Regarding my music finding problems, an interesting quote from Rick Beato: “It used to be that a lot of the things that were great, were also popular; and that’s not really the case anymore.”

I think I might be missing CDs (compilation discs, for the young people out there). There was so much great music I got to know through these, and I haven't had any equivalent for a while. When I still bought music from Apple Music there were some compilations on there. But now that's gone as well.

I’m also going to refer to money as “survival tokens” from now on.

After a lot of trial and error, I might finally have a decent syncing setup for my atproto integration. There's 4 kinds of data: facets, playlist items, settings & tracks. Each kind that is likely to have over a 100 items is grouped into bundles. Bundles are split into groups depending on the data.

This is something I made possible in Diffuse v4, the playlist item lexicon allows you to match to any data associated with a track (album, year, bpm, etc) Haven’t made any UI for this though. First priority is getting the atproto integration stable, pds rate limits you’re killing me 🫠

If anyone is looking for a web engineer, @tokono.ma (ex-Fission & the author of Diffuse elements.diffuse.sh) currently has availability! EU-based but lots of experience working remotely with folks across the pond github.com/icidasset

A history of Apple's HyperCard ("it was the web before the web"), where it came from, and what happened to it. [kottke.org]

I've been so heads-down on my own work that I haven't really kept up with who's been doing interesting things recently. Let's fix that: 💡🗣️ Good people of Bluesky: who's doing interesting work/experiments/creative projects that deserve a shout out? cc @bmann.ca @chadfowler.com

Maybe my dream of artisanal hardware isn’t impossible after all 🙌

That said, we're getting closer to the goal though. Have LLMs generate working UIs in minutes on top of your user-data and the malleable platform underneath it. Just generated this UI, it just borks the code because regular Claude isn't allowed to fetch code without explicitly linking to it.

Claude really doesn't like fetching Javascript does it... I tried having it analyse nested JS code, but it only fetches URLs you explicitly put in the prompt; so it can't resolve any imports. It also removes any scripts from HTML pages. Why is that?! (not talking about Claude Code, just reg chat)

A little retro on elements.diffuse.sh to appreciate what we have so far: (1) A set of custom DOM elements, aka. framework agnostic web components, that allow for web-based music players, audio derived-data (playlists, metadata, etc) browsers, user-data syncing with various services, among others...

A little retro on elements.diffuse.sh to appreciate what we have so far: (1) A set of custom DOM elements, aka. framework agnostic web components, that allow for web-based music players, audio derived-data (playlists, metadata, etc) browsers, user-data syncing with various services, among others...

@rocksky.app Hi! Would love to implement Rocksky, but hitting some CORS errors. Does the API audioscrobbler.rocksky.app/2.0/ support CORS?

@blaine.bsky.social Thank you sir for another great cheese talk 🙏 Love this framing!

@hilk.eu Thanks for your atmosphereconf presentation, great work 🙏 I did not know you were part of Amatorski! That's so cool, "Come Home" is such a lovely song.

I just saw the awesome Web Tiles talk from ATmosphereConf which demos a little audio player and I now wish I could adapt my thing. Still not sure it's actually possible. I worry that it does too much sandboxing to integrate with various api/oauth flows, but maybe I'm thinking of this the wrong way.

Diffuse has turned into this malleable cooperative software environment and I wonder if this should be its own thing. Not crazy complicated I guess, it’s basically a “software loader” that loads html/js/css and injects it into itself. That code is a set of “features” (logic) and an “interface”.

Good good, the #atmosphereconf fomo is almost over 🫠

We're at 48933 redirect rules on Netlify, this content-addressed versioning system might actually hold up.

Aspire your software to be a flexible toolkit instead of a monolithic know-it-all.

Looking for some contract work to pay the bills, anyone got something? Can do cryptography, crdts, protocol work, frontend and backend coding, or some light design. Bonus points if it’s anything that improves user agency, but since I won’t be able to pay rent very soon I’m not going to be picky.

“We gained so much info but lost all our wisdom” 💥

Very good article!! tonsky.me/blog/fall-of...

I might have found a "user-data storage + web software" setup that I actually like and potentially isn't confusing to non-tech people. I think the problem comes from trying to cram every feature into a single web app. What if instead you compose your software, picking the features you like? 🧵1/4

Looking for some contract work to pay the bills, anyone got something? Can do cryptography, crdts, protocol work, frontend and backend coding, or some light design. Bonus points if it’s anything that improves user agency, but since I won’t be able to pay rent very soon I’m not going to be picky.

Yes please, this system would be amazing. Tax global wealth/assets, not income and all the little stuff. We don’t need no stinking billionaires with a 170 million $ bunker.

