Pilot with a few artists?

[zanellia]

@icidasset diffuse seems really cool and I love the idea of having a way for artists to distribute their music directly to fans without any third party invoived 🔥 Do you think it would be possible to run a small pilot with a few artists (10-50) who anyway make very little money from streaming platforms?

Wouldn't the features discussed in #440 already be enough? I feel like diffuse already satisfies most of them? To summarize, I think a lot of people would be interested in a solution with

• decentralized remote storage owned by the artist
• serverless web app
• a diffuse catalogue where you can select artists (i.e. diffuse sources) and listen to their music
• for now it can be for free since anyway they don't make money on streaming platforms, that way more fans will be attracted to diffuse (I do think that, in the long run, artists need to be paid in a fair way, but first an alternative to streaming platforms needs to be created?)

[icidasset]

Thanks again ☺️ I'd love to make this possible! Not entirely sure Diffuse is the right fit here, but can't hurt to experiment.

Just for reference, my interpretation of this idea:

If by catalogue you mean a bunch of "Diffuse sources" in the form of Cloudflare R2 buckets created by you or the artists, then, definitely 💯 With the "adding sources using query parameter" feature you can then add one or more sources at once, by creating a link which you can put anywhere.

I just tested that functionality and it still seems to be working.
Here's an example of the idea layed out in the quote above:

https://diffuse.sh?source=%7B%22kind%22%3A%22amazon_s3%22%2C%22data%22%3A%7B%22name%22%3A%22Eddie%20Vedder%22%2C%22accessKey%22%3A%22c86e133f0e12acdf81d66f5861c3f16d%22%2C%22bucketName%22%3A%22music%22%2C%22host%22%3A%22https%3A%2F%2F67d0565dd63db5e7e864b6cb784bd912.r2.cloudflarestorage.com%22%2C%22region%22%3A%22auto%22%2C%22secretKey%22%3A%2228abc00d8b5172483a6982c33b8d030624ad3f4b523c543f5d24537a4db5514e%22%7D%7D

That should add a source named 'Eddie Vedder' to your Diffuse.
(you can replace https://diffuse.sh with https://nightly.diffuse.sh, or whatever instance you want)
(you can also specify a page, eg. if you'd like to open the sources page instead)

I feel like diffuse already satisfies most of them?

I think so. Decentralised storage ✅ Serverless ✅ free/low-cost ✅

Like mentioned earlier in #440, the "add source via url" feature might need some tweaking for this use case. Couple thoughts:

• What if you clicked the link before and added the source already? It'll currently add the source again, which is not what we want. You'll probably won't notice if the "hide duplicates" option is on (which is the default I think).
• What is the expected behaviour after you click one of those links? Is there something specific you expect to see in the Diffuse app? If you have "group by processing date" on, it'll show up on top (of the tracks page), but that's not on by default.
• Not sure if the stuff in the URL needs a different encoding to hide some of the details (ie. make it nicer to look at and shorter?)

[icidasset]

Here's how you make such a link, you can copy-paste this in the browser developer console (on any page):

json = encodeURIComponent(JSON.stringify({
  kind: "amazon_s3",
  data: {
    name: "Source/Artist Name",
    
    accessKey: "c86e133f0e12...1df5861c3f16d",
    bucketName: "music",
    host: "https://67d0565dd6...4b6cb784bd912.r2.cloudflarestorage.com",
    region: "auto",
    secretKey: "28abc00d8b5172...24537a4db5514e",
  }
}));

`https://diffuse.sh?source=${json}`

Then copy the resulting link.

[zanellia]

If by catalogue you mean a bunch of "Diffuse sources" in the form of Cloudflare R2 buckets created by you or the artists, then, definitely 💯 With the "adding sources using query parameter" feature you can then add one or more sources at once, by creating a link which you can put anywhere.

Yes, exactly!

I just tested that functionality and it still seems to be working.
Here's an example of the idea layed out in the quote above:

https://diffuse.sh?source=%7B%22kind%22%3A%22amazon_s3%22%2C%22data%22%3A%7B%22name%22%3A%22Eddie%20Vedder%22%2C%22accessKey%22%3A%22c86e133f0e12acdf81d66f5861c3f16d%22%2C%22bucketName%22%3A%22music%22%2C%22host%22%3A%22https%3A%2F%2F67d0565dd63db5e7e864b6cb784bd912.r2.cloudflarestorage.com%22%2C%22region%22%3A%22auto%22%2C%22secretKey%22%3A%2228abc00d8b5172483a6982c33b8d030624ad3f4b523c543f5d24537a4db5514e%22%7D%7D

That should add a source named 'Eddie Vedder' to your Diffuse.
(you can replace https://diffuse.sh/ with https://nightly.diffuse.sh/, or whatever instance you want)
(you can also specify a page, eg. if you'd like to open the sources page instead)

This is great 🔥 🔥 I just tested it too.

