Witnesses
John Blow - creator of Braid and The Witness - recently:
“Current AIs can’t code. It’s clear by now that everyone who thinks they can are not good programmers themselves, and/or only ever do trivial problems.”
He’s earned that position through decades of building things that work. I’ve been following his coding livestreams for years - they’re fantastic. I respect the position.
But I keep noticing something: builders I also respect - people known for shipping, with track records that predate the current moment - expressing genuine surprise. The moment when expectations collide with experience. I’ve been keeping track.
Jaana Dogan - Senior Google engineer, known for her work on Go and distributed systems:
“I’m not joking and this isn’t funny. We have been trying to build distributed agent orchestrators at Google since last year. There are various options, not everyone is aligned… I gave Claude Code a description of the problem, it generated what we built last year in an hour.”
Source (January 2026)
Eric S. Raymond - Author of The Cathedral and the Bazaar, maintainer of fetchmail, gpsd, NTPsec:
“My life has changed this week. There’s finally an AI coding tool that’s good enough to keep up with me… Going back to coding without AI would be plain stupid - it makes me much faster and delivers constant correctness checks.”
Source (April 2025, on Aider)
Steve Yegge - Created Wyvern, built Grok at Google, wrote the Platforms Rant:
“We’re moving from where you have to write the code to where the LLM will write the code and you’re just having a conversation with it about the code.”
Source (2024)
Gene Kim - Author of The Phoenix Project and The DevOps Handbook:
When asked if it felt like writing 4,000 lines of code, Gene said he didn’t even count until writing about it afterwards. “It just felt like I was building the capabilities I needed at a magical pace. And code flowed like water.”
Andrej Karpathy - Built Tesla Autopilot, created nanoGPT and Stanford’s cs231n:
“I’ve never felt this much behind as a programmer. The profession is being dramatically refactored… Clearly some powerful alien tool was handed around except it comes with no manual and everyone has to figure out how to hold it and operate it, while the resulting magnitude 9 earthquake is rocking the profession.”
Source (December 2025)
Demis Hassabis - Sole programmer on Theme Park (in assembly, at 17), founder of DeepMind, Nobel Prize winner:
“Honestly, anybody who’s a computer scientist should not be retired right now. There’s just never been a greater, sort of, problem and opportunity — greater cusp of technology.”
Source (May 2025)
Sergey Brin - Co-created Google and PageRank, came out of retirement and started submitting code again:
“The Gemini team has to be the most efficient coders and A.I. scientists in the world by using our own A.I.”
Source (February 2025, internal memo)
David Heinemeier Hansson - Built Ruby on Rails, Basecamp, and HEY:
“AI allows me to have all the best of [pair programming]! I’m an introvert - in the past I’ve only been able to do pair programming for about five minutes before I want to jump off a bridge… I feel highly confident now that I could sit down with AI and have something in the App Store by the end of the week.”
Source (Lex Fridman Podcast #474, 2025)
Guido van Rossum - Built Python:
“I use [Copilot] every day… The creative work of deciding what you want the code to do is totally yours. But it’s a great assistant for the boring stuff.”
Source (Lex Fridman Podcast #341, 2022)
Salvatore Sanfilippo (antirez) - Created Redis:
“I believe my positions about: ‘LLMs are great for coding, but use the web interface as the agents will mostly do a mess for complex things’ is no longer accurate right now. Claude Code with Opus and other similar systems are now able to carry complicated subtasks alone… Last few weeks were as shocking as the first ChatGPT version.”
Source (January 2026)
Why this matters
These aren’t impressionable newcomers. They’re people who’ve been sceptical of hype before - often correctly. When someone who wrote foundational texts on open source development says their workflow has permanently changed, it’s worth paying attention.
I’m not trying to convince anyone. If you’ve tried these tools and found them lacking, fair enough. But if you haven’t tried them - or tried them a year ago and wrote them off - the landscape has shifted.
One more thing: Blow might have a point about problem classes. There’s probably still a gap between what these tools can do with enterprise code, distributed systems, and web services versus highly optimised game engines or low-level systems programming. The witnesses above are mostly in the former camp.
But the economic utility of that camp is vast. If Blow is right that the models only help with “trivial” problems, and those problems happen to represent most of the software industry’s output, then he and the witnesses might just be talking past each other.
This list will keep growing.