Design of This Website (2021)

But that's not quite my point.

The comments helped my realize that I actually had a point I was trying to make. And I believe it's a point worth making, if not exactly a point worth reading... Well, anyway the point is:

_Always write to someone._

    1. Be the person who is listening
    2. and then begin to write
    3. Read what you just created, as the person who is listening

I've been creating web things since 1995, and it always astonishes me, how rarely anyone tests thier end product. Individuals, small business, contracts, major international brand campaigns -- at every level. It's not 100%, and the tech term of art "eat your own dog food" comes from this, so there's the ability to imagine otherwise, but all too often, we just toss it over the wall and hope it lands somewhere.

Two examples that will stick with me forever:

It's 1998, and we're doing a major corporate brand website. The target demographic is low to mid income pet owners, rural and urban alike. It's literally dog food. The design is done in a gorgeous glass high rise office tower, using $10,000 Macintosh computers with 20-inch displays. The target demographic uses Windows 3.1 or Windows 95, at best 640 x 480 VGA 256 colors, often 32-color palette. There is a testing lab closet with some old beater PCs running Mosaic and AOL, but no one goes in there, Windows is scary those machines are junk! I go in there. I try it. The site is a hypnagogic vomit of color on the limited palette display.

It's 1997, and we're building a technology to create web sites that can be downloaded over a slow dialup overnight, then browsed offline. Look up Marimba Castanet. This isn't that product, but that's the idea... The team works for months on this. In an office. In Silicon Valley. With 100 Mb/s network connections. I'm the only person on my team to go home, dial up at 9600 baud, and give it a shot.

We write for many reasons. How do we switch from writing to organize our thoughts, to writing as compassionate communication? Because they are two different things, and I've been all too often doing the one when I think I'm doing the other.

My personal blog! Let me know if you like any of the stuff on it. And what I'm wrong about :)

I'm not sure what you mean by 'lists implemented like divs'. My lists are good old /, as God intended. The fallback fonts and font rendering are also implemented correctly AFAIK: we use a high-quality Adobe FLOSS font, we swap fonts for performance, and we have a full fontstack which falls back through all of it down through system to a last-resort. The analytics is also super-useful for me. I use it all the time for finding broken links, discussions, making design choices about device screen size, etc. I do not regard it as wasted at all. A lot of the WAVE stuff you refer to is things I neither can nor would fix, so gwern.net is never going to be the pinnacle of what you consider accessibility. While none of it is quite as stupid as when accessibility people were telling me to not use hyphens because screenreaders, 20+ years later, still are unable to delete soft hyphens (and this is somehow my fault and I should break hyphenation for 99% of readers), I raise my eyebrow at the claim that justified text is an accessibility problem; and when it flags PDFs, I can only shrug - yes, and I would like a pony too, but that ain't happening either.

I've switched the two navbar divs to s, and added a skip-to-content link. I also ran it through WAVE and fixed a few things. (I did a long time ago but there's been a lot of changes since then. I've added a cron job to check it again every few months.)

> The dialog popups also lack the same usability, they would benefit of clearer aria-labels on the links.

Which aria-labels? There's a ton of them.

I still use archiver-bot etc, they're just not how I do the on-site archives. See https://github.com/gwern/gwern.net/blob/master/build/LinkArc... https://github.com/gwern/gwern.net/blob/master/build/linkArc... for that.

The quick summary is that PDFs are automatically downloaded, hosted locally, and links rewritten to the local PDF; other URLs, after a delay, call the CLI version of https://github.com/gildas-lormeau/SingleFile to run headless Chrome to dump a snapshot, which are manually reviewed by myself & improved as necessary, and then links get rewritten to the snapshot HTML. They get some no-crawl HTTP headers and robots.txt exclusions to try to reduce copyright trouble.