@lynn It just let the controversy die out and everyone moved on. There are too many horrors for a single one to keep everyone’s focus for more than a few weeks. And they no if you want an open laptop you have no other choice anyway.
@lynn Used thinkpads I guess? Lenovo also sucks as a company. But the laptops are easy to fix and there’s a healthy used market for machines and parts.
@fastfinge yeah what i got for myself was a thinkpad t480, which isnt very well performant but it was used. i am really considering framework pro 13 in the coming year or so though because it meets my requirements exactly.
@lynn I have a framework 16. But my plan is to just keep replacing parts every 5-7 years and stop buying laptops at all. I didn’t want to go used because I run a screen reader, so performance matters a lot more for me as the screen reader is an extra resource drain on everything the computer does. I also can’t solder; some blind people can, but I’m not one of them. So I needed something I could fix without ever touching a soldering iron.
@lynn And before you ask: no, a screen reader shouldn’t be a large resource drain. But unfortunately, the people who write accessibility API’s don’t think about speed the same way people developing graphics libraries and low level visual interface components do.
@fastfinge yeah like in my head, i think about being blind and and i think about what is draining the most on my library, and i think things like "i can turn off the display light, which is my biggest energy hog at the moment" but of course i have no idea how a screen reader uses energy on a laptop (i have only used them on a desktop.)
i think people have moral decisions but it doesn't work intersected with accessibility. like the framework issue you mention and the benefits it gives you, and also AI and how much that can help people with accessibility issues. i don't know the solution here but it is something to think about for sure. thank you for sharing though.
@lynn The primary issue is that on Windows and Linux, the screen reader generally uses an entirely separate “off screen model” containing the accessibility tree etc. The idea is that if the application is blocking, the screen reader can use its off screen model to tell me the state of the interface, without just freezing while it waits for the application to respond with the accessibility information it needs. However, the trade off is that accessibility APIs spend a lot of time refreshing, syncing, updating, and traversing the off screen model. And all of this is usually done in Python or some other interpreted language, or C if you’re really lucky, because few enough people want to work on screen readers as it is; we can’t also ask them to be good systems/low level programmers. And I went with the framework 16 specifically because it has a GPU module that’s more than enough to run image description AI locally. I didn’t need the extra screen LOL. Modern image recognition and OCR models are small enough to run on an iPhone. Someone with a modern computer who is still using cloud models is either making a moral choice, or doesn’t have the technical understanding to set up a local model.
@fastfinge lol the idea of you picking up a 16inch over a 13inch for screen real estate gave me a bit of a chuckle though. that makes sense. i do think local models are quite useful. i have fun writing alt text, but i am just a single person who focuses on certain things that may not be as interesting to some other individual. so i see a lot of use there.
i would love to say i would work on screen reader technology but i am not a good systems engineer either. i do wonder what you use on linux though? i want to try and boot it up and run it sometime.
again thanks again for the input this has turned into something neat instead of just a silly post about framework on my part.
@lynn Right now Orca is the only practical option on Linux. Unfortunately, most of my work happens on Windows because the Linux accessibility stack is far less mature. I’d say accessibility in Linux is about where Windows was in 2005. The primary reason is a lack of funding. It’s much easier for NVAccess (makers of the open source NVDA screen reader for Windows) to get corporate money to do accessibility work. Because most workplaces still use Windows, and legally can’t refuse to employ someone because they’re disabled. The problems on Linux could all be solved with a combination of money, and a charismatic project leader who could convince all of the window managers and desktop environments to implement a shared, modern accessibility API. I know of two people who are willing and capable of doing the work. But they can’t both do the work and also fight for grants and funding, so it just doesn’t happen. I really wish a lot of the EU tech funding and sovereign tech efforts would just set a required amount that must be spent on accessibility work. That would then force them to go find things to fund, rather than force the few overworked people qualified to fix the problem to spend all there time filing grant applications.
