Not sure how best to approach this really but I was thinking about discoverability on the Fediverse and taking notice of a lot of the streaming articles coming out of late (plus Liz Pelly’s book). At the same time, I’m also seeing interesting visual platforms popping up like https://www.soot.com and https://rooms.xyz (both backed by private investors though ) and wondering about what it would take to pull people away from streaming, if knowing “it’s bad” isn’t enough. It sort of reminded me of things I liked about the early internet and sites with clickable Flash-based comics and online places such as BowieNet. So, without this turning into a Grandpa Simpson-style ramble, I feel like there needs to be a site that takes the best of the links below and makes a strong, fun visual space where musicians can be found randomly, either by something like location, or by creating clusters of artists or maps of listener recommendations. Mirlo have just started testing out linking musicians to labels (without it being a paid feature like Bandcamp) and I think that’s a really exciting start plus the Fediwall from Indie Beat and @limebar is also really cool (in the last day there is also a live app hoping to launch called Subjam which is aiming to link to music venues and their communities) and I was curious if there’s a way of building on that and even linking a few different ideas together? Here’s a few examples of things that are/were slightly more offbeat ways of discovering things (aside from the aforementioned Soot and Rooms):
Ghostly’s now defunct app, where you could select from their catalogue based on “Mood” (but it was actually a colour wheel, because I downloaded it at the time).
This genre map where you can click around then it gives examples in a playlist.
Every Noise at Once (although I appreciate we don’t necessarily want to fall under genres pushed as part of an end of year ad campaign cough cough Spotify Wrapped).
Radio Garden - imagine this but with musicians and showing lines that indicate links between them. It’s slightly infuriating that you can collaborate with your friends across your projects but on streaming, your musical projects aren’t shown as being related in any way. It’d be a neat way of visualising that context and encouraging people to discover how different people are interconnected (“interdependent”).
A Number from the Ghost is one person’s site but what if there was something showing fedi musicians videos in this kind of way? Or as floating images in a “constellation” (to use @Alex’s terminology) that are scattered around and clickable.
Obviously some are more complex or resource-intensive than others, but it seems like there might be some weirder ideas that could help people get found in the same way you might stumble into a record shop and find something bizarre? It’s hard to articulate, so I’m going to end the post here and let other people chime in.
Edited to add in Emma Warren’s book, which might also have started some of this.
(This isn’t very “discovery” focused, but reminded me of rooms.xyz: https://www.cyberdoof.com/ – found this via Golden Shrimp Guild’s “SynthOn” Twitch event, the person who coded it got involved. Found it really fun to hang out in).
I’d be interested to be a part of this. I think algorithmic recommendation systems tend to completely hide weird music like mine. A discovery method based on real communities, & shared aesthetics sounds good to me!
Oh yes! This is absolutely the kind of thing! I’ll have a better play around when I get home. This is probably closer to the BowieNet idea too (with a pleasant sprinkling of retro point and click games).
Maybe it’s two separate points as well - a way of organising music communities for fun, randomised discovery by listeners AND an online hangout (whether it’s for folks not gigging due to health/cost or just as an additional “live” experience on top of gigging normally). It’s a shame that Metaverse talk has soiled talk of virtual spaces a little and that Spotify Wrapped has crept in on music data visualisation to a certain degree but they’re both concepts that I think Fedi could do better and organised around topics or areas that artists actually want. I have zero technical expertise for this kind of thing but do have design experience and plenty of research ideas to contribute to a group if a bunch of us are keen on exploring any of this.
…I know, I know, but there are algorithms and then there are algorithms.
I am a fan of Plex’s “sonic analysis” apart from it being closed source and paid. The main thing I like is that it does not care whether tracks are indie or major label – if two tracks have acoustic similarity it can find them and play them together in a playlist. The best thing is the sonic journey - pick few tracks and it fills in progressively similar track between them. If the open source community could replicate this capability it would unlock an automatic “Soot” style discoverability / spatial relationship kind of world – forget genre, what does it sound like? It will happily play a Prince track and then Mel, for example, and rightly so.
It also enables “mood based” and other affinities which some of the links above are doing.
I think there are real challenges in that tho… compute and storage intensive and how do you reliably crowd-source and share the data while making sure it can’t be poisoned. I think this is why AcousticBrainz shut down. Plex does not crowd-source for this reason as well-- when you turn the feature on it fires up your CPU for days computing those signatures locally (depending on how big your music collection is). Wasteful really since somebody somewhere has definitely already computed the data for Stairway to Heaven.
If you had that data you could seemingly easily build those UIs from it.