Like mentioned earlier in #440, the "add source via url" feature might need some tweaking for this use case. Couple thoughts:

• What if you clicked the link before and added the source already? It'll currently add the source again, which is not what we want. You'll probably won't notice if the "hide duplicates" option is on (which is the default I think).
• What is the expected behaviour after you click one of those links? Is there something specific you expect to see in the Diffuse app? If you have "group by processing date" on, it'll show up on top (of the tracks page), but that's not on by default.
• Not sure if the stuff in the URL needs a different encoding to hide some of the details (ie. make it nicer to look at and shorter?)

These things seem very relevant, but I guess they can be figured out later on?

[icidasset]

Awesome, glad it works! Yeah that stuff could be figured out later, but I was very curious to what you think about the second item. I mighttt enable “group by processing date” by default (not sure if you’ve even found that in the UI, it’s a bit hidden)

[icidasset]

I do think that, in the long run, artists need to be paid in a fair way, but first an alternative to streaming platforms needs to be created?

Kind of? The way I see it is that everyone should go back to buying music. Though there are some rough edges there which I'm not entirely sure about, like, what about artists who passed away, should their music become public domain?

Anyhow, that aside, I like that we can buy music and download it DRM-free on a bunch of individual stores, eg. bandcamp, subvert.fm, etc. And the ability for artists to sell their music on their own by just setting up a website. The problem then however becomes, I now own music on a bunch of different platforms (or no platform), how do I gather all my music in one place? I can probably stream it on each individual platform/website, but I'd like to have a single place where it's all gathered.

This seems especially difficult for non-tech folks. Sure, downloading music files is easy, but what then, where do I put those files?

I wish we could figure out where to go from there.

---

Regarding the approach we talked about above, Cloudflare R2 + Diffuse; I do like that the artist doesn't need to pay for the bandwidth. It doesn't encourage people to buy the music though. Not sure how we solve that.

[zanellia]

The way I see it is that everyone should go back to buying music.

I agree 100%. In a way we are still "buying" music, and the revenues associated with streams are not that low. The problem is that this money does not go to the artists. If we could just go back to buying physical supports or files that would be awesome, but we are so used to the fact that every album ever published should be available to us in a couple of seconds and at a low cost that I feel like any realizable transition to a healthier music industry might have to go through streaming at some point. Once a considerable portion of streaming takes place in a safe place (and to me the safest place for artists is something like diffuse, i.e., no third parties involved), then we can probably rethink monetization paradigms: donate, buy after a certain number of streams, pay per stream, etc. or a combination of all the previous. But first and foremost streaming needs to take place in a fundamentally different context?

Anyhow, that aside, I like that we can buy music and download it DRM-free on a bunch of individual stores, eg. bandcamp, subvert.fm, etc. And the ability for artists to sell their music on their own by just setting up a website. The problem then however becomes, I now own music on a bunch of different platforms (or no platform), how do I gather all my music in one place? I can probably stream it on each individual platform/website, but I'd like to have a single place where it's all gathered.

That's definitely super cool, but I feel like too few people buy music on those platforms at the moment? Most people are addicted to a Spotify-like experience. That's why I think something like diffuse can make an even stronger impact. In the end it's a matter of removing (unnecessary) third parties from the picture and redistribute the revenues? If Tidal or other managed to make deals with labels after Spotify even a third-party-less platform could do that? In that case, Spotify's revenues (~30%?) would go directly to the artists. And that's still a bad deal probably, but it's a deal that labels would do?

Regarding the approach we talked about above, Cloudflare R2 + Diffuse; I do like that the artist doesn't need to pay for the bandwidth. It doesn't encourage people to buy the music though. Not sure how we solve that.

That's already something very valuable in my opinion. Once a technological alternative exists, maybe ways to make people buy music can be found? Just daydreaming :p But hey diffuse is exciting!

