Conversation

Notices

1. Embed this notice
   Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 21-Aug-2026 06:16:55 JST Paul Cantrell

   I just registered my kid for fall school athletics, and now feel like Heracles returning from the underworld with Cerberus on his back.

   WOW is that online registration platform bad software.

   1/3

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 21-Aug-2026 06:20:48 JST Paul Cantrell

     in reply to

     The software in question is Arbiter Sports (arbitersports.com).

     The headliner among the •numerous• software flaws I encountered in this single registration process: the site fails validation when making subsequent edits to a form that the system •had already accepted•.

     The only user-visible message is “Required Information Missing.” (There was no required info missing; again, it already passed validation.) The problem, visible only in the Javascript console, was that the site was rejecting non-alphanumeric characters in a phone number it had pulled from its own database.
     
     2/3

     Rich Felker repeated this.
   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 21-Aug-2026 06:30:48 JST Paul Cantrell

     in reply to

     I’ve been thinking a lot about the gulf between people who think LLMs can cut 90% of the cost of software development or whatever and people who think that’s a pile of catastrophes balanced on top of a nonsense wish.

     I think a lot of that chasm boils down to the fact that software is already •terrible•, and people have very different feelings about that. I don't just mean “Liquid Glass is ugly” terrible; I mean “it’s broken and insecure and unreliable and just doesn't flipping WORK” terrible, like Arbiter Sports.

     We are right now in a war between people who think software should treat humans humanely and people who are perfectly willing to offload their development costs onto captive users in the form of wasted time, anger, anxiety, and even more severe forms of personal harm. That war was raging well before LLMs entered the picture, but they’re exaggerated. I’m extremely skeptical of the idea that LLMs make any significant positive contribution to the long-term speed or cost of writing •good• software, but they clearly have the potential to lower the floor for •bad• software to unheard-of levels at unheard-of speeds. How you feel about that probably predicts a lot about how you feel about the “LLMs can 10x software dev!” hype.

     3/3

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 21-Aug-2026 06:44:11 JST Paul Cantrell

     in reply to

     OK, I guess I wasn't done after all.

     Much of the opining about developer backlash to LLMs falls into two loose theoretical categories:

     (1) Labor problem: developers now experience employment threat, loss of job security, layoffs, new levels of worry about the future

     (2) Identity problem: developers used to feel special because generating code was placed on a cultural pedestal (sign of a special third brain lobe or whatever), that perceived specialness was a pillar of our personal identity, and now our egos are bruised because machines can do it and we don’t feel special anymore

     While there’s doubtless truth to both of those in some / many cases, I posit a third theory:

     (3) Purpose problem: contrary to popular belief, developers actually care quite a lot about the people who use out software and derive a sense of purpose and meaning from doing what we believe to be good work; the current environment undermines that, not just LLMs but LLMs as part of a broader vortex of enshittification, and our world is effectively saying “everything you cared about doesn't matter, and you can't fight to make it matter”

     I think there's evidence for that last theory in the kinds of despair you hear from developers: not just anxiety about paying the bills, not just angry feather-puffing, but a kind of existential despair that comes from the unraveling of one’s internal sense of purpose and meaning.

     4/3

   • Embed this notice
     Þommy (thomasjwebb@mastodon.social)'s status on Friday, 21-Aug-2026 06:50:54 JST Þommy

     in reply to

     @inthehands I keep having this exact experience of not even being able to submit a form and not even being able to make it work by trying other browsers and randomly typing and deleting in fields. There really is an epidemic of poor quality software. And it doesn't matter if theoretically AI could increase quality rather than speed because in the real world, it's used to increase speed and lower quality, like so many shortcuts the industry employs before it.

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 21-Aug-2026 06:50:54 JST Paul Cantrell

     in reply to

     • Þommy

     @thomasjwebb

     Yes, that. I occasionally surprise people who think I’m hard-core anti-AI by saying that yes, of course one •could• use LLMs in ways that would •increase• software quality (and slightly slow its development in the near term, with compounding benefits in the long term). But that’s not even remotely like how people are using it in practice, and that’s not a multi-trillion dollar industry to dangle in front of investors.

     (There are also myriad environment and ethical showstoppers too, but just in terms of the raw tech…sure, of •course• one could use it in a completely different way that’s focused on detecting quality problems instead of speeding output.)

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 21-Aug-2026 06:52:42 JST Paul Cantrell

     in reply to

     • Large Heydon Collider

     @heydon

     I’m arguing it goes further: it can make software at levels of shittiness humans have barely even contemplated, at a tiny fraction of the cost. That’s what I mean by “it lowers the floor.”

   • Embed this notice
     Large Heydon Collider (heydon@front-end.social)'s status on Friday, 21-Aug-2026 06:52:44 JST Large Heydon Collider

     in reply to

     @inthehands The real promise of LLMs/agents is they can make software just as shitly as humans, at a comparable cost.

   • Embed this notice
     Jonathan Schofield (urlyman@mastodon.social)'s status on Friday, 21-Aug-2026 06:55:38 JST Jonathan Schofield

     in reply to

     @inthehands thank you for putting so eloquently how I feel – as a designer who codes more than as a developer developer.

     The phrase I come back to as my attempt to encapsulate it is *the lack of care economy*.

     That’s always been underlying financialised modernity but it’s currently running unabashedly rampant

   • Embed this notice
     Chris Adams (acdha@code4lib.social)'s status on Friday, 21-Aug-2026 07:00:06 JST Chris Adams

     in reply to

     @inthehands I maintain that people would be far less upset about this if LLMs weren't the tool being used by companies to say quality, accessibility, security, etc. don't matter any more. If we had healthier markets this would be far less pronounced: companies only cut QA when they think customers won't stop buying.