It would awesome for radio stations like TIBR as well… a station could pick two random tracks from the portfolio, generate a 10 song playlist journey and ensure a smooth ride for listeners, before picking two new tracks and repeating… some similarity, some variety, without a constant DJ or doing playlist management, which frankly, burned out the hamster wheel at RFF, I think.
Goodness gracious! An algorithm that actually agrees with some of my reviews that I am somewhat Prince-adjacent! That’s good to hear… what a relief lolll
Yeah, it makes sense that it being resource-intensive would massively put anyone off. Need to do some more reading on all of these places being mentioned as I haven’t poked around with MusicBrainz since 2013, when I saw it mentioned here:
Interesting reading the updates at the bottom of Every Noise at Once in relation to discoverability too. Glenn should join the Fediverse!
Edit: he is on the Fediverse and so is Every Noise but neither account appears to be active right now.
There are many videos on this channel just posted.
Thinking of this being a possible building block using open web tech to achieve the kinds of things you are talking about. Immersive, hyperlinked, etc.
MusicBrainz was cited before but this would be a fantastic data-viz project based on their data ! Anna’s Archive recently put up a call for a paid project like that, explicitely to help them in their fight for piracy.
Also, going back to the first MakerTube video around Matrix rooms, @limebar, I can totally see this as a way of displaying a FediWall or something like NHAM, but I might be misunderstanding its potential use? I clicked on the A Number from the Ghost site again too and under the “what is this?” link, you can click through to loads more sites using three.js. This electronic music site is mesmerising.
Just chiming in quickly to say that I think the more artists can be connected TO something the better. Label, location, story, technology, famous artists etc. the easier it is to get into them.
So I think this is somethingg that discovery could use in some way.
F.ex. I discovered synth pioneers like Suzanne Ciani and Morton Subotnick because of Buchla synths, and I discovered Yann Tieresen because of his Youtube videos.
I keep going back to the thought that many music makers are going solo and doing a lot of non-musical work themselves that traditionally a label would do. But labels also had to reinvent themselves, and just like with music distribution platforms, some of them also aim to be social, fair, cooperative, even artist-run.
I miss these labels in TSMN, or at least discussion about them.
I also think that much of the function of a label could be recreated in different ways. Like webrings, collectives, some sort of shared social media account or blog etc.
Not saying it is easy though as people needs to organise and agree.
Wow. What a collection of links! I could look at that stuff ALL DAY AND NIGHT.
As for the potential use… I may be way wrong about this… but… the XR Fragments video I posted was less about “Matrix” and more about an open web toolset. As I understand it, XR fragments are a way to pass 3D spatial data openly between resources to collectively define an interactive space. So… looking at that link with all of the three.js sites you posted – imagine they could be interlinked in some way and coexist in the same space?
I think this is what XR fragments, as a tech, is trying to do.
So I was thinking of it as a potential building block for a universe of interrelated music sites… a space for music discovery where users can hop from place to place by some kind of affinity.
Also, I am entirely full of shit. I don’t know for sure if this is what it means to do but I think so…
If that “some kind of affinity” could be something like the similarity cloud site you sent or the plex sonic analysis type data etc… that could be great, or could be artists linking to one another…
Yeah, you’re exactly right in terms of potential benefit to fedi folks (and I can picture it), but I have zero technical expertise to even begin describing how to do it (aside from proving examples of stuff that might be close)! To use that “constellation” analogy again and maybe referencing something like Soot, the first “cluster” visual could be something a general Fediverse overview, with offshoots for musicians, labels, radio or whatever, and when you click on the “radio” node, you see Indie Beat, Audio Interface and NHAM Mixtape grouped together, or if you click “musicians”, you can search by genre, location or by affinity (something like artists who have collaborated or remixed one another, or grouping by CC and all right reserved even - there’s so many parameters it could work on).
) and wondering about what it would take to pull people away from streaming, if knowing “it’s bad” isn’t enough. It sort of reminded me of things I liked about the early internet and sites with clickable Flash-based comics and online places such as BowieNet. So, without this turning into a Grandpa Simpson-style ramble, I feel like there needs to be a site that takes the best of the links below and makes a strong, fun visual space where musicians can be found randomly, either by something like location, or by creating clusters of artists or maps of listener recommendations. Mirlo have just started testing out linking musicians to labels (without it being a paid feature like Bandcamp) and I think that’s a really exciting start plus the Fediwall from Indie Beat and @limebar is also really cool (in the last day there is also a live app hoping to launch called Subjam which is aiming to link to music venues and their communities) and I was curious if there’s a way of building on that and even linking a few different ideas together? Here’s a few examples of things that are/were slightly more offbeat ways of discovering things (aside from the aforementioned Soot and Rooms):
I don’t know for sure if this is what it means to do but I think so…