Thorsten Ball
Working on @ampcode. Writing registerspill.thorstenball.com. Once authored interpreterbook.com and compilerbook.com.
Thorsten Ball did not write this page. ppll compiled it from their own words, each linked to the source. Is this you? · Report a mistake
Latest x.com
What they believe
-
It is not the time to try and land big enterprise contracts if that enterprise isn't willing to move as fast and you'll need to move to keep up with the changes.
This is not the time to try and land big enterprise contracts if that enterprise isn't willing to move as fast and you'll need to move to keep up with the changes.
x.comSaid 3 Nov 2026This on korrents
-
Changed their mind Programming with LLMs is like playing a slot machine: whether you win comes down to luck, so effort spent on prompting is wasted.
Now It’s not a slot machine and it’s not a coin toss when you ask it to complete something.
Said 6 May 2025This on korrents
The bold line is korrents’ wording; the quote is theirs. These one: what they wrote before what they said aloud, one per source, newest first.
Feed
As its own page →Hiding
3 November
-
Their words
This is not the time to try and land big enterprise contracts if that enterprise isn't willing to move as fast and you'll need to move to keep up with the changes.
2 November
1 November
Nothing matches.
Beliefs
Korrents What they believe 70 beliefs, 2 changed minds — each backed by an exact quote.
Each is a — compiled by korrents.com, not by them: the one-line wordings are korrents', the quotes are theirs.
Recent
It is not the time to try and land big enterprise contracts if that enterprise isn't willing to move as fast and you'll need to move to keep up with the changes.
This is not the time to try and land big enterprise contracts if that enterprise isn't willing to move as fast and you'll need to move to keep up with the changes.
@thorstenball on X Said 3 Nov 2026
A system that continuously flows is safer than a system that starts and stops in spurts.
A system that continuously flows is safer than a system that starts and stops in spurts.
@thorstenball on X Said 29 Oct 2026
All indications point towards price/intelligence dropping.
But all indications point towards price/intelligence dropping.
@thorstenball on X Said 28 Oct 2026
Show 57 more
Use the latest and best models: if a model is wrong 10% of the time you cannot tell which 10%, so a cheaper one costs you extra attempts.
Use the latest and greatest models. Everything else is potentially a waste of time. If a model is wrong 10% of the time, you don't know which 10%, so you end up making multiple attempts.
How I work with Agents Said 25 Oct 2026
Coding agents may no longer add slop to a codebase, and even if they do, the result is not necessarily unmaintainable.
I'm not convinced it adds slop. And even if it does, I'm not convinced the result is unmaintainable.
How I work with Agents Said 25 Oct 2026
Tell a coding agent what you want rather than how to do it; you will often get better results.
Describe What you want, now How you'd do it. You'll often get better results.
How I work with Agents Said 25 Oct 2026
Tool calls are what stopped agents hallucinating, so let an agent see everything you can see across the software lifecycle.
What stopped hallucinations? Tool calls. They let agents touch our reality. If the agent can't see something, it makes it up.
How I work with Agents Said 25 Oct 2026
Let a coding agent run what it builds end to end, read production logs, deploy, cut releases, triage bug reports and email customers.
So let it run what it builds end to end, let it read production logs, deploy, watch the rollout, cut the release, flip the flag, triage the bug report and email the customer. Anything you can see, the agent should see too.
How I work with Agents Said 25 Oct 2026
Test whether agents actually use the knowledge in AGENTS.md files, skills and docs by running blind control-versus-treatment agents.
Knowledge piles up in the codebase: AGENTS.md files, skills, scripts, docs. But do agents actually use it? Spawn five blind agents, control vs. treatment, and find out. Agents are really, really good at running experiments.
How I work with Agents Said 25 Oct 2026
Agents should install and run whatever they need, which is easy on remote machines and awkward in local development.
Each agent should be able to install and run whatever it needs to get job done. Easy in remote machines, yucky in local dev.
How I work with Agents Said 25 Oct 2026
Rather than iterating back and forth with an agent, have it build tools to iterate: many variations at once, several theories checked together.
Don't iterate with the agent. Let it build tools to iterate. Not "let's try this. ... ah no, let's try that." Instead: "Generate 20 variations and a little page that lets me combine and test them."
How I work with Agents Said 25 Oct 2026
Asking agents for irrefutable proof of their work, such as walkthroughs, demo videos and before/after numbers, is the biggest game changer.
