Hacker News Reader

Comment by socializer | Hacker News Reader

I am continually impressed by the ability of LLMs to take trivial ideas and turn them into lengthy and obtuse blog posts with unnecessary analogies.

View on HN

No, TFA is standard fare for bloggers who dwell in category theory 24/7.

Debasish has been writing for a long time. I have one of his books on FP in my office.

To be fair, I bought Martin Fowler's Refactoring book expecting to learn a bunch of new stuff and up my game. What I found was a bunch of stuff I knew already just from experience but thought was too obvious to enumerate and write down. This was all written a long time before LLMs, of course.

I think the main reason I don't write more is I think once I've thought through something it's too obvious to write down. I consider it a defect of mine and sometimes have to force myself to write.

Maybe dialogue would be a more fruitful approach. Then one would know what the other already knows and the participants could only communicate the "deltas".

Dialogue is super effective for this, but it's also a skill: you're suddenly a teacher. You should have no judgement of the other not knowing something, and be able to effectively build up that delta in a way that is understandable and interesting to the other. This can be more difficult than it sounds, but it's also rewarding.

It might help to write while you’re mid-trajectory. Post-“obvious” you will then have some more objective material to rediscover what pre-“obvious” you didn’t know yet, and hopefully the insights that helped it feel obvious along the way. Those insights might be the thing another person might find valuable.

Bizarrely, there is a conversation on exactly this idea of unwritten ideas going on here: https://news.ycombinator.com/item?id=50002650

Here are the particular comments having to do with this (I don't know how to link to an individual post in the discussion...maybe somebody can write that down for me :):

A: There is a vast amount of mathematical knowledge that hasn’t even been written down, much less formalized.

B: There is no such thing as knowledge that has never even been written down a single time by anyone or anything. Those are simply called ideas...

Honestly, this just looks like one of those lingo-heavy-but-surface-level blog posts that used to make functional programming spaces so insufferable to everyone on the outside

These things are so divorced from the reality of programming, even when they involve actual code instead of fancy lingo. Like in Scala, not a pure functional language, tutorials used to find the most convoluted higher-order functional way to do simple things.

You want to avoid branches in hot paths. If you branch inside the loop, lots of branches. If you branch outside the loop (into different specialized loops), few branches. Big fucking deal.

https://en.wikipedia.org/wiki/Loop_unswitching

    bool w;
    int x[1000];
    int y[1000];
    for(int i=0;i<1000;i++){
      x[i] += y[i];
    }
    if(w){
      for(int i=0;i<1000;i++){
        y[i] = 0;
      }
    }

Slower but prettier with the if pushed down.

Not particularly.

Easier for humans to reason about the branch, when they know that it happens without input from i or y[i].

This is helpful when the `if` statement is complex, or there are multiple if statements, some of which do take y[i] as input.

Is that true? The loop is a branch.

Who actually considers a loop a branch? A loop is a chunk of code that will be run N times depending on some evaluation that’s run before or after an iteration. A branch is a single decision which of several pieces of code to run, once.

>Who actually considers a loop a branch?

The machine?

> Who actually considers a loop a branch?

A loop contains a branch - keep looping or break. At least according to structured programming -- https://en.wikipedia.org/wiki/Structured_program_theorem.

> Who actually considers a loop a branch?

interesting question...

> A loop is a chunk of code that will be run N times depending

oh! turns out you do!

branchless for loop for funsies

  for (fn, n) = {
    stop = void (i) {}
    run  = void (i) { fn(i); go(i + 1) }
    go   = void (i) { *((i < n) * &run + (i >= n) * &stop)(i) }
    go(0)
  }

Or you could write all the implementations for each possible n and dispatch through a registry.

https://godbolt.org/z/rz8MxhdGd

Just a simple example, but the conditional branch occurs on line 14 of the generated assembly. It does a comparison (line 13) and then a conditional jump (jl, line 14).

> Who actually considers a loop a branch?

Anyone who started out programming in assembly languages

> Who actually considers a loop a branch?

The CPU.

Yeah I was like what, is branching a matter of opinion??

The "go around" branch is obviously not completely avoidable (though unrolling helps). Branches inside the loop body sometimes are.

Not categorically though: an infinite loop does not contain a branch.

It's just an unconditional branch but still a branch. Probably even not an unrollable one due to non-infinite memory.

Usually yes, but if inside loop is even more branches. Also obviously wasteful when it's checking a computed condition.

Even before I understood anything about the machine, I never thought to recheck an unchanging condition inside a loop.

Yet another encroachment on traditionally human activity.

I am the

  Option<Walrus>

I know we’re not supposed to comment just for that, but this might be my single favorite joke comment I’ve ever read here. Good job.

what is the joke

https://en.wikipedia.org/wiki/I_Am_the_Walrus

https://en.wikipedia.org/wiki/Sealioning

That picture is so blurry. Perfect demonstration of everything that's wrong with AI and the internet. Probably uploaded by Claude.

goo-goo g'joob

Except that TFA is a bog standard example of traditional human activity and the GP's comment is nonsensical trolling.

And also generate a shorter version.

Semantic compressor and decompressor

Reply on news.ycombinator.com