Thread

1. alkonaut · 1785999287 · focus · HN ↗

   Programming has five phases effectively:

   1) You figure out what problem to solve.

   2) You figure out HOW to solve the problem.

   3) You actually implement the solution.

   4) You see the solution work, for yourself.

   5) You ship/deploy/publish the program. This means you see people be happy users and/or you get paid for it and so on.

   If you're an entrepreneur type, you probably enjoy the first and last steps most, and you see steps 2-4 as mostly a chore. If you're a tinkerer, you don't care much for 1 and 5, and you see 2-4 as the whole point of programming. I'm a tinkerer. I'd be happy to just write code and throw it away. Coding is like solving sudokus. I could skip steps 1 and 5 forever. I don't ever need to show any code to anyone. In fact, most of the time when programming I do steps 2 and 3 and even skip 4. I don't even finish! I work weeks on something until I lose interest, and I know that in order to even run it, it would be several more weeks. A PoC is enough. Or just a half one. It's just code-to-structure-thoughts, not to create anything finished.

   The 5 phases look kind of symmetric. The outermost layer (1 and 5) are the entrepreneurial steps. If you're a product owner or CEO, you might work strictly at steps 1,5. Then steps 2-4 are the managerial/architectural steps. If you're a very senior IC at a large company, you might work at this level, without actually doing much coding. Only the inner most step (3) is the manual creation of source code. Even though it's 5 different phases, it's just "3 layers" of programming.

   The problem as I see it is that I enjoy step 3. And LLMs are good at step 3 almost exclusively. So they just pick the best bit of this dish, and leave me with the rest.

   If you're an entrepreneurial type, the LLM appears to take the _worst_ bit of the work from you. Great.

   1. yunwal · 1786026177 · focus · HN ↗

      I enjoy the process of solving problems with LLMs. I understand it's chaotic (what many people call non-deterministic), and that doesn't appeal to some people, but solving problems with statistics is problem solving just the same. Just because medicines don't work with 100% reliability doesn't mean Jonas Salk didn't solve a problem. The same is true with LLM solutions that don't work 100% of the time.

      I think most people will come around to this way of thinking once the pro-LLMers stop pretending like the statistics don't exist too (i.e. that ML is the solution to any problem, despite how unreliable it is).

      1. thesz · 1786094142 · focus · HN ↗

           > The same is true with LLM solutions that don't work 100% of the time.
         

         In my experience, about one of ten LLM suggestions during code review is important and valid. Nine of ten are either unimportant or not valid. Yet, I have to read them and rule them out.

         Also in my experience, having my colleagues do root cause analysis (RCA) with LLM brings to me invalid RCAs which I have to rule out, instead of working on finding true RCA.

         I saw part of documentary about Charlie Abrahams, son of Jim Abrahams. Charlie had epileptic seizures as a child, hundredths a day. In the documentary, Charlie's doctor said that no drug in his experience worked as fast as ketogenic diet (seizures stopped in three days) and as reliably as ketogenic diet (works on more than 80% of people). When he was asked why he did not applied ketogenic diet, he said there still were drug options to try.

         "Medicines do not work 100% of time," my bald cranium.

         1. yunwal · 1786112016 · focus · HN ↗

            > Nine of ten are either unimportant or not valid. Yet, I have to read them and rule them out

            You don't have to do this? Maybe someone is making you, but it's literally possible to just stop using LLMs for purposes where they fail reliably.

Unofficial Hacker News client; not affiliated with Y Combinator.