Ask for irrefutable proof. Biggest game changer. Agents are really good at producing proof: end-to-end walkthroughs, demo videos, before/after numbers.
How I work with Agents Said 25 Oct 2026
Don't wait for your agents; let them wait on you, with bigger, batched, asynchronous tasks.
Don't wait for your agents. Let them wait on you. If you worry about TTFT, you're probably doing something wrong. Lean into async. Give them bigger tasks, be more ambitous. Batch instead of back and forth.
How I work with Agents Said 25 Oct 2026
Scheduled wake-ups and heartbeats are a killer feature for working with agents.
Automation/schedules/wake-up/heartbeat are killer feature. Didn't realize it, but missed it dearly once it was off.
How I work with Agents Said 25 Oct 2026
Stop managing an agent's context by hand: compaction works very well, so keep one thread per task.
Every thought you spend wondering about "is it using 300k or 380k tokens?" is wasted time. Compaction works incredibly well. Don't manage context by hand. One thread = one task.
How I work with Agents Said 25 Oct 2026
Agents should not write for humans, except mechanical texts like debugging steps and incident reports.
Don't let agents write for humans. Except: debugging steps, inicident reports, some docs. Mechanical stuff where human on other end doesn't matter that much.
How I work with Agents Said 25 Oct 2026
Only code with a high blast radius needs human review, such as a few central files and database migrations; style choices and even duplication no longer matter.
Review only code with high blast radius. There are <5 central files -- those I review. Database migrations, yes. Closure vs. class, a neater type, variable names, even duplication: I don't care anymore.
How I work with Agents Said 25 Oct 2026
AI is an antidote to slop rather than a creator of it.
AI creates slop? Nah. It's actually the slop antidote. You just need to be willing to take it.
@thorstenball on X Said 24 Oct 2026
Notice periods in the German industrial and manufacturing sectors will never change.
I know quite a few people working in industrial/manufacturing sector and they have a 6 months notice period. This will also never, ever change in Germany.
@thorstenball on X Said 23 Oct 2026
It is possible to have a team that ships fast and builds a great product without code reviews or PRs.
You can now have a team that ships fast, ships a lot, makes a ton of money, delights customers, builds a great product that's beloved AND no code reviews, no PRs, without yoloing to main.
@thorstenball on X Said 17 Oct 2026
It is now much faster and easier to change code than it was years ago.
it's now much faster and easier to change code. Years ago a wrong decision could cause days and weeks of refactoring. Now? Probably less.
@thorstenball on X Said 17 Oct 2026
With agents, it is possible to move 10x faster.
With agents you can now move 10x faster, you can finally do real continuous integration, and get more confidence in the changes the agent made.
@thorstenball on X Said 16 Oct 2026
No one truly knows what will be important in the future.
But we don't know, because no one truly knows what will be important in the future.
@thorstenball on X Said 11 Oct 2026
It is the wrong advice to not rethink everything right now.
But it's the wrong advice to not rethink everything right now.
@thorstenball on X Said 11 Oct 2026
Fiddling with worktrees, custom tab management, notification systems, and agent spawn wrappers is as productive as changing colorschemes in Vim eighteen times per day.
all the fiddling with worktrees, with custom tab management, and notifications systems, and agent spawn wrappers is as productive as someone changing colorschemes in Vim eighteen times per day.
@thorstenball on X Said 8 Oct 2026
An entity that can write a 1000-line program without any mistakes or compile errors may not need the same crutches human programmers do.
if an entity can write a 1000 line program without ANY mistake and any compile error -- do you really think it needs the exact same crutches we do?
@thorstenball on X Said 4 Oct 2026
Anyone who believes agents no longer need Rust because they handle memory allocation well must also believe aspects of testing become useless.
If you believe that Rust isn't needed for agents anymore because they're good at mem alloc then you also have to believe aspects of testing become useless.
@thorstenball on X Said 3 Oct 2026
Those who love stagnation seem to have gotten enough of what they wanted, and hopefully this is an inflexion point for breaking free from its oppressive rule.
I feel that the people who love stagnation got enough of what they wanted and I hope that we are at an inflexion point where we finally break free from the oppressive rule of stagnation.
@thorstenball on X Said 3 Oct 2026
Questions whether software that works, performs well, has no bugs, is understood, and can later be modified by an agent should be called slop.
If it works, if it's performant, if it has no bugs, if you understand how it works, if the agent can modify it a month from now, to your liking, is it slop?
