--- Log opened Fri Nov 05 00:00:23 2021
00:22 < fenn> The National Rifle Association (NRA) is also against the Superhuman Registration Act, stating that the Second Amendment to the United States Constitution's protection of the right to keep and bear arms in the United States applies to superpowers. An NRA representative testifies to Congress that "if powers are outlawed, only outlaws will have powers".
00:23 < fenn> not the real NRA, who are a bunch of unprincipled wussies, the fictional NRA
00:46 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has joined #hplusroadmap
00:47 < fenn> .
01:35 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has quit [Quit: bye bye]
02:46 -!- Codaraxis [~Codaraxis@user/codaraxis] has quit [Ping timeout: 268 seconds]
03:02 < maaku> "I’d suspect the design of an object would need to include instructions to make and use a tool to wrap wire or do whatever niche thing." <-- that's called constructivist design, and it is something that is generally good to start with
05:27 -!- spaceangel [~spaceange@ip-89-176-181-220.net.upcbroadband.cz] has joined #hplusroadmap
06:51 < kanzure> computation can be done by fluidics, pneumatics, mechanical motion and a number of other things
06:51 < kanzure> a little weird to suggest that the chatters don't know about reprap but if you wanna take that bet....
06:52 < muurkha> we wouldn't be here if we weren't weird
06:52 < kanzure> biological platforms are already self-replicating and only require a gazillion dollars to control and reprogram
09:58 -!- spaceangel [~spaceange@ip-89-176-181-220.net.upcbroadband.cz] has quit [Remote host closed the connection]
10:30 < docl> constructivist design is kinda hard to google
10:49 < maaku> docl: https://en.wikipedia.org/wiki/Constructivism
10:50 < maaku> linked to the fan-out article because the other defintions are interesting but this one is probably most relevant: https://en.wikipedia.org/wiki/Constructivism_(philosophy_of_mathematics)
10:51 < docl> specific-use assemblers that do one thing well instead of being able to build anything in your tech tree would make the game a lot more authentic. but there's a lot of complexity being abstracted away, like there being fundamentally different kinds of circuits and gears. and the inserters/belts thing is unrealistically easy to do, even ignoring the wear and tear question. a realistic bootstrap would 
10:51 < docl> have you doing all the logistics manually, at least until you have motors. small enough motors/engines to integrate directly in the machines was a big part of the second industrial revolution -- belts connecting a big steam engine to everything was a huge logistical tradeoff
10:51 < maaku> in case it's not clear, "constructivist design" is not a term. I meant "constructivist" design
10:51 < docl> I was wondering which one of those was most relevant
10:53 < maaku> a constructivist theory is one in which things are proven by showing the algorithm for constructing an example
10:54 < maaku> a constructivist design, by logical extension, would reject plans which do not include manufacturing steps. a blueprint isn't a blueprint unless it shows how to make the thing
10:54 < maaku> it's a term I've been using in my own notes and writing for the better part of a decade, long enough that I forgot I probably coined it
10:55 < docl> nice!
10:56 < docl> I sort of use that as litmus test for whether I really understand a thing, can I actually come up with an example for how it would work
10:58 < maaku> Yeah in the late 19th/early 20th century a bunch a mathematicians felt similarly--that you shouldn't believe something unless you could come up with examples
10:58 < maaku> To their surprise, they found many widely accepted theories which could be "proven true" but in a way that didn't lend itself to generating examples
11:01 < maaku> The maths you can prove true in a generative way (such that the proof itself is a algorithm for "constructing" examples) is a subset of widely accepted math, but arguably far more interesting and useful in applied contexts
11:01 < docl> here's a paper that seems to be talking about this https://sci-hub.se/10.1016/B978-0-444-64235-6.50138-8
11:01 < docl> .title
11:01 < saxo> Sci-Hub | Process Innovation through bio-inspired design. 28th European Symposium on Computer Aided Process Engineering, 785–790 | 10.1016/B978-0-444-64235-6.50138-8
11:03 < maaku> Some CAD data structures are generative/constructivist in the sense that they specify machining operations. "Start with cube or sheet of metal, apply these mill/lathe/extrusion operations..."
11:03 < docl> I think their argument is that processes should be modular like cells (hence the bio-inspired part of the title)
11:04 < maaku> If we specify all blueprints that way, architecting a mechanical self-replicating system becomes much easier.
11:05 < maaku> Not just self-replicating systems though, also things like automatic design of factory eqipment for mass production of a part. You should be able to just specify a part constructively and have a "compiler" generate the factory design automatically.
