
I ❤️ @amandaharlin, Ada & Dorian And: Programming, Composing New Music, Synths Owner/Engineer @simiancraft Formerly/Cofounded: @techlahoma @okcjs @thunderplains
Norman, OK
Joined January 2011
- Tweets14.2K
- Following1.2K
- Followers1.5K
- Likes11.6K
Pinned Tweet
After 10+ years, finally gave jesseharlin.net an update. built a celestial music machine with Fable and Astra, both working from my own code conventions and a decade of my talks and patterns. ( Sound ON 🔊)
the_simian retweeted
The time has come
You've never needed experienced devs more, and if they aren't reading the code, I'd fire um on sight
I have never seen so many dumbass bullshit mistakes disguised as solutions that pass all the AI checks
Yeah, they pass the checks. No, they shouldn't
Replying to @vyrotek
it turned a 3 file PR I had published into a 32 file PR today for me right after lunch and *everything* it did was literally retarded
the_simian retweeted
The loudest mouthed, most ai-pilled voices on here constantly yapping about “code review is obsolete just have an extra agent check the first agent’s work” etc would be slightly less obnoxious if not for
* waves hands *
This constant enshittification
the_simian retweeted
Every time I've looked at the code, the LLMs have snuck something really, really nefarious and awful.
Seriously, every single time
It's been over a year now. And I still have to read the code.
the_simian retweeted
They certainly don’t produce better architecture or more coherent code that the average engineer I’ve interacted with
I’ll stick with calling it vibe coding ty
it’s clear we need to retire the term vibe coding because the models now produce better architecture & way more coherent code than an average software engineer.
perhaps even better than top software engineers too because unlike a human, ai has a much much larger context window.
this ain’t vibe coding at all. it’s precision engineering.
the_simian retweeted
I’m speaking at ThunderPlains on October 21, and my talk is called Confessions of an AI Vampire: Building Software When the Agents Never Sleep.
AI was supposed to help us work less, but I’ve found it can make it harder to stop. There’s always one more thing I could get done, and the agents don’t need to go to bed. I do, but that doesn’t always win the argument. That’s where the AI vampire part comes in.
I’ll walk through how I went from asking AI for help with a function to managing work across agents, the processes I use, and how I’m thinking about working this way. Whether you’re still writing most of your code by hand or already juggling agent threads, there should be something you can take back and try.
October 21 · 1:30–2:30 p.m. Central
Oklahoma City Convention Center · Track AB
2026.thunderplainsconf.com/
Note: No Carries were harmed in the making of this photo
AI is making a lot of engineers, even famous and good ones, literally and actually incompetent right now. Its simultaneously a quantum leap in technology, and a thing you should use. But "the AI said" should be met with skepticism and rigor. Read the code, and be still be curious
Yesterday, someone posted a task runner claimed to be 39x faster than Turborepo.
I got really excited, thinking they found some edge case in Turborepo and that I could make something way faster.
I open the benchmark repo, and its just straight up, flat out incorrect. Like, not even sorta useful. Not only was it not measuring the task runner parts of the script, it was overfit to a highly specific, completely unrealistic workload. 💀 The entire project is obviously Opus 5.5 vibe coded and Opus cheated the goal. Easy to miss if you're not experienced in build systems. Forgivable!
So, I open a PR to show that the algorithm being used can also make scheduling slower, by any arbitrary factor, with only a tiny adjustment. I chose 100x slower than Turborepo for demonstrative purposes.
The author fixed some of the problems but there are still grave issues. They're still comparing on an overfit bench that doesn't hold up when you insert the slightest variable, and *insisting* that its right. It is very obviously incorrect, but oh, well.
After revision, Turborepo is only 1.36x slower. Down from 39x, but a more realistic number nonetheless. So I figure: "Great! There must be something I can learn to make Turborepo faster."
I go digging, and it turns out the other tool is only faster still because it has correctness bugs. Once they correct those, they'll be roughly on par with Turborepo.
On top of that, the numbers they post now report with a mix of "% faster" and "✗ faster", using "% faster" whenever the difference isn't big enough to make their tool look good. 1.36x gets turned into "36%", which looks like more than it is to a casual observer.
This is a story of how I thought I was going to speed up Turborepo by 39x, and got nothing out of it.
I'm not mad; I'm just disappointed. 😄
the_simian retweeted
Korg DS-10 for Nintendo DS running as an unmodified ROM inside a html file inside jweb~ inside a max for live device inside of Ableton. By writing to specific points in RAM, we can play notes via MIDI, modulate and control parameters, trigger drums etc. github.com/little-scale/ds10…
"you refactor ideas before you refactor code" is something I've taught every person I've ever mentored or trained for the last 15 years. Along with "there are no good ideas for paintings, only paintings".
Nothings changed much in these ways.
This is because you refactor ideas before you refactor code, and this has always been the actual job.
1. Oh shit, AI can do my job better than me
2. Software engineering is dead!
3. I can vibe code as much slop as I want!
4. Oh, things are not going well. Fable/Sol/Opus all suuuck
5. But what if I constrain the agent? Put it in a straitjacket?
6. Oh, the results are pretty good now. How do I go deeper?
7. *Orders 12 software engineering books from Amazon*
8. Software engineering is alive!
The folks that were saying humans don't even need to read the code are backpedaling. Everyone sane has been saying human-in-the-loop with AI assistance is most effective right now, to believe human alone can do more is delusional and equally so that you can remove humans entirely
There's certain categories of work that AI can do insanely good things with. It should be sprinkled throughout your workflow, and you should use judgement in when and how you use it. You're still responsible for the output; you should assume this; take pride in understanding it
the_simian retweeted
I’ll be blogging about this very thing soon but unless you’re a rich millionaire with no boss nothing in life has changed.
Life still has consequences.
Bad code can lose your job.
Bad code can lose your customers.
How we make code has changed but the consequences have NOT
the_simian retweeted
I made another VST plugin. And this one is also free. It’s an 808 bass plugin called Big Booty Bass. Enjoy:
github.com/kenwheeler/big-bo…
the_simian retweeted
I am not touching the operating system built from someone who cares so little and is proving to be the worst programmer I’ve ever seen.
Most 'vintage' VST enjoyers know about Delay Llama; pretty infamous for being a Very Strange VST™. That said, I used to just hunt the internet for the weirdest VSTs I could find: this one was one of the weirder ones I remember (warning sounds terrible):
youtu.be/NYXVNYOZV2M?t=22


















