Post #3290837

2026-06-08 12:05 UTC

@fireborn@dragonscave.space I think the vibe coding trend in the blindness tech community is at odds between necessary empowerment and ignorance and disdain for expertise. I notice a lot of experts, in open source and in commercial tech companies, have treated blind users with disdain, laying the groundwork for a backlash as soon as this technology became convincing enough. Not good enough, convincing enough. Basically we’re now faced with a group of people who deeply emotionally distrust the ones who know what they’re doing, and can’t necessarily tell apart the experts who care from the ones who don’t. They also lack the ability to distinguish between what’s simple or complicated, since the whole process is a black box. Chances are nobody who was in a position to teach them five years ago bothered to. Ultimately, I think the solution is the same that was always proposed by the open source movement: community. The question is how to establish the necessary emotional bonds. #blind #FOSS

Replies (2)

@fireborn@dragonscave.space 2026-06-08 12:12

@MostlyBlindGamer@dragonscave.space I agree, but that isn't really the point of the post here. I understand why people do the things they do, but it's not the right way. Building a community of experts that actually care is, but that comes with time, and the first step is to accept that these hundred projects made by AI is not the solution.

Open ##3290838
@zersiax@cupoftea.social 2026-06-08 12:19

@MostlyBlindGamer@dragonscave.space What exaserbates this is the lackluster attempts at #accessibility we're used to, I think. I'm branching off something you said slightly but LLM-generated output in the #accessibility space often doesn't need to be *good*. It needs to exist and work, to begin with. Because the alternative is zero access at all, with no seeming recourse particularly from non #FOSS projects. Accessibility "fixes" that are just patched on in hindsight, which is most of them, rarely work elegantly or efficiently, they just ... work, ish, kinda. Compared to an LLM that you can iterate with until it works good enough for you specifically, that's almost not even a choice worth considering for some.

Open ##3290842