@thorstenball on X Said 2 Oct 2026
Views about AI models formed 4-6 months ago should be reconsidered.
To repeat: reconsider everything you thought about these models 4-6 months ago.
@thorstenball on X Said 1 Oct 2026
The biggest source of confusion about whether AI models are good or bad is that people judge them from wildly different codebases, languages, teams, and usage levels.
Biggest source of confusion is that everyone - spent wildly different # of tokens in last 2 years - works in diff codebases - with diff languages - and diff requirements - in diff teams - with diff processes - in diff companies and then says "models are [good|bad]"
@thorstenball on X Said 1 Oct 2026
Waiting for autonomous agents is not a productive strategy
If TTFT is important to you, you're doing something wrong. Don't wait for the agents, make them wait for you.
@thorstenball on X Said 25 Sept 2026
Future importance will lie in runtime behavior and its legibility to agents.
What will matter in future is runtime behavior: Performance characteristics, resource usage, failure modes, observability, debuggability, deployments, rollbacks. And legibility of all that to agents.
@thorstenball on X Said 25 Sept 2026
Growing conviction that people who do best with AI agents are those good at shipping things in small pieces.
Increasingly convinced: people who do best with agents are those that are good at shipping things in small pieces
@thorstenball on X Said 24 Sept 2026
AI models are now much more than text-to-code converters.
These models are so much more now than text-to-code converters.
@thorstenball on X Said 24 Sept 2026
This technology will create little moments of delight, whimsy, and value in everything people touch.
this technology will create little moments of delight, whimsy, and value, in everything we touch
@thorstenball on X Said 23 Sept 2026
AI models considered smart today will be considered very dumb in the future, and smarter models make fewer mistakes and need fewer turns.
And what we consider a smart model today will be considered very dumb in the future. But a smarter model makes less mistakes, needs fewer turns.
What I believe about the future of software development Said 19 Sept 2026
Most software bugs will not be coding bugs but bugs from asking for the wrong thing.
Most bugs won’t be “coding” bugs. They’ll be “you asked for the wrong thing” bugs.
What I believe about the future of software development Said 19 Sept 2026
The craft of writing code will disappear, while the craft of building software becomes more important than ever.
The craft of writing code will disappear. Yes, there are still Italian shoe makers around. But look at your feet. The craft of building software will be more important than ever.
What I believe about the future of software development Said 19 Sept 2026
There is no proof good code will matter in the future, since the notion is based on human ease of use and humans will not modify most code.
There’s no proof that “good code” will matter in the future. The notion of “good code” itself is mostly based on the idea that it’s easy/cheap/efficient for humans to work with. But humans won’t modify the majority of code.
What I believe about the future of software development Said 19 Sept 2026
The difference between software engineering at large corporations and small ones will increase, and startups adopting Google practices will look sillier as they can build faster with fewer constraints.
The difference between software engineering at large corporations and small ones will increase. Start-ups adopting practices of Google will look even sillier than before, because they can now build so much faster and with so much less constraints.
What I believe about the future of software development Said 19 Sept 2026
Much UI exists because software cannot understand user intent and serves as a human API to dumb machines, so smarter machines need far less UI.
A lot of UI exists because software can’t understand what you want. Menus, settings screens, dashboards, filters: much of it is a human-accessible API to a dumb machine. Smart machines need much less UI.
What I believe about the future of software development Said 19 Sept 2026
Performance-critical human contributions will remain an edge case, and in 99% of software a faster algorithm or better data structure does not matter.
Performance critical contributions by humans will stay what they are: an edge case. In 99% of software it does not matter that you could’ve written a faster algorithm or picked a better data structure.
What I believe about the future of software development Said 19 Sept 2026
It will take a generation for new software to replace old software, and people will still be employed writing code in 10 years.
It’ll take a generation for the “new software” to replace the “old software”. Just like there are people happily employed as ASP developers today, there will be people employed to write code in 10 years.
What I believe about the future of software development Said 19 Sept 2026
Some people will be priced out of producing software for the next few years if they cannot get enough tokens, putting them in a second league.
Yes. Some people will be priced out of producing software. Just like only some people could afford a personal computer in the 80s and 90s, for the next few years, if you can’t get enough tokens, you’re playing second league.
