I ran a quick test, averaged 268 milliseconds. Yep, that's "essentially immediately.

We forget how fast the web can be, because of all the bloated sites out there.

to go with sibling, i ran a few tests, and 73% of the site is jscript, which contributes to half the load time. The tracker uses 40% of the load time.

Remember when sites used to serve requests based on the browser identifier rather than trying to contemporaneously display and format text with jscript?

This site is "fast" - it fully loads in 1 second, sure. one of my own purely nodejs + jscript site also finishes rendering content in 750ms, with 480ms of that spent decrypting into the drag-resize container. Meanwhile, Dan Luu's site completes rendering an entire blog post in 300ms, first byte to fully rendered.

this doesn't mean anything, really - i just expected it to waterfall differently based on "how fast it felt".

> Remember when sites used to serve requests based on the browser identifier rather than trying to contemporaneously display and format text with jscript?

I remember when my window to the world was a 56k modem, which took its sweet time to load the page I wanted.

i've had to use 56k in the last ten years for a primary connection. I used pfsense to cache and compress as much as possible, but more interestingly to me, i use 3K with PACTOR radio modems way more often. A "you have no new mail" handshake and chatter takes minutes[1].

This all ignores that spoken word over a mediocre connection (anything except actual copper landlines, in my - get off my lawn - opinion) is way lower bandwidth than even early modems in the 300-2400 baud range. Realistically, my favorite method of communication is to actually have a spoken word conversation with people, in person. This is untenable for my friends that have moved away from me, and me from other friends. That in mind, the matrix people have worked on and have published extensions to the matrix ecosystem that allow sub-kilobyte messages to pass on the network, fully encrypted and fully fledged messages. They are targeting sub 4000 baud connections (in general, i think their actual goal is much slower transmission than that).

I don't have the time or inclination to try and run a matrix "node" on ham bands, but i do know it is technically possible. And that's almost good enough for me.

For your edification though, the last time i used only a 56k connection was back in the very late 90s, and i spent a lot of time on IRC rather than "the world wide web"

[1] https://www.youtube.com/watch?v=DA6JRB560eM A youtube video shot handheld of a radio and a modem checking email wireless-like. The modem, i've heard, is around $800. I got it for free to participate in emergency communications and teach other people how to use it.

We’re targetting 75-100bps for ultra-low-bandwidth Matrix for HF. The main missing piece has been sync v3, but this is almost done now.

I don't know if you will see this, or if you're using the "royal we", but if there's anything i can do to test using ultra-low bandwidth matrix over packet/PACTOR radio on HF; i'd be very interested.

Unfortunately, none of it can be encrypted (not with my license, at least) - but for sending telemetry or simple "i'm still alive" messages i think it would be pretty cool. Not having to write custom tooling off ham radio "tools" like fldigi, but rather have a bot sitting somewhere else monitoring the channel seems like a huge, huge win.

we = matrix core team. totally understood that crypto is not an option, which is arguably good news for ultra low bandwidth: compresses better and no aead headers :)

there should be another round of work on this fairly soon driven by our p2p work - #p2p:matrix.org is probably the best place to follow along, likewise arewep2pyet.com