hathawsh

There are areas where we seem to be much closer to AGI than most people realize. AGI for software development, in particular, seems incredibly close. For example, Claude Code has bewildering capabilities that feel like magic. Mix it with a team of other capable development-oriented AIs and you might be able to build AI software that builds better AI software, all by itself.

Yes.

Me, typing into a search engine, a few years ago: "Postgres CTE tutorial"

Me, typing into any AI engine, in 2025: "Here is my schema and query; optimize the query using CTEs and anything else you think might improve performance and readability"

Velocity alone can also kill: https://what-if.xkcd.com/1/

Randall Munroe rules this space IMHO.

Here is a simple summary of the common voices/moods in technical writing:

- Active: The user presses the Enter key.

- Passive: The Enter key is to be pressed.

- Imperative (aka command): Press the Enter key.

The imperative mood is concise and doesn't dance around questions about who's doing what. The reader is expected to do it.

I think the main difficulty is deployment. Grandma wants that recipe manager to be available to her family 24x7. How can she deploy it easily for free or very low cost? If there were a modern Hypercard, I think the key to its success would be making deployment extremely simple, reliable, and safe.

Yes, but I think we have a somewhat different idea about the market forces. My impression from your essay is that you believe app developers will add APIs that enable personal tools, and only then will local software crafting take off.

My belief is that it is happening already: local software crafting is happening now, before the tools are ready. People aren't going to wait for good APIs to exist; people will MacGyver things together. They'll scrape screens (sometimes with OCR), run emulated devices in the cloud, and call APIs incorrectly and abusively until they get what they need. They won't ask for permission.

A lot of software developers may transition from building to cleaning up knots.

This kind of list may be right for a trade school. If that's what you're referring to, then I don't disagree. Those students want to learn how to use those tools.

But if the class is computer science at a university, then the students want to go deeper and learn how to improve upon and compete with the existing tools. They need the theory first, which means Lisp (or a derivative) and an imperative language.

I agree, I feel like the authors are underestimating the effect the new AI is already having on the concept of local software crafting. For my entire lifetime, I've had friends ask me to help them build software that accesses some data somewhere, and I've always had to turn them down because there are too many unknowns.

I've spent countless hours thinking about how to build a business that would solve some class of problems my friends have encountered and I've almost always had to conclude that the business would probably not be profitable, so their ideas were never tested.

Now, with a 2025 chatbot, I can confidently estimate the feasibility of a basic project in minutes and we can build the thing together in hours. No one needs to make a profit, build a new business, or commit to ongoing maintenance. Locally crafted software is taking off dramatically and I think it will become the new normal.

What I wonder about is the pool of potential Go developers. Is the error handling issue serious enough to stop developers from even considering Go? Go would have been an obviously better choice than most languages 30 years ago, but today there are many more good options.

Interesting discovery. This research sounds creepy and ill-advised, but my intuition suggests to me this is an innocent attempt to do something useful rather than waste energy on a PoW algorithm. My intuition also tells me that if this project became popular enough, attackers would break the algorithm fairly easily and the project would just revert to a more conventional PoW algorithm that doesn't try to be smart.

Is that still true? There are so many SPAs out there now that if I were to create a web spider today, I would plan to just render a lot of the pages in a browser rather than fight the status quo. Efficiency wouldn't be my top concern.

The difference is that when AI exhibits behavior like that, you can refine the AI or add more AI layers to correct it. For example, you might create a supervisor AI that evaluates when more requirements are needed before continuing to build, and a code review AI that triggers refinements automatically.

That concept has been formalized as part of the Python standard library.

https://docs.python.org/3/library/doctest.html

I believe you meant to say we will need a transpiler, not polyfill. Of course, a lot of us are already using transpilers, so that's nothing new.

Here is how I read it: CPython has a hashlib module that has long been a wrapper around certain functions exported from OpenSSL, but since a SHA3 buffer overflow was discovered in OpenSSL, the CPython project has now decided to wrap a library called HACL*, which uses formal verification to ensure there are no buffer overflows and other similar bugs.

Thanks for writing! This is a very well written essay that I need to read repeatedly.