CID profiles, the white whale of 2025, has landed! 🐳 More soon but if you've ever been confused about why the same data yields different CIDs, this should help. One million blessings to @lidel.org 🙏🏽 specs.ipfs.tech/ipips/ipip-0...

How do we get to more than just three web engines owned by three US companies? It's a gargantuan question, with no easy or right answer. I've put together a draft report, thinking about it through a very specific approach - please enjoy: Servo Readiness Report webtransitions.org/servo-readin...

This is what some people have been trying to do for at least 10 years. I feel like the hard part is the UX? Also, why is atproto supposedly better in this regard than remotestorage.io which has been around for ages. Even though the former is solely public data. Is it features, or timing?

For the very few people that do still download music and other audio files, where do you keep them? Are you happy with that system; if not, what would the ideal audio file management system look like to you? Also, do you wish you could do more with those files?

I was looking at an older version of the software I'm yet again working on from what, 15 years ago, and I can't believe we're still mostly stuck with the same private blob storage options that can serve as user-data storage. S3-compatible storage + Dropbox was the list: github.com/icidasset/on...

Guilty 🙈 What’s the best way to go about this, without introducing too much centralisation? The PDS flow is so straightforward, the rest is not (I’m an atproto newbie) Also, how do people write/read a large set of records without getting rate-limited by the pds? eg. for importing/exporting data.

Doing this while we wait for private records on atproto. Optional Passkey PRF encryption for individual record properties. Context: I needed some way to hide potentially sensitive authentication details (eg. s3 secret). Only thing I don't like about this is that it's now tied to a specific domain.

Working on cooperative audio playlists in Diffuse, curious what people think of this lexicon. This works a bit different than a CRDT-based approach, so I'm curious how it'll turn out. I'm guessing it does mean you can't reposition other people's items, only your items in respect to other list items

How are atproto apps dealing with conflict resolution? Or is there no such thing basically because everyone uses the firehose, or polling, to get the latest records? Or should I use jetstream?

In the process of adding ATProto to Diffuse v4. One tiny issue though with the OAuth ... it clashes with my system that allows you to pick a version of the software on the website 😩 I don't know the exact redirect URIs in advance, so I can't specify all of them in the metadata json file.

The quality on this is crazy 😱

Oh this might work nicely for non-forced upgrades for software on the web. elements.diffuse.sh/bafybeihj5zj... (See upgrade icon next to the logo on top, might take a few seconds to show up) This then upgrades to the latest version by just redirecting to: elements.diffuse.sh/bafybeihby6n...

help me upvote unstream and get more tech bros to support music artists directly www.producthunt.com/products/uns... #musicsky #fairtrademusic #bandcamp #mirlo #faircamp #qobuz

Not sure if my thinking here is valid, but I'll give it a try anyway. My goal with Diffuse + LLMs is so that people can build a Millennium Falcon model/replica using the provided Lego set, as opposed to building it from scratch with raw materials. (disclaimer: prob obvious but this is an analogy)

Do you like a free tool? Send money. Do you use a tool at work? Convince your employer to sponsor it. Stop marketing OSS as free stuff to your employer, and treat it as a cost for the company. You only endanger the OSS project, if you don't give something in return.

I think I've started to figure out that gentle slope the malleable software essay from @inkandswitch.com talked about (where the user gradually becomes an editor or creator). www.inkandswitch.com/essay/mallea... I'm comparing Diffuse (a set of components and tools) to food, copy need work tho 😅

Can’t we just all stick with distributed compute in order to create & use these LLMs? I mean, wtf with all these new huge data centres. Fuck big tech so much. There’s so much farm land disappearing, that’s gotta be bad for the future. This techno feudalistic capitalism is just … so depressing 😒

Are there any decent local AI setups yet for coding with open-source models? Thinking of models like the ones from @ai2.bsky.social and @nousresearch.com I tried to setup jan.ai on my Linux machine, but after downloading any model it just sits there as if it hasn’t downloaded anything 😬

Are there any decent local AI setups yet for coding with open-source models? Thinking of models like the ones from @ai2.bsky.social and @nousresearch.com I tried to setup jan.ai on my Linux machine, but after downloading any model it just sits there as if it hasn’t downloaded anything 😬

I like the internet handle thing except maybe for the fact that it looks so much like your email address. I suppose if I say my internet handle is @tokono.ma and my email is [email protected] That kind of makes sense. But what if my handle was @steven.tokono.ma and email [email protected]? 🤷‍♂️

Happy Holidays and may all your protocols in 2026 be open. 🍾🥂

Whelp, my domain got blocked, no incoming email 🥲 My fault for picking the top-level domain of Morocco I guess.