Post #115705
2026-01-06 19:52 UTC
Replies (36)
-
@crazyeddie__dup_521@mastodon.social 2026-01-06 22:15
@Chip_Unicorn@im-in.space While I have no problem creating 3k lines of code in a day that only I can understand and describe how it works.
-
@shadowcheets@meow.social 2026-01-06 19:56
@Chip_Unicorn Vibe coders are kind of a menace to a code base. ^^; I'm sorry you have to deal with that!
-
@munin@infosec.exchange 2026-01-06 20:00
@Chip_Unicorn @cstross yeah, the people - veeeeery limited number here - I know who use these things ... I hesitate to say "correctly" but "with deliberate skill and intention" mb? ... use them for basically two use-cases: prototyping new stuff quickly as a "will this approach work at all" check, and for generating easily-predictable boilerplate (e.g. unit tests) that they can check quickly. But in both cases, none of the LLM generated code gets committed unless they've gone over every line to understand what it's doing.
-
@cora@hachyderm.io 2026-01-06 20:04
@Chip_Unicorn @cstross I like to think of vibe coders as fertilizer nourishing a future bountiful harvest of desperately needed critical maintenance by those who kept their tools sharp.
-
@TimWardCam@c.im 2026-01-06 20:18
@Chip_Unicorn Plus why is it a good thing that he's produced 3,000 lines of code where 27 would have done the job? And does nobody care about performance either?
-
@mirabilos@toot.mirbsd.org 2026-01-06 20:30
@Chip_Unicorn someone who cannot read code is not a programmer.
-
@Crissa@meow.social 2026-01-06 20:56
@Chip_Unicorn I feel like a Xavier when faced with many modern implementations. There's no syntax glossary, no 'why are we doing it this way' and they've done things just slightly differently and I don't know why a bunch of it is interacting like it does. I hate feeling like a Xavier.
-
@greenpepper22@mastodon.social 2026-01-06 20:57
@Chip_Unicorn 3000 lines of code a day is insane, how do you even review that? They must be constantly violating DRY and all sorts of other good practices in order to get that volume, you’d think eventually they’d be able to make a new feature only doing a little bit
-
@ThePolishDispatch@mstdn.social 2026-01-06 20:59
@Chip_Unicorn Even more, programmers who can troubleshoot.
-
@andrewault@twit.social 2026-01-06 21:00
@Chip_Unicorn I'm now an agentic dev on large projects. My experience as a caveman engineer is extremely valuable. I know how to structure projects, can identify failure patterns, can READ code... I make systems that are self-monitoring, create detailed error dialogs and logging. So when there is a failure, I have lots of data and context. Consequently failures are root caused and fixed properly. Often, so easy to fix that a generated JSON is fed to an agent that creates an analysis.
-
@nyx_lyb3ra@tech.lgbt 2026-01-06 21:09
@Chip_Unicorn problem is when nobody hires someone who doesn't submit features in less than 1 or 2 days
-
@FeloniousPunk@beige.party 2026-01-06 21:19
@Chip_Unicorn We have a jr developer that started a couple months ago. My dude is vapor locked without his LLM tools. I mean yeah, we all use them to some extent, right. They CAN be useful for isolated tasks but, if you don’t understand libraries, architecture, tests, or even how docker works, you’re gonna have a bad time in the real world. Personally, I would LOVE to see a ban on LLM usage through at least the first year of the job. You gotta be this tall to ride this ride, kid.
-
@burnitdown@beige.party 2026-01-06 21:51
@Chip_Unicorn and then the tests are generated by LLM so they're also telling you nothing useful. garbage out, garbage in, it's very clever you see.
-
@dark_stang@kind.social 2026-01-06 22:08
@Chip_Unicorn when I did a lot of interviews, my Sr Dev tech interview process was basically "read this code and walk me through what it's doing" in 5 different languages. I don't expect people to be a deep expert in every language. But I do expect a Sr to at least be able understand the gist in most languages they see.
-
@wrosecrans@mstdn.social 2026-01-06 22:24
@Chip_Unicorn I've been saying for a while that spamming out a bunch of extra code and complexity that nobody wrote is gonna reach a tipping point where it will become impossible for humans to ever unwind it, and that's scary. I think I am in the last generation of humans that understood what computers were doing.
-
@dankmane@mastodon.social 2026-01-06 23:20
@Chip_Unicorn Sounds like I might be able to do what Xavier does. What company is it???
-
@iron_bug@friendica.ironbug.org 2026-01-06 23:37
@Chip_Unicorn sounds stupid and really scary. if such "programmers" appeared that means we will work till the end, without pension, because nobody else can anymore.
