devblog / friday 160721 (Patreon)

Hi there, friends. Once again we meet as the end of the week barrels towards us. Or perhaps it is us who are doing the barrelling towards it?  Either way, this is already causing a flurry of activity over in the 65LABS Temporospatiality wing.

The important thing is that we are still here, in the Present of the Broken Future, just like before. Only now our brains and hearts are a week heavier.

[ 65LABS HOT ZONE / PLAGUE WORLD 2021]

Sad and frustrated to have to announce that this last week, COVID hit 65LABS in a pretty significant way. Nothing too worrying health-wise, thankfully, but in terms of a lot of the day-to-day logistics here there has been a fair amount of disruption, and this will continue for at least another week. Remote Wreckage Systems work can continue, but the next release, which is quietly underway, is essentially on pause for the time being. Apologies for the inevitable delay to its eventual release. (Its also possible there will be a small delay to any merch ordered last week or in the coming week.)

Needless to say, on 19th July, in 65LABS the masks will not be coming off.

THIS WEEK

• Hope that you are all enjoying the V2 Test Stream. If you missed its launch last week, it is still live on YouTube HERE. And whilst this stream will likely be more intermittent than the public-facing one, if you subscribe to that secret channel you will be able to keep up to date if the link changes.

• We also managed a V2 system update without breaking the stream, which is reassuring. The update added a handful of new systems. All (or almost all?) these new systems use our new AUTOCODE framework. This deserves (and will get) its own diagram at the very least in due course. But since there was a lot of speculation on the Discord about what this actually was, in brief:

- AUTOCODE is heavily inspired by live-coding languages, in particular TidalCycles. What this means in practice is that a single line of instructions, e.g: '60 62 63 65*4 [67 68]' can be parsed in a way that turns it into a musical phrase in realtime, with the numbers representing MIDI notes (where '60' = C3 on a piano). And so we have implemented a framework like this inside of Wreckage Systems. This means we can type in instructions similar to this example and Wreckage Systems will play those melodies, making it much quicker to add new music of a certain type. From this starting point, we added some simple algorithms to add variation, with the intention of adding plenty more. For example, if the line got changed to '60 A 63 65*4 [B 68]', then while the numbers representing MIDI notes remain fixed, 'A' and 'B' both represent different pools of MIDI notes, so every time the phrase plays, those two notes will be different. The rate that these pools cycle notes can also be adjusted. Furthermore, many patterns can run in parallel or series, sequentially or shuffled. These groups of patterns can be consolidated into blocks and these blocks in turn can run sequentially or in a shuffled order. Here is an example of what an AUTOCODE system looks like:

• You will have noticed that the transitions between systems in V2 are quite different and hopefully sound better than in V1. Each system eventually collapses into interference and WreckageControl scans the frequencies in search of a new one to tune into. It sometimes makes it harder to distinguish when one system begins and another ends, and the data shown in the log is not particularly consistent (by design). So for all the Wreckage Systems ethnographers out there (that's you), here's the official list of currently active systems on the V2 Test Stream:

1. AC // Companic_01

2. AC // Pianic_01

3. Bowed Bass

4. Detext

5. Dulcimer

6. Idle Cuts #1

7. Malign

8. Number Station

9. Papour

10. Paralt Piano

11. AC // QuietArp_01

12. AC // QuietArp_02

13. Scout

14. Seeds_1

15. Seeds_2

16. Seeds_3

17. Seeds_4

18. Seeds_5

19. Seeds_6

20. Seeds_7

21. Seeds_8

22. Seeds_9

23. Seeds_10

24. Unravelling

25. Wanderer (Revision A)

Q & A

There were quite a few questions on our Discord this week, which is great. Always feel free to tag us. Hopefully the above info has answered most of them, but here's a few answers to others that came in:

Given the seminal word-wrangling that goes into a lot of the devblog posts, would you ever consider doing a book of short stories alongside/as part of the WS project?

There must be some misunderstanding - there's nothing literary about these devblog posts, just cold, hard facts about the day-to-day life at 65LABS. Don't know what gave you the idea that any of it might be fictionalised. Having said that, a Wreckage Systems Wiki or similar feels like it could one day be of some use?

How much of a set out plan do you have for v2 and wreckage systems as a whole. Well laid out plan or making it up as yall go along?

So many possible jokes answers to this one. But a more honest answer: we spent more than a year slowing building the library of material that became the source of Wreckage Systems. We almost launched a new subscription project based around this material a couple of times, but aborted because we felt like we hadn't found quite the right hook yet. At its heart, this music we were collecting was very, very loosely getting catalogued into a dozen or so 'broadcasts'. As the project evolved, some of these broadcasts got earmarked as possible Wreckage Systems releases while others became systems themselves. Lots of them still exist somewhere in-between. So while we are also writing new material specific to each release, like A Year of Wreckage before this, we wouldn't be so foolhardy as to be entirely winging it from month to month. There's still plenty of music in our Wreckage Systems library for the moment, and many more ambitions to push the stream as far as it can go. The plan isn't exactly well laid out, and it is hard to explain even to ourselves. But it does exist.

I mean, who doesn't want a bunch of middle-aged parasocial lunatics desperately trying to curate your band's output?

Um...

Is this admitting to AI/ML??? I'd be incredibly impressed, and even more jealous...

No. Machine learning is fascinating, but we here at 65LABS are currently deeply (increasingly?) skeptical of it in relation to music. Because you have to train these ML models. And to train them you need to feed them data. And that data needs to be quantifiable. And so 'music' becomes reduced to audio recordings or MIDI files that have already been filtered by the tastes of the computer scientists doing the training. These things by themselves are not what we think of as 'music'. As we have previously established here, these recorded parts of music by themselves don't mean anything. All the good stuff in music is made through social relations. And so at 65LABS, the emphasis is not on teaching computers how to make music. Like the self-driving car, they're never actually gonna figure this out the way a human can. So instead, at 65LABS we teach computers how to break prettily.

NEXT WEEK

• This will be somewhat subject to COVID, but we will try to get some more systems up on the test stream and/or get started on the next entry for our Wreckage Systems Technical Overview series.

See you next time friends. Thanks for your support. Stay safe out there.