I Want to Love Linux. It Doesn't Love Me Back: Post 3 – Speakup, BRLTTY, and the Forgotten Infrastructure of Console Access
53 points by abhinav
53 points by abhinav
I think these posts are very valuable, and I wish Linux / Open Source ecosystem accessibility story improved. Before reading them I wasn’t aware how sucky the accessibility experience is.
But it needs to be recognized that Linux / Open Source ecosystem is decentralized and has very limited resources. It just isn’t a multi-trillion corporation, making money on every user. Realistically, for things to improve there’s got to be some funding for organization / dev teams focused on improving them, getting private donations and government grants, etc. Then it can collaborate with mainstream projects, and even help maintain distros specifically optimized for accessibility, etc. Otherwise things will stay inevitably sucky.
IBM has a revenue of $62 billion per year and sells RedHat linux. IBM purchased Red Hat for $34 billion, and Red Hat is responsible for $6.5 billion per year of IBM’s revenue. Canonical has a revenue of $250 million per year.
There’s plenty of resources in the Linux / Open Source ecyosystem… if it became illegal to profit off of Linux without first making it accessible (or to the extent that it already is, said laws were enforced) these problems would disappear in a week. It’s a question of the allocation of resources, not the lack of them.
Both IBM Red Hat and Canonical Ubuntu are products primarily but not exclusively for servers. They target a truly absurd range of hardware, and represent a combined revenue of $7B. Compare, for example, Apple iOS and Apple Mac OS, often held up as the gold standard for accessibility, which targets a tiny range of devices and represents some fraction of a revenue of about $76B (that is, 69B for iOS devices plus 7B for Macs). We don’t know what that is, but even if we imagine that Mac OS’s development only sees about a quarter of that, that’s $19B. Way more money, spent on mostly end-user systems, targeting way less hardware. This is not to justify their lack of focus on accessibility, but I do think it’s unreasonable to imagine that they’d accomplish similar outcomes on their own. That money needs to make its way into either dedicated teams improving accessibility, or into the community.
I think that dpc_pw’s suggestion of funding organizational structures to lobby for regulations, advance interfaces that promote accessibility, and fix issues like those highlighted by the OP is exactly the right way to a) move some of the existing funding towards accessibility improvements and b) make targeted interventions to improve accessibility outcomes in important places in the stack. Otherwise, what’s the action item here? Yell at volunteers for not volunteering right?
I don’t buy that MacOS sees a quarter of that, hardware is really expensive. But regardless, it’s not the right question, Apple is wildly profitable despite being accessible, not barely scraping by. There’s no reason to think that they couldn’t support the same level of accessibility with much less revenue.
The right question is “how much would it cost” and “would IBM and Canonical bear that cost or exit the business of distributing and supporting operating systems that run in a non-accessible manner”. The answer’s in my opinion are “a couple million per year, or less” and “definitely not, that would be the end of a business worth more than 100x the cost”.
Otherwise, what’s the action item here? Yell at volunteers for not volunteering right?
I thought I was rather explicit about that, lobby for new legislation and/or the enforcement of existing legislation requiring people making money only do so off of accessible systems. It’s hard to get more specific without knowing what country someone is in. I’m not asking for volunteers at all, the legislation I am suggesting targets only professionals profiting off of the software.
This is the only thing that has ever worked to get companies to allow disabled people access at expense to themselves, and there’s plainly enough money involved here that it would work to do so. This is why when you walk around (assuming you live in a city similar to mine) you see wheelchair ramps and elevators made at (in total) significantly greater expense than it would take to make Ubuntu and Fedora accessible.
I think we are vehemently agreeing, then - there should be an organization, staffed and funded, with the charter of “lobby[ing] for new legislation and/or the enforcement of existing legislation requiring people making money only do so off of accessible systems” as well as making targeted interventions in hardware and software to solve issues that aren’t “owned” by any of these companies. Right?
Yell at volunteers for not volunteering right?
Hmm. I would never demand that volunteers go out of their way to work on any particular feature, but I do think it’s reasonable to expect anyone authoring software for more than themselves* to (a) not break existing accessibility features in the software, and (b) not make choices that break accessibility when other reasonable choices are available (e.g. writing a GUI app using a framework with no screen reader support).
* or themselves and a few specific people they know don’t have particular accessibility needs
Absolutely agreed. Morally, I think it’s the right thing to do. However, my experience has been that this is generally not very effective in specifically motivating software engineers to work on specifically accessibility features. Hell, a comment from a few months ago on here where I did exactly that is highly flagged. People don’t want to hear it.
I think it’s fair to point at these big companies that do actually have resources and ask them to do better.
if it became illegal to profit off of Linux without first making it accessible
I think promoting regulations to improve it is very much in the scope of an organization I mentioned.
Or someone could try to sell Linux desktops to governments that have accessibility requirements. Or try to sell Linux desktops at all. I think there’s very few people trying to make money selling Linux desktop.
(I believe there’s some workstation Linux market in some markets.)
If the people putting the money are only getting money from servers, that’s where most of the improvements will land.