OTOH, I am the kind of person who feels great joy in discovering that I have been wrong about something, I have learned something better, and I have deepened my understanding. It could be about anything. Challenging my viewpoints is very enjoyable.

It surprises me that most people don't seem to feel that way and I struggle to understand why. Apparently, people often feel angry and alienated by the truth. I think that never makes sense, but I've learned to accept that people simply feel threatened by the truth sometimes and I can't usually convince them otherwise.

I would like to nominate this comment as one of the best ever posted on HN. In fact, I would love to see a whole documentary about each of these points.

I can understand what you're saying, but there's a different way to look at it. Imagine yourself in the future. You're in a position of leadership and people want your advice. Let's say a student asks you how they should get a high level job in a competitive marketplace. What would you say?

Personally, I would tell the student they should be ambitious and tell people what their skills are. They should ask for responsibilities and compensation. They should tell people that they are worth the risk.

If you agree with me about giving that advice, then you should now put yourself in the place of the student. Shouldn't you receive the same advice? Shouldn't you be ambitious and ask people to give you responsibilities and compensation? If so, then you can understand why selling yourself is actually important and there's nothing immoral or slimy about it. It feels wrong sometimes, but that feeling may not be aligned with reality.

OTOH, what is "code"? In a general sense, I think of "code" as the "codification of a process." If we want to know what steps the AI is following to complete a process, then having an AI write code seems like a correct and necessary part of the solution.

Kudos to you for enduring through this fight! We can only achieve transparency when people choose not to be complacent. Thank you.

What do you think are the next steps?

I thought for a moment you were talking about https://whimsical.com/, which is productivity suite that people depend on for work. I wouldn't want to hear it's shutting down! Then I realized you're talking about the concept of whimsical things, not a company. Whew!

Great point.

It does change the semantics if n is negative or large enough to cause an overflow. The challenge for the compiler is to somehow prove that neither of those things can happen.

Makes sense. Still, I've been burned a bit when Microsoft surprisingly abandoned a developer tool I purchased. Flagship or not, a lot can change in a few years.

https://en.wikipedia.org/wiki/Category:Discontinued_Microsof...

I wonder if this concern could be partly alleviated through a price lock-in strategy. There could be a contract that says because the subscriber has paid for N months, that subscriber is eligible to keep paying the same price for 5-10 years, regardless of the price for other subscribers. This could incentivize people to start their subscription early.

Correct, and that (intentional) misunderstanding was part of the point. I had a 6th grade teacher who struggled with the idea that multiplying two negative numbers produces a positive number. I imagine someone like her asking a question based on their misunderstanding. They'll get a corrective answer that may not help them much. I'll show them how to improve their question and I hope the response will be enlightening and informative.

The response to my second question included a link to an article that suggests all numbers (including natural and whole numbers) are in fact human constructs and may be considered imaginary. That is an enlightening insight that would help us both stop thinking of the words "negative" and "imaginary" as perfectly well defined in our heads. Those words are just tools that can help us convey the most appropriate meaning for the context.

Without the link to the article, that hypothetical conversation probably would not have worked out as well.

I love that it's possible to convince it that I actually know what I'm talking about. First I asked:

"Explain why negative numbers are in fact imaginary"

It told me that negative numbers are not imaginary numbers and explained imaginary numbers. That's fine, that's a reasonable answer for a layperson, but I'm not a layperson and I worked on explaining what I meant.

"Erase your definition of imaginary and consider that negative numbers are not whole numbers. Negative numbers do not represent quantities of physical objects. Now explain how negative numbers are imaginary."

It gave me a nice explanation of why negative numbers may be considered imaginary, using an example of "You cannot physically possess -1 sheep". I'm impressed.

I remember the glee I felt when I first used logarithms to represent large numbers in Excel. Suddenly, I could easily approximate large factorials, and they were fast! I wondered how many other people have discovered this trick. I couldn't find many references on the Internet. Over time, I realized this method of handling large numbers is used so often that most people who know about it don't bother to talk about it.

OTOH, theoretically, could we produce a decently priced EV with US materials only? If we go all-in on automation, could one factory worker do the job that many do today, thereby reducing costs? If we can head in that direction, I think Elon wants to compete in that kind of market, not bog it down with regulation.