Understanding is the new bottleneck

(geoffreylitt.com)

40 points | by sebg 1 hour ago

10 comments

  • w10-1 20 minutes ago
    I agree with the problem but not the solutions.

    The problem pre-dates LLM's: writing code that "works" but breaks the underlying model. Because it works, it always sounds reasonable and doesn't raise any flags.

    Only someone - human or LLM - who holds the model as the standard would see that this working solution breaks the model.

    (In theory, the model is to preserve scaling, flexibility or some other systemic feature not immediately invalidated by this working code, but as always the model itself could be bad.)

    LLM's are not bad at giving an account of the model; indeed, fighting with the LLM over what the model is can clarify things. But LLM's will happily hold on to a stream of inconsistent statements as their model, so they are not the authority.

    • cyanydeez 13 minutes ago
      understanding a different modality of model interaction gave me proper insight into the specific problem. In visual models, even if the model understands the concept of face, or hand, or whatever, it doesn't know how to de-dupe a statement like "count the number of faces" until you give it a countable reference frame, so it can internally, place a box around a face and give that a coordinate, and then it can collect all the coordinates, and suddenly it's counting face in a picture.

      The same thing happens in code. Things we're happily shifting from context to context, the model itself isn't doing. When it reads file1 for the main() clause, it will easily read file2's main() clause as the same. It'll internally merge these.

      So if you do want to work with these models to achieve complex tasks, you basically do have to go reverse centaur and bend the code base to it's blindness. You can't use the same function names across the code base; each one needs to be dstinguishable; same thing with variables that represent seperate entity relationships.

      You do that, and it suddenly because a whole lot smarter.

  • alecbz 48 minutes ago
    We have LLMs try to generate descriptions of PRs for us and they're pretty universally disliked. They're always overly-complex descriptions of the mechanical changes and have no sense of motivation.

    Also, a huge reason to understand the code yourself is to make sure the LLM isn't wrong, but this doesn't work if an LLM is itself generating the understanding.

    • avaer 7 minutes ago
      Maybe on your team, but I don't think AI PRs are universally disliked. The people that submit PRs without understanding them are universally disliked.

      Have you tried writing in AGENTS.md or whatever to exactly explain what you like/dislike about the PR descriptions?

    • dylan604 27 minutes ago
      Are these generated descriptions of LLM submitted changes or of human changes? If a human, shouldn't they be putting the motivation into the PR?
      • alecbz 23 minutes ago
        LLM-generated (or at least LLM-assisted), but a human's still the one submitting the PR for review.
        • dylan604 18 minutes ago
          A PR with a minimal title and empty description should be refused at submission. If the human is so disinterested that they're using LLM generated code and then can't explain the purpose, that human should be prevent from making the PR. Working as a solo dev, it is very easy to be lazy like that, and I'm as guilty as anyone. Working in teams with actual reviews should absolutely have much more strict policies of what is considered a valid PR
          • alecbz 1 minute ago
            I mean, the “meta” right now is trying to automate as much of the software engineering flow as possible, including writing PR descriptions.

            They’re shooting for LLMs being able to one-shot PRs or need minimal oversight. But yeah, in practice LLMs are not there IME.

    • vjvjvjvjghv 45 minutes ago
      I don’t know. From my experience I get pretty good descriptions of PRs if I ask the right questions and provide some context.
    • morkalork 37 minutes ago
      I am so very tired of 2 page long PR descriptions for a 5 line change.
  • iainctduncan 54 minutes ago
    I am so dying to read more about the new/current/real bottleneck!

    Where is the bottleneck? WHERE?? Tell me! No evidence needed, just lay it on, man to man, thought-leader to thought-leader!

    • euroderf 24 minutes ago
      > thought-leader to thought-leader!

      This is my new chat-up line at networking events.

    • iainctduncan 50 minutes ago
      Oh wait, there it is, sitting over top of the fat part and under the cork...
  • euthymiclabs 1 hour ago
    "I read the code." -Mitchell Hashimoto

    Great code needs great understanding and agents need excellent guidance. Even in my current solo-dev work, I can't imagine making a production commit I haven't read until I understand it. I own the consequences of my code; that's a responsibility AI agents can't take.

  • fabiensanglard 52 minutes ago
    While the tips are good to handle the volume, I still think this sets code owner on a dangerous path.

    AI have limitation and hallucinate. Complex code will be explained in hallucinated way. At some point AI will be unable to write more because the arch has become too complex or the volume of code will be to high.

    The article I would like to read would suggest how to force LLM to architect the code like a solid tower instead of a pile of unstable mud.

    • vjvjvjvjghv 41 minutes ago
      I feel if you still architect the code and guide the LLM, it will do a pretty good job. Maybe one day the LLM will be able to do all the system architecture but that’s probably still quite some time out. I don’t even know if that’s possible considering different business needs and other factors that aren’t technical.
    • alikhater30000 22 minutes ago
      [flagged]
  • hk__2 40 minutes ago
    For me the solution has been to throw away the code I don’t understand. I let the agent write the code, and if when I read it it seems unclear or needs a lot of explanation from the agent, I just throw it away and start over, or do it by myself.
  • wseqyrku 27 minutes ago
    If you try to spec the problem with all the painful details for the machine to understand, you will end up with a rust codebase.
  • bigstrat2003 21 minutes ago
    Understanding has always been the bottleneck. That's why LLMs aren't actually helpful: they speed up the part which is easy (typing characters into your editor), but are neutral or even harmful on the part which is hard (understanding the problem and how best to solve it).
    • stronglikedan 7 minutes ago
      The part which is easy is still time consuming, so LLMs are helpful. They're just not a silver bullet.
  • elendilm 6 minutes ago
    Understanding is expensive. And hence valuable.

    LLMs usually points to the most idiotic future trajectory on my work, and I have to curse it inorder to let it keep up with my refined understanding.

    But what else would one expect from a probabilistic weighted next token predictor, other than to conduct probabilistic search which are 99.99% deadends.

    But LLMs can pave the way towards constructing resilient and correct architecture which can be iterated fast by a human.

    Architecture and determinism is where my money is in.

  • jbdamask 46 minutes ago
    pfft...I'm way past understanding
    • ch4s3 37 minutes ago
      Overstanding is the new horizon
      • stronglikedan 6 minutes ago
        That's already been co-opted by the sovereign citizen movement, so use with caution lest you be judged for it.