I do like Gemini but still question whether the development of a protocol was necessary. Couldn't you replace it with webrings and a crawler to index them?
Gemini is _incredible_ for accessibility, because there's no disconnect between what's displayed and what a blind person can consume, and it's not complicated by layout things or developers who failed to account for the fact that maybe you need huge fonts.
Also, the moment you use normal HTML someone is going to stick an image in it. Now, i 100% support this idea. I've never embraced Gemini because the idea that the web is better with NO images is just absurd to me. I don't care how good of a writer, no textual description of something you've seen will come close to conveying what a photo can, and if you're trying to describe abstract art you're doomed to failure.
Which is probably a relatively small subset of images on the web.
It depends on how you use it. If you want to make ASCII tables, then a screen reader will not have an idea of how to navigate the table. However a simple HTML <table> is well understood by screen-readers. I would contend that properly structured HTML documents are far better for accessibility.
For example, the lack of simple text styling makes Gemini significantly less accessible than it could've been. Because now a screenreader will say something like "Significantly asterisk less asterisk accessible" instead of "Significantly emphasis less accessible".
I don't think images can have alt-text in Gemini so that blind users could get info from there.
Etc.
Edit: someone else has a much better answer: https://news.ycombinator.com/item?id=37052729
Gemini does not prevent anyone to put images. What it doesn't provide is the semantic separation between a link to an image and embedding that image into the page. The choice is up to the reader. Actually the default setting of the lagrange gemini browser is to download and show the images by default.
For strictly descriptive writing, maybe. But novels remain popular even in 2023 and while they mostly do not have illustrations of any kind, good ones paint some very vivid pictures.
Incorrect. It is actually bad for accessibility.
Because gemtext was so deliberately hobbled in functionality, it is actively hostile to users with accessibility needs.
No tables means people have to do things with ASCII art which makes them incomprehensible to screen readers.
Standalone images (i.e. not online with the content) means you cannot provide any textual clues or context to what the image is about - you just get the image on its own. And the pervasive use of ASCII art "images" obviously is terrible.
No way to specify descriptive labels for links.
No way to visually-hide content so that it is only visible to screen readers.
No way to use aria-style tags to semantically mark up the different sections of a page (this is the navigation are, this is the gemlog listing etc)
A bit web-centric but I think important considering the ASCII table issue, no way to use JavaScript to provide what is known as "managed focus" handlers to allow navigation of cells/content areas using cursors+space+enter etc.
I can't remember, but I don't know if it even has in-page anchors?
They should have made gemtext slightly more sophisticated - inline images, proper tables, and basic visual styling would solve a lot of the a11y issues.
It is a shame. I got into Gemini a bit a while ago (even wrote a basic library for deno @ https://github.com/matt1/deno-gemini (essentially a fork of my own https://github.com/matt1/deno-gopher) and started work on a GUI client). I started to experiment with a totally-non-spec "style" spec that my client would support - essentially a line-by-line thing similar to CSS to apply the most basic of styles (colour, font weight, size etc) to gemtext lines, and which would be loaded from a well-known file (similar to robots.txt etc - compatible clients would load styles.gem or whatever).
But ultimately I was scared off by the short-comings of gemtext, and the staunch refusal to consider any changes or improvements. Which is fine, it is their spec and they don't have to change anything, but its why I left.
An all new protocol stack focused on accessibility could be pretty cool. Flip the problem. Instead of layering accessibility on top, make it the foundation.
We might learn that an accessibility-focus stack proves better for all use cases. Kinda like accessible consumer goods are also better for every one.
Something fun to think about.
The idea that accessibility means everyone's experience is nerfed into the ground just seems fundamentally wrong. We need to do much better than that and enable a richer experience that still is accessible. There's really no reason that can't be the case.