Hacker News Reader

JohnKemenyon Teacher AI use is already out of control and it's not ok

Technology is making humankind lazy. First physically, then mentally.

Learning is hard, it's a struggle. Why learn when you can not learn?

JohnKemenyon Hiroshima (1946)

"Look what you made me do!"

By the way: that same argument is used by Netanyahu and Putin today!

JohnKemenyon PHP 8.5 adds pipe operator

Thank you for your insight.

JohnKemenyon PHP: The Toyota Corolla of programming

You sound very sad that your Corolla isn't good. I would get rid of it, if I were you.

However, 6 issues with a 20 year old car doesn't sound that bad to me.

Edit: By the way, you can easily get a 2023 Corolla for less than 23k euro.

JohnKemenyon The Ski Rental Problem

Suppose you repeat this every day, and every day the bus arrives a random time between "now" and in (let's say) 30 minutes.

There is a strategy that allows you to never be worse than 2x if you knew exactly when the bus arrived: Wait for 10 minutes, and if the bus didn't arrive, walk home.

In all cases when the bus arrives between now and in 10 minutes, you do the optimal thing, and whenever the bus arrives after 10 minutes have passed, you will be home after 20 minutes, which is not worse than 2x worse than optimal.

JohnKemenyon The Ski Rental Problem

It's a computational problem, I thought (mistakenly) that the HN crowd would understand.

JohnKemenyon The Ski Rental Problem

It's a problem from computer science, not Dr Oz.

You want to optimize for when you get home, not for your health or environment.

JohnKemenyon The Ski Rental Problem

Here's a different version of the problem.

It takes 10 minutes to walk home from the bus central. The bus is late but should be here any minute now. The bus takes one minute. Do you wait or walk?

JohnKemenyon At 17, Hannah Cairo solved a major math mystery

> She could just use her publication as a dissertation and be done with it!

The purpose of a PhD is not writing a dissertation. It is a research school, and I'm sure she could still learn a thing or two about research (and teaching).

JohnKemenyon A Python dict that can report which keys you did not use

Do you seriously have difficulties explaining when to use a class and when to use a dictionary?!

JohnKemenyon Ambigrammia: Between Creation and Discovery

I think he straight up doesn't care. He knows that intelligence won't come from a Markov chain. He's always been working on more fundamental stuff.

JohnKemenyon Revisiting Moneyball

https://www.vox.com/videos/2017/6/5/15740632/luck-skill-spor...

Why it’s so much harder to predict winners in hockey than basketball

JohnKemenyon Bus Bunching

Always upvote Futility Closet!

JohnKemenyon Quantitative AI progress needs accurate and transparent evaluation

Don't put Google and ChatGPT in the same category here. Google cooperated with the organizers, at least.

JohnKemenyon Python classes aren’t always the best solution

namedtuple is not a replacement for a class, but for a tuple. It is a tuple with named accessors.

You can also argue that you don't need a Pair type since you can simple make a class for the two things you want to contain.

JohnKemenyon Compression culture is making you stupid and uninteresting

If you want to learn something new, don't bring your phone up the toilet. Sit with your own thoughts.

(Or: shit with your own thoughts, as my auto-correct wanted it.)

JohnKemenyon OpenAI claims gold-medal performance at IMO 2025

We have been doing automated theorem proving since 1954.

JohnKemenyon Nobody knows how to build with AI yet

Tesla drivers are involved in more accidents than drivers of any other brand.

JohnKemenyon “I noticed a clear violation of our contributing guidelines”

commit number n: add tests

JohnKemenyon What the Fuck Python

If you use a floating point number as a key in a `dict` the wtf should not be pointing at Python.

These are not wtfs, please.

These examples read like someone who is learning to program and who is confused.

    x is not y
    x is (not y)

How can you confuse these two?!

JohnKemenyon Evaluating publicly available LLMs on IMO 2025

The last time someone claimed a medal in an olympiad like this, turned out they manually translated the problem into Lean and then ran a brute force search algorithm to find a proof. For 60 hours. On a supercomputer.

Meanwhile high schoolers get a piece of paper and 4.5 hours.

JohnKemenyon I'm switching to Python and actually liking it

    exit("Missing ...")

This prints the message and exits with code 1.

JohnKemenyon To be a better programmer, write little proofs in your head

I once needed to write binary search for a microcontroller in C (no libraries). The routine ran about every hour, with appx 4M data points.

JohnKemenyon To be a better programmer, write little proofs in your head

That is stretching the definitions of proofs.

    assert random() != 0.

QED

JohnKemenyon To be a better programmer, write little proofs in your head

You’ve missed the most important point of TDD—and indeed, of tests.

Tests do not ensure that your functions are correct; they ensure that you are alerted when their behavior changes.

Second, TDD is a way to force dogfooding: having to use the functions you’re going to write, before you write them, helps you design good interfaces.

JohnKemenyon Reading Neuromancer for the first time in 2025

Turn in your geek card on your way out.

JohnKemenyon I used to prefer permissive licenses and now favor copyleft

Crypto bros rediscovering RMS 50 years later.

JohnKemenyon I don't think AGI is right around the corner

I'm a scientist with many published works, albeit in the theory of computation. I have many colleagues publishing in top AI/ML conferences and journals.

Let me tell you: nobody listens to us.

JohnKemenyon I don't think AGI is right around the corner

You're not making the point you think you're making.

JohnKemenyon I don't think AGI is right around the corner

Whose opinion would you want to hear? The CEO of an AI company?