Speaking as someone who was born blind Braille may not be worth using if you went blind later in life. I was forced to use it heavily until about 10 or 11 in spite of the fact that I wanted to use text to speech since it's quicker. It was totally worth it since it gave me a basic literacy foundation that enables me to type out sentences like these and be able to spell well enough that less then 1/5 of everything I type has to be corrected with Spellcheck. I have an 18-year-old braille display I got in school when I started programming more. It still works but I only use it once a month if that. I do everything with text to speech. The only time it was really useful was when I was doing assembly programming and had to look at hex values in a debugger. If you go blind later in life it may be pretty easy to avoid learning anything but enough braille to read signs for room numbers and bathrooms and do just fine.

Wouldn't a large braille display be very useful for getting an overview of something? When consuming a big chunk of material, audio is very convenient, but being "linear" seems to make it quite impractical when you need to jump back and forth, or get an overview.

I'm particularly bad at piecing together a larger context from smaller pieces, so maybe it's easier for others, especially if they are forced to?

Large Braille displays basically don't exist, they'd be too expensive.

Navigation with a screen reader isn't just next item / previous item, though. There are hotkeys to jump between headings, links and so on. That's why good markup is so important for accessibility.

No sure, I didn't mean that navigation itself was hard, more that it's hard to know where to navigate to. If you can see, you can quickly scan a page and get some sort of understanding of what is where, but at least with audio that's very hard.

It seems that it should be possible to make cheap full screen braille displays, I suppose it's just that the volumes are not big enough to drive down prices.

The issue is that to move 80x24 dots independently you need 80*24 physical actuators (or an enormous mechanical multiplexer), which is very expensive. Industrial looms and knitting machines have that but nothing much else does. However, the technology to solve computer braille without needing width*height mechanical actuators actually exists - the fluidistor, or liquid transistor - but is only just now getting serious development, and it's mostly in haptic feedback labs.

If you want 8024 characters, you actually need at least 8024*6 dots, each Braille character, also called a cell when Braille displays are concerned, can consist of up to six dots, eight in computer Braille.

In my experience, anyone with any kind of physics background thinks that Braille displays could be radically cheaper.

Anyone who spends a couple hours on researching the problem thinks that the price could be improved, maybe by half, but that the problem isn't as easy as it seems.

Anyone who tries building one eventually arrives at the conclusion that making it cheaper requires a bunch of tradeoffs which they're not willing to make.

As far as I know, the central problem off Braille Display engineering isn't making the dots, but putting them so close to each other. If the gap between dots would be half an inch, let's say, a braille display would probably be a hundred-dollar affair. It would be completely unusable for any kind of reading, though.

Orbit Research managed to find a way to make Braille displays somewhat cheaper, which is now incorporated into the Orbit Reader 20 and 40, but that technology has a much slower refresh rate than traditional Braille displays and causes them to be much louder.