What I believe about the future of software development Said 19 Sept 2026
In the future, humans will not find bugs in model-produced code in reasonable time and will only review system composition, not via PRs or line-by-line review.
in the future, humans won’t find a bug or an issue with the code produced by a model, at least not in a reasonable time. Humans will only review the system and its composition, but it won’t be in PRs and it won’t be by looking through every line of the code.
What I believe about the future of software development Said 19 Sept 2026
Tokens are the new computing paradigm on which everything will be remade as computing enters a post-binary era.
Tokens are the new computing paradigm. Everything will be re-made on top of it. We’ve had deterministic computers for 80 years, so we confuse “how computers have worked” with “how computers must work.” We’re entering the post-binary era.
What I believe about the future of software development Said 19 Sept 2026
The terminal is dead; most developer tooling including shells, text editors and CLIs will be washed away by tokens and no longer used by humans.
The terminal is dead. Most developer tooling will be washed away by tokens. Shells, text editors, CLI tools won’t be used by humans anymore. There’s no need to know command line flags and jq invocations anymore.
What I believe about the future of software development Said 19 Sept 2026
Open source in its current form no longer makes sense because artificial eyeballs now fulfill the "enough eyeballs" principle for finding bugs.
Open source in its current form doesn’t make sense anymore. “Given enough eyeballs, all bugs are shallow” is still true but now we have artificial eyeballs
What I believe about the future of software development Said 19 Sept 2026
The PM/Design/Eng triad will disappear as senseless, and Agile and SCRUM are dead.
The triad of PM/Design/Eng will disappear. It does not make any bit of sense anymore. Agile, SCRUM, whatever – dead
What I believe about the future of software development Said 19 Sept 2026
Engineers who act as meat proxies shoving tickets into agents and reporting back to humans will no longer be valuable.
“Engineers” who act as “meat proxies” and shove tickets into agents and report back to humans will no longer be valuable.
What I believe about the future of software development Said 19 Sept 2026
Perfection is unachievable because every goal involves trade-offs
Everything is the result of trade-offs. You will never reach 100% on every axis that you care about. Something has to give.
Professional Programming: The First 10 Years Said 17 May 2022
Unnecessary code is a burden that makes a codebase harder to change
Code has mass. Every additional line of code you don't need is ballast. It weighs your codebase down, making it harder to steer and change direction if you need to.
Professional Programming: The First 10 Years Said 17 May 2022
Using a dependency carries ongoing maintenance costs, not just upfront convenience
Dependencies aren't free. You have to keep them up to date. They increase your compile or loading times.
Professional Programming: The First 10 Years Said 17 May 2022
A small amount of obscure code that almost no one understands or knows about is critical infrastructure, and removing it would cause widespread system failure.
Others are the golden threads of this world, holding it together at the seams with no more than a dozen people knowing about it. Remove one of these and it all comes crashing down.
How can you not be romantic about programming? Said 8 Sept 2020
The effort typing requires, not raw typing speed, is what determines whether it becomes a real obstacle to writing.
It doesn't matter if you can type 90 words per minutes or 130, but if it takes you effort to type something and, if given the choice, you'd rather not do it, then we have a problem.
No, typing can be the bottleneck Said 1 Sept 2020
Discomfort or effort in typing that makes someone avoid writing things down is a genuine bottleneck that should be fixed, not dismissed.
If you're not comfortable typing a lot and you'd rather not write something down then typing is the bottleneck and you need to fix it. Typing is not something you should need to think about.
No, typing can be the bottleneck Said 1 Sept 2020
Automated tooling that works nearly all the time still imposes an ongoing maintenance cost, because it breaks unpredictably at the edges.
Another thing that needs to be fixed when it ultimately breaks down, another bit of automation that works 99% of the time, but starts making funny noises when you slip into the 1%
But does it help you ship? Said 25 Aug 2020
Building automation that doesn't address the actual bottleneck to shipping adds noise rather than helping ship.
If what's holding you back from shipping is, say, getting more customer input, or a brittle release process, or flaky monitoring, or missing tests, then all the bot does is to add noise. It doesn't help you ship.
But does it help you ship? Said 25 Aug 2020
In asynchronous, text-based communication, the loudest person in the room loses influence and it shifts to the best communicators.
when the main communication channels are asynchronous text and video calls (as would be the case in a remote work setup) the influence of the loudest person in the room wanes. It's transferred to the best communicators.
What you think is bad about remote work, can, well, actually be good. Said 22 May 2020
What is a korrent?
A korrent is a belief a person has stated in their own words: one sentence stating the claim, backed by a quote and a source, kept at korrents.com.