Several companies sell pre-configured Linux machines, which actually sidestep a lot of these issues by having sighted employees get the system into a state where the screen reader works - and anyway, a lot of them ship software that works pretty well without that configuration, as long as the hardware is supported. In my testing, for example, Pop!_OS 22.04 (installer ISO hash 701c6ed55ecbc5bac9ecfe876e9efb6405dab2d6d94f49338f7d9f78cea17e68) dropped me into a graphical installer where Meta + Alt + S started Orca working well enough that I was able to install the OS with my eyes closed (it has a few papercuts; I filed an issue.)
Upon restarting, I was presented with a login screen whose Orca toggle keyboard shortcut worked and dropped into a desktop I was able to navigate with only the keyboard. They are shipping an accessible-by-default OS, which absolutely does not have the audio issues discussed if you buy hardware from them.
That is very interesting and surprising.
Which kinds of companies do that? Is this something they advertise?
I’m sure many people using Linux for a long enough time remember when practically everything required making “it work through manual configuration, scripting, rebuilding broken packages, and sheer force of will.” Over time people came up with ways to make things work more smoothly, but only after it bothered somebody enough to do something about it.
The strength of the open source ecosystem is that people can solve their own problems. A pitfall is that’s usually the only way things get done.
Is the windows or macos version of this a better story?
According to the documentation, macOS has voiceover available in rescue mode and on the boot menu.
Entering rescue mode is easy on the arm models: hold the power key.
I haven’t seen single-user-mode for over a decade and I’m not sure you need it because recovery mode has a terminal.
It’s a shame that folks with disabilities are not more considered in the process. It’s a human attribute that if I don’t have the problem, I don’t think about it. It would be nice to see more accessibility features in Linux. It’s one of the few things that Apple does pretty well.
Honestly, I think a huge part of the issue is that there is no real “process” that spans the entirety of the Linux desktop. There’s just not enough funding for that kind of infrastructure to exist. I think these posts are really valuable, and fireborn’s writing is quite persuasive and shines a light on important topics, but treating “Linux” as synonymous with free desktops, and free desktops as some kind of cohesive entity, has always been a mistake.
This article is mostly not about the desktop, though. Console access should be universal.
It’s extremely telling that the way to work around the PipeWire session lock for audio isn’t meaningfully documented anywhere but in a chance comment on the author’s blog. Linux accessibility depends on so much word of mouth, and a whole extra heap of never letting anything upgrade automatically thereafter.
“Desktop” is an overloaded term; I intended, here, to use it to refer to the entire system that makes up a usable personal computer’s software, meaning bootloader, kernel, drivers, and userspace programs.
Console access should be universal.
It’s all very well to put it that way, but in practice “universal” means “everyone who distributes Linux systems should have the same opinions about this.” That’s not enforceable. What we need is interfaces and protocols that are designed to enable access tools, and funding for that work.
Linux accessibility depends on so much word of mouth
True, and I am not arguing against this is my original comment.
“Desktop” is the graphical thing. “Console” is the text thing you get at boot up, sometimes via a tty. Every Linux system has a Console.
It being accessed directly in this way is pretty much thing only on Desktop systems though. Nobody reads console output directly on an embedded device, almost all servers, … – if at all it is accessed over network or serial ports and then the accessibility-concern is with the client used to do that.
So talking about Desktop systems as the whole “usable personal computer” makes perfect sense, plenty things that are not specific to the graphical desktop environment are relevant to the overall Desktop use case and experience.
I do wonder if the majority of the audio-centric issues could be solved with stacking pipewire daemons, a global one, then per user instances which instead of plugging into hardware, send their stuff off to the global daemon instead? (Doesn’t work currently I’d imagine).
I wonder if a dedicated distro or more would be a better option at this point to have a fully integrated and safe environment, though I do wonder if that would give permission to the other distros to slip further.
I think having a Qubes OS design might also help, a hypervisor that is pretty much eternally unmoving, almost no updates, etc etc. a keyboard shortcut to drop everything into the guest VM, and a keyboard shortcut to pull everything back out, need to poke around? mount the guest storage and cat the logs.
At the risk of sounding like a zealot: I wonder if NixOS might address some of the issues outlined in the blog post. It’s pretty common to make custom install ISOs, which are configured very similarly to your regular NixOS config file, so someone presumably could make a NixOS installer that has Fenrir included by default. Furthermore, whenever you change or update your OS, a boot entry is kept with the previous “generation”. This means that if something does mess up your setup, you’re not forced to boot into recovery mode and hope for the best, you can just boot up a previously working version with all the accessibility tools properly configured.
To be clear, I’m not arguing this is a perfect solution. NixOS is a pretty esoteric system, and saying that people should learn this esoteric system just to be able to use Linux in an accessible manner is not the message I want to bring across. If, however, there are people that enjoy tinkering with their OS and appreciate declarative OS configuration, then NixOS might be something to check out.
If I wanted to write a driver for braille systems, can anyone suggest a common device that is used by those with vision impairments that’d be a good entry point into the world of accessibility tech?