   • Embed this notice
     Greg Whitehead (grwster@mastodon.social)'s status on Friday, 21-Aug-2026 07:00:20 JST Greg Whitehead

     in reply to

     @inthehands Yes to #3. We used to have this frustration when timelines or budgets or crappy outsourcing resulted in poor quality software and now it’s just the baseline state-of-the-art

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 21-Aug-2026 07:03:37 JST Paul Cantrell

     in reply to

     Put more concisely:

     I’d bet money that developer reaction to LLM-driven development would be •very• different if its practical effect were to force management to slow down development and sink more money into quality.

     5/3

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 21-Aug-2026 07:23:25 JST Paul Cantrell

     in reply to

     • Robert Lee

     @robertlee

     Good luck and godspeed. You are not alone in this!

   • Embed this notice
     Robert Lee (robertlee@mastodon.world)'s status on Friday, 21-Aug-2026 07:23:27 JST Robert Lee

     in reply to

     @inthehands

     Number 3 is me.
     I've been writing my software for 23+ years. As an employee, a contractor, and then as a partner with a larger percentage of ownership each year until I owned it all.

     It's my baby, it feeds the employees in my tiny company, and runs the businesses of my ~85 clients.

     I'm going out of my way not to enshitify our product, and to continue to provide quality and value, and I'm not going to let LLMs near our code (for both the ethical and technical reasons).

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 21-Aug-2026 07:24:47 JST Paul Cantrell

     in reply to

     Yes. “Moral injury” is indeed a relevant concept here.

     https://infosec.exchange/@EINS_Institute/117130150598686968

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 21-Aug-2026 07:25:14 JST Paul Cantrell

     in reply to

     • slackline :emacs: :orgmode:

     @slackline

     Saw and appreciated that too, and yes, it's another part of the picture for sure.

   • Embed this notice
     slackline :emacs: :orgmode: (slackline@mastodon.social)'s status on Friday, 21-Aug-2026 07:25:15 JST slackline :emacs: :orgmode:

     in reply to

     @inthehands In a related vein I read this the other day and thought it was good

     https://ryancheley.com/2026/08/17/what-if-maintainer-burnout-isn-t-burnout/

   • Embed this notice
     Steve Purcell (sanityinc@hachyderm.io)'s status on Friday, 21-Aug-2026 07:42:28 JST Steve Purcell

     in reply to

     @inthehands Yes, absolutely tagging myself as 3 here. Nicely expressed, thank you.

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Friday, 21-Aug-2026 07:55:27 JST Paul Cantrell

     in reply to

     • V

     @miss_rodent

     I mean, there are some absolutely absurd egos in the tech world, so. But big picture yes.

   • Embed this notice
     V (miss_rodent@girlcock.club)'s status on Friday, 21-Aug-2026 07:55:28 JST V

     in reply to

     @inthehands As an added note; The labour issues were starting to grow as a problem anyway - we passed the point where "Learn to code" was a path to a stable career before the slop, as far as I can tell;
     and the identity issues... really, how can you have that much of an ego about it when you occasionally get outdone by - or seriously work alongside in the FOSS side of things - a literal child or teenager, who's not even out of grade school yet?

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Friday, 21-Aug-2026 08:23:11 JST Rich Felker

     in reply to

     @inthehands I think the (3) "purpose problem" doesn't adequately characterize it.

     It's not about our purpose. If our non-programming selves and everyone else would be fine without us having to make software, this "loss of purpose" would not be a big deal.

     The problem is that the things we care about are bound to software working right. Not exposing out privacy. Not breaking and losing our data. Not exploiting wetware bugs in our brains to serve evil people's political and economic goals. Not policing who we can communicte with and what we see. Not erasing our history and rewriting it out from under us. Etc.

     Software will not work right when it's vomited by machines owned by and serving the interests of our enemies.

   • Embed this notice
     Yeshaya Lazarevich (alter_kaker@hachyderm.io)'s status on Friday, 21-Aug-2026 09:49:14 JST Yeshaya Lazarevich

     in reply to

     @inthehands on point 3, agreed. And, a secret: there might be a national campaign coming this fall aimed at addressing just that. Stay tuned.

   • Embed this notice
     Raphael (omegapolice@hachyderm.io)'s status on Friday, 21-Aug-2026 15:40:38 JST Raphael

     in reply to

     @inthehands Tracks.

     Also, we know how shaky much of tech already is, and some of us don't believe that rejection-sampled code will improve on that. We're _scared_ of the inevitable failures.

   • Embed this notice
     Ken Brucker (sigsegv@hachyderm.io)'s status on Friday, 21-Aug-2026 23:23:00 JST Ken Brucker

     in reply to

     @inthehands I’m among (3). I resent those that are using LLM to put slop in my review queue but don’t know what I can do about it. I look at my woodworking and at least there, that’s still a realm that is void of the AI pollution that I can escape to (at least at the execution side, I’m sure there’s plenty of AI slop in articles about woodworking).

   • Embed this notice
     davecb (davecb@hachyderm.io)'s status on Saturday, 22-Aug-2026 01:25:07 JST davecb

     in reply to

     @inthehands Outgoing inspection *can work*.

     My VP tried out a demo of a program we were writing for a major-dollar customer and found something like 80 bugs before he gave up.

     QA isn't the answer. A good Technical Product Owner is. My VP was good at that.

   • Embed this notice
     davecb (davecb@hachyderm.io)'s status on Saturday, 22-Aug-2026 01:36:55 JST davecb

     in reply to

     @inthehands I have found some cases where LLMs can be used to improve quality.

     An example from the ACM is in my blog, at https://leaflessca.wordpress.com/2026/06/07/hedgewitch-part-3-llms-should-challenge-not-obey/