Drew Mochak1d ago
CodeByJeff

As a programmer, I wish I could help people understand that the fastest part of any project is writing the code...

Non-programmers always want to throw more code writers at a project to speed things up. The "Mythical Man Month" problem aside,

the bottlenecks are all the legwork involved in defining the problem, the scope of this particular task, and fully testing it.

Even before AI, writing code was fast and we already have a lot of tools to crank it out...if we're clear on *what* to crank out.

This is prolly true in all fields, this difference in understanding between the people who actually work in a field, and the people who are adjacent to that field.

Mar 19 at 9:12pmWeb
Show threadjeremiah2d ago
@codebyjeff The only part of code writing that has historically not been fast for me is writing tests but that might be more a problem of knowing what to test and how to accurately capture that in an automated fashion versus the writing itself.
Show threadCodeByJeff2d ago
@jeremiah that's absolutely me, but thankfully I've been showing improvement in that area lately
Show threadPadraic Harley2d ago
@codebyjeff "Measure twice, cut once" =)
Show threadAlanna 🏳️‍🌈🏳️‍⚧️2d ago
@codebyjeff writing code with the so-called "AI" is probably slower TBQFH.
Show threadCodeByJeff2d ago

@kelpana the small bit I've done has proven to take about the same amount of time, but been incredibly boring

I would love to try it out on a large factoring task, though - there's where I think it would really shine

Show threadAlanna 🏳️‍🌈🏳️‍⚧️1d ago
@codebyjeff every time my IDE updates it switches "AI" autocomplete back on. It is extremely annoying more than anything else.
Show threadIvan Sagalaev 2d ago
@codebyjeff what's absolutely *amazing* to me is when actual experienced programmers start singing songs about increased productivity and give examples like "C compiler in two weeks from scratch". They should know better, but it looks like there's something about those tools that captures them on a deep psychological level.
Show threadJack William Bell1d ago

@codebyjeff

Watching the British show 'Grand Designs' has taught me how many parallels there are between a coding project and building a house. And it's pretty much everything you described.

You want to see a house build go badly over budget and schedule? Just do it on your own without retaining an architect past the 'getting permits' stage, manage it yourself, and change your mind about *anything* as you go.

Show threadTom 🇵🇸 🇺🇦 🏳️‍🌈 🏳️‍⚧️1d ago

@codebyjeff
Maurice Wilkes, an early computing pioneer, reflecting on the early days, said, "I remember the exact moment when I realized that a large part of my life from then on was going to be spent in finding mistakes in my own programs."

With AI-generated code, programmers give up the creative pleasure and joy of writing code, and multiply the amount of tedious and difficult debugging instead. What a bargain.

Show threadCodeByJeff1d ago

@tom we also give up a lot of the tedious boilerplate parts of our jobs...

I think it can be argued both ways, depending on how you personally are using it

Show threadRobert Brewer1d ago
@codebyjeff yes, and the REALLY hard part is making the design coherent, which tends to become harder after about 3 people working simultaneously. AI can never good cohere, it can only bad cohere.
Show threadHanno Rein1d ago
@codebyjeff it’s also the really fun part
Show threadWesDym1d ago
@codebyjeff Logically, nine women working together should be be able to produce a baby in just a month.
Show threadPeter Ludemann1d ago
@codebyjeff Probably true for the majority of software, that the biggest problem is figuring what the problem is (business rules, use cases, UX, etc)
But some problems are mostly about the code ... compilers, databases, webservers, browsers, operating systems, paxos/raft, end-to-end cryptography, etc, etc.
Show threadJage1d ago
@ChanceyFleet @codebyjeff I feel this. Even working with AI coding, I find that at least half or more of my time is spent planning, testing, etc.