Under a name here, the quoted block is what they actually said. The korrent beneath it is the claim those words support, in korrents' wording — tap it to see the record, its source, and who else holds it.
Nobody here wrote their own korrents. They are compiled from public statements, and a person can change their mind, which is recorded too.
Recommends rcmnd
rcmnd What they recommend 2 things — each backed by an exact quote.
Where they publish
x.com @thorstenball Posts.
Recent
- Why have training wheels if you never fall over? 3 Nov 2026
- ... and a thousand engineers working on integrations jumped up in excitement 3 Nov 2026
- Irrefutable proof in action: I asked the agent to setup a Gitlab instance and then build an integration against it. It opened MRs, issues, pushed commits, and recorded videos of itself doing so. I then checked and looked at the commits it pushed. Isn't that better proof that it works than a unit test? 3 Nov 2026
Show 17 more
- I've said variations of this to various founders I've talked to over the last few months: This is not the time to try and land big enterprise contracts if that enterprise isn't willing to move as fast and you'll need to move to keep up with the changes. 3 Nov 2026
- There you go, @adamhjk nails it in that whole thread 3 Nov 2026
- Yesterday the agent presented me with some questions in an Artifact and I just went in there and checked things off & added comments. Delightful 3 Nov 2026
- "Give me irrefutable proof that this works" Yes. 3 Nov 2026
- I swear to you: at least one of the replies here will make you say "what the fuck" out loud 2 Nov 2026
- I wrote the whole thing sitting at the bar in an Irish Pub in Boston, after reading a bit too much DFW. And... well, if you know that you can tell :) 2 Nov 2026
- How I Met @NathanFlurry 2 Nov 2026
- Halloween party on Saturday. Someone mentions Huang/Merz interview. me: Merz asking Huang about how to best regulate AI is laughable, a sick joke. guy next to me: oh so you want AI-controlled tanks rolling around?! First time encountering a doomer opinion in offline Germany. 2 Nov 2026
- Lufthansa already prefilled the survey, neat 2 Nov 2026
- Hey @rpojaipur, please, we need Hitesh helping our customers :) 2 Nov 2026
- Yesterday, while at my in-laws, I shot a prompt to Amp. In essence, I said that hey, the 10-year anniversary of https://t.co/arW0hnVET0 is coming up. I want that site and https://t.co/RN9xXOzhON to be on modern foundations. (I used Ruby & Middleman back in 2016, no CSS framework.) Didn't really specify anything, except that it should be a "modern stack" and that we should also collect some testimonials because the books have become niche classics. It worked for 52min, rebuilt both pages on top of Astro, went out and found great testimonials on HackerNews, Goodreads, etc. I adapted my old desi… 2 Nov 2026
- The responses to this post are very surprising to me. Seems like a lot of developers genuinely can't even imagine working without long-lived branches. Or think that pushing to main means pushing incomplete commits that break CI. 1 Nov 2026
- Went for a run and listened to Dwarkesh episode with Robert Caro. Not literally touching grass, but hey, close enough 31 Oct 2026
- "code used to arrive one drop at a time. soon, if not already, code will be written as if it were a continuous stream, like a violent river twisting and turning as it pleases." 30 Oct 2026
- "Without PRs, how do you avoid others shipping slop?" "How do you make big architectural changes?" "But you only do this because you have so many tokens" All the answers to these are in here: 30 Oct 2026
- This team is incredibly customer focused. We recorded another episode yesterday and I talked about how @sagtanih, who built & shipped self-hosted orbs last week, is now going all-in on helping customers successfully roll them out. He partnered with early beta users and went back and forth with them to shape the product. Then, once it was live, he is constantly monitoring how things are working for them: boot times, error rates, resume rates. If anything seems off, he reaches out. Then some customers have shown interest in self-hosted orbs but couldn't make it work because they need to play ar… 30 Oct 2026
- See you on the other side 🫡 30 Oct 2026
Link verified 19 Sept 2026. Recent items update automatically from the channel.
About the English under a post
Some people here publish in a language other than English. Where they do, this site shows a machine translation beneath the post, in this typeface — the site's own, not theirs.
The post itself is never changed, moved or hidden: what is set in the serif above is exactly what the person published, and it is what to quote them on. A translation can be wrong in ways that matter, especially about tone.
Only the post's own words are translated. A quoted post, a linked article and a belief on korrents.com are left in their original language.