Conversation

Notices

1. Embed this notice
   Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 15-Jul-2025 01:25:42 JST Paul Cantrell

   Code is a confusing engineering object for our human brains. It is a bit like building a bridge or a kitchen utensil: it’s an object without a mind, it has a function, it can fail, people do unexpected things with it. But it can also feel a bit like a person given a task: it has behavior, it •decides•, it •acts•, it •causes•.

   (All arguably true of a bridge too, but most of us don’t think of bridges that way!)

   The lay understanding of code leans heavily on the idea that programs are anthropomorphic, little homunculi with agency.

   2/

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 15-Jul-2025 01:31:54 JST Paul Cantrell

     in reply to

     Where that understanding goes wrong is that humans have perception and experience and common sense. And yes, people are foolish and fallible — but ultimately humans are the adaptable element of complex systems[1], and when we design processes involving humans, we always, always lean on that adaptability.

     If we say “walk out the door,” humans generally will not just walk face first into the wall just because the door is behind it.

     Code will — unless you tell it not to.

     [1] https://how.complexsystems.fail/#12

     3/

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 15-Jul-2025 01:34:13 JST Paul Cantrell

     in reply to

     So much of the work we do in software is about asking not just what •does• happen when the developer runs their own code, but what •could• happen when the code is running in the wild, out from under the watchful eye of its author.

     That’s really, really hard. It’s a large portion of what makes development time-consuming and labor-intensive.

     And it’s something you can only do if you actually understand what the code •means• — both to the computers and to the humans.

     4/

   • Embed this notice
     Yeshaya Lazarevich (alter_kaker@hachyderm.io)'s status on Tuesday, 15-Jul-2025 01:35:36 JST Yeshaya Lazarevich

     in reply to

     @inthehands this is why I always say that the best thing for me as a dev is to partner with a great tester

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 15-Jul-2025 01:38:33 JST Paul Cantrell

     in reply to

     A lot of the design work that goes into programming languages and tools is about prompting developers to •think about meaning•: tests, types, scope, compile errors, runtime errors — all about •preventing code from running• in the presence of an expectation/reality mismatch.

     I’m always on high alert for tools that promise to speed development by letting developers skip the thinking.

     5/

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 15-Jul-2025 01:41:43 JST Paul Cantrell

     in reply to

     That’s not necessarily the only way to use machine learning to assist coding. It is, however, one of the “magic diet pills” promises at the heart of the current hype bubble.

     Whatever the dev tools of the future look like, until we have Lieutenant-Commander Data, difficult human thought will be at the heart of writing code.

     6/

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 15-Jul-2025 01:43:21 JST Paul Cantrell

     in reply to

     In a saner environment, we’d be having a reasonable conversation about that: in what ways, if any, can a machine that repeats contextual patterns with no sense of meaning augment humans thinking through tricky things? In ways can it mislead? When, if ever, is it worth the tradeoffs? the resource costs? etc etc.

     Right now, the off-the-charts money and hype make that reasonable conversation impossible except perhaps in hushed corners. (Please do not have it in my replies. I am tired.)

     7/

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Tuesday, 15-Jul-2025 01:45:26 JST Paul Cantrell

     in reply to

     (Footnote: code that only has to run once is a different beast from code that has to work in a production environement. I’ve noticed a mismatch between data science / research types using LLMs to jump start one-off code and developers writing code for unknown end users / prod env. Understanding that mismatch may help some of these conversation be less unreasonable!)

     8/