acquired by Harness.
Sanket and I started DeepSource on one bet: every line of code shipped should meet a high bar for quality and security, without a human having to review it.
Over the last eight years, we built DeepSource from the ground up: static analysis across 20+ languages, Autofix in 2020, before automated remediation was a category, code formatting, coverage, SCA and security, and eventually AI-powered code review and remediation.
The idea never changed much. Find problems before code ships, and fix as many of them automatically as we can. With agents producing code faster than humans can review it, there is a lot more of that problem to solve now.
I’m proud that everyone from open-source maintainers to enterprises relied on DeepSource to ship better code. More than anything, I’m proud of the team that built it.
Harness is building the Autonomous SDLC Platform to catch risks in software before it ships, and our vision aligns perfectly with theirs. I'm excited that DeepSource is joining forces with them and continuing our work.
deepsource.com/blog/deepsour…
shorturl.at/4am48
Feel free to DM, or email [email protected]
Another fun fact: two randomly sampled vectors are almost certainly near orthogonal in high dimensional spaces.
Meaning if you need properties that fall out of orthogonality for two independent vectors, higher d_model helps.