11:06 < maaku> docl: thanks for the link. the abstract seems relevant. I'll file it away to read in full
11:11 < docl> they mention the main disadvantage is that it's highly combinatorial... I think that's accurate. like if you can't just use generic assemblers you end up needing a whole new assembler for each product (really some stuff does generalize in real life though, like a lathe can make a lot of stuff... but you need ways to constrain it to a mechanical pattern that creates a given thing)
11:25 < muurkha> one of my big revelations in Dercuano is that wear and tear has a totally different role in replicators than it does in industrial-age machinery; replicators just have to last long enough to produce on average more than one viable offspring.  dying after spawning is totally legit
11:30 < docl> yes but if you have a thousand processes and one goes bad and the unit doesn't have redundancy you lose the unit. and if you have 2 of every process that's 2000 processes (not necessarily active, but switching over and powering up is another thing you have to have a mechanism for)
11:30 < muurkha> if you have 400 20mm M6 screws in your self-replicating factory, it's okay if your M6-screw-making subassembly wears out after making 500.  but maybe not if it sometimes wears out after making 350
11:31 < docl> trade might be a workaround here, but then you need trade logistics
11:32 < muurkha> yeah, that's true.  living creatures tend to have a lot of cells that do pretty much the same thing
11:32 < muurkha> rather than just one
11:34 < muurkha> so you might have 10 or 100 M6-screw-making subasssemblies whose mean time to failure is 50 or 5 screws
11:35 < muurkha> you don't necessarily have to switch over and power up but you do need some way to deal with overproduction
11:39 < muurkha> but if you only have one M6-screw-making subassembly it has to have super reliability — like you said, if you have 1000 reproduction-critical processes, each with some failure rate, you need the average number of reproduction-critical processes that fail to be less than one per reproduction
11:41 < muurkha> which means that the failure rate for each of those 1000 processes needs to be 1000 reproductions
11:41 < docl> if it's enough less than 1, you can replace the whole unit, but if it's a close call you want process redundancy. this can be added as part of a robustication phase perhaps (e.g. start with only 2, but scale up to 10 before trying to replicate)
11:42 < muurkha> probably better to have many fewer nonredundant reproduction-critical processes
11:43 < muurkha> sure, like stomach cells
11:44 < muurkha> instead of having an M6-screw-making subassembly that lasts 500 screws, you have an M6-die-cutting setup that makes new M6 dies when one fails
11:51 < maaku> docl: I think it's a matter of scale.  Individual parts are produced by specialized assemblies. But when you get to the level of a pick-and-place, it can be made generic and reprogrammable
12:23 -!- archels [~neuralnet@static.65.156.69.159.clients.your-server.de] has quit [Ping timeout: 260 seconds]
12:25 -!- archels [~neuralnet@static.65.156.69.159.clients.your-server.de] has joined #hplusroadmap
12:42 < muurkha> in normal manufacturing practice this is largely a question of part volume, although there are some exceptions
12:43 < muurkha> if you need one or ten or maybe 100 of something, you produce it with a generic machine like a lathe or a 3-D printer
12:44 < muurkha> or you put together the assembly by hand on a bench with a screwdriver, or solder it together with a soldering iron
12:45 < muurkha> if you need hundreds or thousands of them, it may become worthwhile to make up molds, dies, jigs, stencils, and other specialized tooling
12:46 < muurkha> but there are things where you never reach that crossover point, and things for which you reach it much earlier, because of the particular processes you're using
12:50 < muurkha> like, if you want a cold-forged part in order to get a better strength-to-weight ratio, you pretty much have to make up dies unless it's something you can actually make by hammering, even if you only want one of them.  and if you need micron-level tolerances on a part, you're going to have to machine it (even investment casting won't do), even if you're making a hundred thousand of them (although 
12:50 < muurkha> with modern CNC this doesn't necessarily mean that the costs are through the roof the way it used to)
12:50 < muurkha> I'm just a muurkha, though, I've never made anything on a lathe or by investment casting.  so I might be missing some basic knowledge
13:45 -!- chiastre [~chiastre@user/chiastre] has quit [Quit: chiastre]
14:03 -!- chiastre [~chiastre@user/chiastre] has joined #hplusroadmap
14:27 -!- berndj-blackout [~berndj@ns1.linksynergy.co.za] has joined #hplusroadmap
14:27 -!- nsh [~lol@user/nsh] has quit [Ping timeout: 246 seconds]
14:27 -!- nsh [~lol@5.135.157.17] has joined #hplusroadmap
14:28 -!- rndhouse [~rndhousem@2001:470:69fc:105::1:1cfd] has quit [Ping timeout: 264 seconds]
14:28 -!- berndj [~berndj@ns1.linksynergy.co.za] has quit [Ping timeout: 264 seconds]
14:31 -!- berndj-blackout is now known as berndj
14:33 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::f2f0] has joined #hplusroadmap
14:34 < muurkha> nsh: your IP leaked when you rejoined without a cloak
14:44 -!- rndhouse [~rndhousem@2001:470:69fc:105::1:1cfd] has joined #hplusroadmap
19:17 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::f2f0] has quit [Quit: Leaving]
22:19 -!- strages [uid11297@id-11297.helmsley.irccloud.com] has quit [Ping timeout: 260 seconds]
22:20 -!- EnabrinTain [sid11525@id-11525.helmsley.irccloud.com] has quit [Ping timeout: 260 seconds]
22:21 -!- EnabrinTain [sid11525@helmsley.irccloud.com] has joined #hplusroadmap
22:21 -!- strages [uid11297@id-11297.helmsley.irccloud.com] has joined #hplusroadmap
--- Log closed Sat Nov 06 00:00:24 2021