Looking forward to when people realize that human programmers are not deterministic and never have been.

Understanding all of this requires a lot of nuance and familiarity with both OSS and Bay Area lingo. Fortunately for those of us in the alliance who need the support of the simple consumer, music visualization requires no such deep understanding. Joe Consumer is powerful, and we need them onboard.

Federation & P2P etc are mainly held back because of the dilemma where being open makes a platform much better for consumers and therefore viable yet allergic to capital because open platforms sabotage their own competitive moats. This is why multi-party funding and incidental funding are key!

Taking things way further, why not use such tools to fund the development of their open kin? Create viable open platforms. The decentral kids have been working hard on the foundations all along.

All open source boosts competition and inoculates the market against crappy business models, so it's easy to reason that a ton of revenue going into consumer open source means replacing a whole of bad with a whole lot more good, and operating fancy MasterCard to make that happen is viable.

In these times of broad cynicism towards the Bay Area, that idea is able to cut through. If the revenue going into making open source suddenly jumps and the "cost" (revenue necessary to motivate operating the platform) is barely a tenth of the new revenue going into open development, huge win.

IIRC it was Bill Gates who said a platform is when the economic value of everybody that uses it exceeds the value of the company that creates it. I'd add that the value created by the users that would not be created must far exceed what is captured by the operators.

Some people are afraid to learn things because of who they learn them from. Knowledge isn't colored that way. In the real world, Yoda can learn force lightning, a darkside skill, without his eyes suddenly glowing and his face melting.

That kind of quick value means getting people's foot in the door. It also means the value of simple data or render techniques will scale quite a bit better. More value from the same data always bodes well for any software ecosystem.

One of the problems with Milkdrop's architecture is that 3D data and render techniques are not really interoperable. You can't re-use meshes across presets. Nothing is really very interoperable. With MuTate, making a few 3D models will just automatically compose.

Be a #3D artist or graphics programmer. The minimum thing with commercial value is quite a lot of work for the individual. Instead building just some demo toys for a music visualizer is instantly viable.

First proc macro of the day shaped up and pushed. Really looking forward to getting out of the boring updates phase. It's nice that the code is moving, but obviously posting videos of new effects is where we want to be. #rust #vulkan

Honing the executive capacity of organic online community into a sharpened instrument that can do real things is a bypass. We don't need to tear down the entire internet and online culture at once. We need only a properly placed tiny channel from the lagoon to the sea.

To overcome choked flow in a supersonic nozzle, a tiny bypass is all that necessary, leaking the high pressure behind the bow shock so that the wave front is swallowed supersonically, reaching the unchoked flow condition. The bypass flow is rather like a chemical catalyst.

The outcome we see is not always proof that one is more natural. The other can survive scrutiny, a tantalizing mathematical certainty that would just as easily be the object of many just-world rationalizations if only it could be reached for a moment.

Rather than nostalgia or even artifacts of declinism, the sense that captures my instinct is that there were two distinct stable futures of online chemical chain reactions, and that they are separated by a bifurcation. Both are highly stable, but only one is true at any time.

There's an interesting effect in real systems called bifurcation. I first encountered it with the phenomenon of choked flow in a Busemann biplane. There can be multiple distinct stable solutions for the same far-field conditions, and it's very difficult to get from one to the other.

Lean back, smell the stack of unburned CD ROMs, and count the rows of light through slits in the vertical blinds. Greenery off the balcony is exploding. The smell of an AC running signals the change of seasons. The digital world is but a frontier embedded in the physical one. #lexan @eff.org

If anything, continuing to purge planning is the right move. There's plenty of good stuff to make the necessary milestones that kicked off this MuTate adventure. The DSP progress is just sitting there being cool but invisible until it's wired into the display.

Settled on a few syntaxes for pipeline macros yesterday. Looking very orthodox ie other Rust users will "just get it." Going to avoid looking at anything architectural for a minute.

Going to sleep on the macro syntax for pipeline composition. Can see either a Lisp or a Nix style (Json with functions (lies! it's Pascal or something IIRC)) syntax working. As much as some people dramatize parentheses, I have never met anyone who prefers #Nix syntax to anything. #Rust #Vulkan

I'm quite positive on ML development, but it's clear that certain things cannot remain. It will be a blessing when traditional social, which is essentially nothing more than a vehicle of intentional and unintentional distortion, gets dethroned.

How do feed your algorithms when you can't Turing test anything? Where is the ground source input if the users, having been bludgeoned by disinfo for years, provide no useful signal even if you could sort them from the AI?

There's no way we don't go back to that very primitive style #social where you connect through people you know in meat space.

Currently working on a proc macro declaration for Pipelines. I had earlier looked at how Bevy does various things only to realize that it does not do them at all and just lets the user find out about problems when the shader explodes. We're onto something. #rust #vulkan.

Work on MuTate was in danger of live-locking on too much architecture uncertainty to digest at once, so I've been aggressively under-engineering (the 🤠 counterpart to over-engineering) to unstick things. A couple files nearly evaporated under pressure to do something concrete. #rust #coding

If the input ratios are favorable because the details you omit only have 1-2 good ways to do it and no real tradeoffs, the expanded code generally contains a lot of useful or at least interesting tidbits. Requires editing, which hand coding doesn't avoid because we're not deterministic at all.

You start off rubber ducking in natural language. By the time you get the problem description lined out, instead of writing the code, you just press "send" because this duck can actually scratch together a draft.

Another ancient behavior that emerged from email was chain mail, the original news feed. At least you knew who was falling for everything (or superstitious enough to believe what would happen if they did / didn't press send).

Email made us aware that non-revocable global message routing is a bad idea. The common name for this behavior in a messaging system is spam. Or at least any messaging that has global routing is guaranteed to invite spam.