-
@t3rcermillenium@mastodon.social 2026-01-06 23:51
@Chip_Unicorn Ok, so old enough to have transitioned from asm to C in the 80’s. The first compilers were shit; I spent a load to time looking at the generated asm for the C code we wrote. It was shit. But that passed. Now we “trust” compilers. Now I spend a lot of time looking at the code generated by Claude, and it works. But it’s shit code. But this too, will pass. Soon, I won’t look at the code generated by Claude. At least, that’s the ideal. Let’s see how that works out.
-
@collimated_thought@defcon.social 2026-01-07 00:00
@Chip_Unicorn I was actually reading a Cory Doctorow piece today on how code is a liability, but business types think it's an asset. I believe this story is a case in point. https://pluralistic.net/2026/01/06/1000x-liability/#graceful-failure-modes
-
@n_dimension@infosec.exchange 2026-01-07 00:05
@Chip_Unicorn Sounds like Xavier needs to improve his prompt skills. OR you can hang his corpse over the door as a message to other humans who betray us to the machine. All of the issues you listed are avoidable with better prompting (source: me ape #vibecode) LLM is just another layer of abstraction. You can't toggle address, instruction and data values on the bus and clock pulse them either.
-
@TahomaSoft@puget.social 2026-01-07 00:49
@Chip_Unicorn Eeek!
-
@cap_ybarra@beige.party 2026-01-07 01:08
@Chip_Unicorn holy shit look at some of these fuckin slopmongers in the comments
-
@pdd@mastodon.social 2026-01-07 01:35
@Chip_Unicorn this is fucking terrifying. The amount of unreadable, untested code that AI is regurgitating is a huge liability to any software project. No wonder @pluralistic was so emphatic at pointing out “software as a liability” in his last talk: https://mamot.fr/@pluralistic/115848731714392646
-
@richpuchalsky@mastodon.social 2026-01-07 02:30
@Chip_Unicorn This person isn't a programmer or a developer. You can't be one if you can't read code. I do admire him as a con man, though. High salary, no skills, not on the hook for anything that goes wrong.
-
@sabrinaweb71@sociale.network 2026-01-07 02:47
@Chip_Unicorn sometimes I'm tempted to ask an llm to write some code just because I want to learn how to fix slop code. I suspect we will need slop code fixers in the future
-
@waffles@masto.yttrx.com 2026-01-07 03:48
@Chip_Unicorn how did person become a developer if they cannot read or write code?
-
@condret@fedi.absturztau.be 2026-01-07 05:21
@Chip_Unicorn This is what Marx meant by the term "alienation". Xavier is an alienated developer
-
@dianshuo@mstdn.io 2026-01-07 06:59
@Chip_Unicorn this problem has existed before LLMs. We would see it in people who “tweak” or “fix” until it works and stop as soon as the “test” passes.
-
@megatronicthronbanks@mastodon.social 2026-01-07 09:01
@Chip_Unicorn This is going to crash the world.
-
@fishidwardrobe@social.tchncs.de 2026-01-07 10:33
@Chip_Unicorn thank the gods he's using tests. but, even so: gaah
-
@byteborg@chaos.social 2026-01-07 12:13
@Chip_Unicorn sigh, I hope I'm not the only one running out of spoons for cluebatting questions on failure mode behaviour into codeprompters. It's an uphill battle until the system implodes. I silently hope that this will happen soon before many many more billions are stuffed down the "AI industry's" throats. @cstross
-
@draeand@social.the-gdn.net 2026-01-07 16:36
@Chip_Unicorn Here here! I rely on LLMs for some stuff but I always understand it (and sometimes even rewrite it). I've never understood those people who just put all their faith and trust blidnly into LLMs thinking that they can just do their job for them.
-
@markotway@mastodonapp.uk 2026-01-07 22:02
@Chip_Unicorn Said it before and will say it again. If your level of programming expertise is such that it will be enhanced by using an LLM, you don't understand the dangers of LLMs and should not be coding with them. And if your level of expertise is enough that you understand the risks of LLMs, then you don't need an LLM to code.
-
@computernut43@nexto.my 2026-01-07 22:25
@Chip_Unicorn unfortunately....I feel like that is the future of coding Now... should someone with 0 coding experience get a coding....no. That's like hiring a surgeon to work in the Operating Room with just an LLM.
-
@Gustodon@mas.to 2026-01-08 10:11
@Chip_Unicorn I hope Xavier gets taken by wolverines.
-
@carl@chaos.social 2026-01-08 10:14
@Chip_Unicorn this