[icidasset]

I'd like to listen to music before I buy it, in that regard streaming is indispensable for me as well. I think that's how Bandcamp does it, allows you to stream a bunch of times and then you need to buy it. But yeah, you're right, most people opt for the streaming experience.

In the end it's a matter of removing (unnecessary) third parties from the picture and redistribute the revenues?

Maybe yeah, it'll be much easier once there's a protocol or software that enables this. I feel like we're slowly heading in that direction, eg. with the ActivityPub (Mastodon) and AT Protocol (Bluesky) protocols for public content discovery. Hopefully there'll come an equivalent for music streaming? It's also interesting to see how new ways of monetisation arise with the introduction of protocols (eg. https://www.techticia.com/2025/02/graze-pioneering-monetization-on.html)

I suppose there's a distinction here between people who'd like to curate their music collection and people who prefer others to do that for them? I'm the former, I'll never opt for a way to listen to music where the music could disappear at any time. Most people are the latter then I guess? Though I guess Spotify is partially curated. So much convenience in not having to manage music files.

[zanellia]

Maybe yeah, it'll be much easier once there's a protocol or software that enables this. I feel like we're slowly heading in that direction, eg. with the ActivityPub (Mastodon) and AT Protocol (Bluesky) protocols for public content discovery. Hopefully there'll come an equivalent for music streaming? It's also interesting to see how new ways of monetisation arise with the introduction of protocols (eg. https://www.techticia.com/2025/02/graze-pioneering-monetization-on.html)

Thanks for the pointers, this is very interesting.

I suppose there's a distinction here between people who'd like to curate their music collection and people who prefer others to do that for them? I'm the former, I'll never opt for a way to listen to music where the music could disappear at any time. Most people are the latter then I guess? Though I guess Spotify is partially curated. So much convenience in not having to manage music files.

Unfortunately, I feel like the vast majority of the people indeed belong to the second category at the moment. But that's also the case because of the way streaming is structured at the moment and because of the distorted perception of music as something that "can be made almost for free" and can be accessed anytime, everywhere at low cost. The moment these perception will go back to matching a bit more reality, I think more people will care more about what music they listen to - fingers crossed.

[zanellia]

Also, streaming could really be just a way to monetize listens before a fan buys the album? After X streams? My point is that most people would not buy records because of the current dominant perception of music and they need to be "guided" back to a meaningful perception?

[zanellia]

Regarding monetizaztion, is there an obvious way to track access to files on R2?

[icidasset]

I think we can also generate access keys that work on specific objects in buckets. And I suppose this is something you could do without a server, but I don’t see a way to do this using the Cloudflare dashboard.

So using that, you could, for example, pay an artist for whatever tracks and then the artist could give you access to those items that all live in the artist’s r2 bucket. And this could be something done manually (without a server).

I suppose you could make a static website or app that is given privately to the artist which could do that for them.

Docs: https://developers.cloudflare.com/r2/api/s3/tokens/#temporary-access-credentials
It says temporary here, but I suppose you could make it non-temporary somehow too.

[zanellia]

That's pretty cool. At the same time, do you think it would be possible to do something in the direction of streaming? Since that's what people are used to, it would make much easier to attract people to Diffuse, no?

[zanellia]

It seems that Cloudflare has Instant Logs that log access requests https://developers.cloudflare.com/logs/instant-logs/, but it's a paid feature.

[icidasset]

Depends on what you mean exactly with "streaming". You mean like, paying artists based on how much their music was played? If so... I'm not sure, that's a more difficult problem to solve. I'm not all familiar how those streaming platforms work, how do you determine if a request is for playing a track or for something else (eg. downloading embedded metadata or album art). Also, what if you cache/download a track in Diffuse, then it streams it locally, not from Cloudflare 🤷 So yeah, not sure how to best approach that problem.

Not in a hurry to attract more people to Diffuse 😄 This is just something I've been using personally for the last ~12 years (there have been other iterations of this idea before this one). And if other people can benefit from this, or pay me for it, then that's great.

[zanellia]

Not in a hurry to attract more people to Diffuse 😄 This is just something I've been using personally for the last ~12 years (there have been other iterations of this idea before this one). And if other people can benefit from this, or pay me for it, then that's great.

haha yeah I get it :) At the same time, it's not entirely unreasonable to think that a "healthy" solution in the near future would/should be based on something like Diffuse? With "something like Diffuse" I really mean decentralized, no third parties involved. I think that's a pretty strong point.