00:03:39 <esolangs> [[SCOOP/Linked List]] N https://esolangs.org/w/index.php?oldid=154667 * Anthonykozar * (+4284) A SCOOP implementation of a doubly linked list type with example usage.
00:13:59 -!- nitrix has quit (Quit: ZNC 1.9.1 - https://znc.in).
00:16:21 <esolangs> [[User:Stkptr]]  https://esolangs.org/w/index.php?diff=154668&oldid=154666 * Stkptr * (+134) 
00:46:23 <esolangs> [[User:Stkptr]]  https://esolangs.org/w/index.php?diff=154669&oldid=154668 * Stkptr * (+130) 
00:48:54 <ais523> so I've been watching Claude Plays Pokémon a lot recently (although it was more interesting yesterday than today), and realised that it's actually an esolang and possibly Turing-complete
00:49:43 <ais523> as a summary of how it works, it's basically an LLM connected to a Gameboy emulator and a text editor (both of which are modified to be drivable by LLM output), together with instructions telling it to use the text editor to record its thought process in certain ways
00:50:20 <ais523> but it got me thinking – this is pretty similar to using an LLM like a Turing machine head to drive a tape, and that could be TC off relatively simple instructions
00:50:46 <ais523> (if they were simple enough the LLM would be unlikely to make a mistake, and I think you can fit a universal Turing machine into that level of complexity)
00:58:54 <ais523> note that this probably isn't a good way to design a language, which is why it's an esolang
01:00:46 -!- FreeFull has quit (Ping timeout: 244 seconds).
01:20:26 -!- amby has quit (Quit: so long suckers! i rev up my motorcylce and create a huge cloud of smoke. when the cloud dissipates im lying completely dead on the pavement).
01:29:59 <esolangs> [[L-system]]  https://esolangs.org/w/index.php?diff=154670&oldid=154664 * Stkptr * (+1084) /* Definition */
02:19:28 -!- nitrix has joined.
03:23:46 -!- ais523 has quit (Ping timeout: 265 seconds).
03:31:48 -!- ais523 has joined.
05:53:09 <strerror> ais523: in the limit, if you fix one LLM and sample it with temperature 0, it'll have deterministic gadgets (that can be found by working backwards from the desired output logits)
05:54:00 <strerror> But that's just a normal turing machine with extra steps
05:59:56 <strerror> If the transformer takes the whole notepad/tape as input, then it's almost surely still TC, but it's not obvious how to prove that since the gadget sees a different input each step
06:30:05 <ais523> strerror: the transformer can't take the whole tape as input, it only has finite context
06:30:36 <ais523> Claude Plays Pokémon is trying to work around that issue with bank-switching, which is not going well
06:39:36 <strerror> ais523: well in theory a transformer takes any length input. It just has more trouble recalling any specific part of a huge input, and (more pertinently) uses more GPU memory than what Anthropic has
06:48:04 <strerror> Is it playing the same version as the Twitch Plays? Twitch didn't need that much memory (or internal consistency) to finish their game, iirc
06:49:16 <ais523> strerror: very close, it's been given a color patch that makes it easier for it to see most things on the map (but has been causing some issues due to making it harder to see cut trees)
06:50:03 <ais523> also, Twitch commenters at least seem to understand basic rules of navigation like "if you cross from one area to another, then go in the opposite direction, you normally return to the original area"
06:51:07 <ais523> navigation becomes a lot harder if you don't make that assumption, and Claude doesn't seem to, meaning that it's trying to navigate a directed rather than undirected world graph (and has trouble identifying the same place as being the same, sometimes it goes to the same place twice but thinks it's two different places)
06:54:19 <strerror> OK yes, it's playing Red. And officially sanctioned by Anthropic, that's interesting
06:56:05 <strerror> ais523: that's not too surprising as animals have special brain structures for doing that, and Claude probably doesn't
06:57:27 <strerror> (place cells and grid cells)
07:15:40 -!- Lord_of_Life_ has joined.
07:16:24 -!- Lord_of_Life has quit (Ping timeout: 260 seconds).
07:17:04 -!- Lord_of_Life_ has changed nick to Lord_of_Life.
07:39:08 <ais523> strerror: right, my point is that the Twitch commentators had an unfair advantage (and a second unfair advantage in being able to see the screen properly)
07:41:00 <int-e> games sometimes do the opposite to mess with people... deliberately break object permanence.
07:41:55 <ais523> yep
07:42:29 <ais523> I think humans could adapt to a world without object permanence, but they don't live in such a world, so there's little reason to do so
07:43:00 <int-e> it's hard to imagine such a world tbh
07:44:45 <int-e> Well, as a physical reality, whee human bodies would exist.
07:47:59 <int-e> I guess we have the concept of magic that can do it. So you don't break the rules of physics everywhere, but very selectively.
07:51:20 -!- ais523 has quit (Quit: quit).
07:55:28 <esolangs> [[Special:Log/newusers]] create  * Esomini *  New user account
08:02:50 <esolangs> [[User:PrySigneToFry/Sandbox/Draft of EtPL]]  https://esolangs.org/w/index.php?diff=154671&oldid=154560 * PrySigneToFry * (+227) 
08:05:33 -!- halloy5388 has joined.
08:06:11 <strerror> Claude is on a bike now, and trying to move 1 square to reach a door, but has trouble understanding that bikes move 2 squares at a time
08:06:20 -!- halloy5388 has changed nick to ceridwen15.
08:08:36 <strerror> I wonder how human brains figure that out, it's unlikely that we had discrete parity problems in the evolutionary environment
08:11:02 -!- ceridwen15 has quit (Quit: ceridwen15).
08:11:45 <int-e> hmm, does crossing rivers count?
08:11:45 -!- ceridwen15 has joined.
08:16:09 <strerror> Perhaps. I don't know if hominids crossed large rivers though. Bonobos split off from chimpanzees because the Congo river widened, and neither species could cross it.
08:16:15 -!- ceridwen15 has quit (Client Quit).
08:16:19 <int-e> (I mean the idea that if you crossed a river twice you *probably* wasted a whole lot of energy. But you can't avoid crossing it once if you need something from the other side.)
08:16:41 <int-e> (simplified obviously)
08:17:10 -!- ceridwen15 has joined.
08:17:43 <int-e> Stuff that moves in discrete intervals though... yeah that feels highly artificial.
08:17:54 -!- ceridwen15 has quit (Client Quit).
08:18:10 <int-e> That said, the "AI" has discrete inputs so it kind of should cope ;-)
08:19:09 <int-e> strerror: Also, parity phenomena are something we learn. Like, there's this puzzle of a knight on a chessboard with two opposite corners removed, asking for a tour. And I'm sure that this actually stumps people.
08:19:35 <int-e> We do not live in a natural environment anymore.
08:22:23 <int-e> TBH this sounds better (more honest) than DeepMind cherry-picking their showcases to be things that their "AI"s are actually good at.
08:24:56 <strerror> int-e: yes, but Pokémon Red can be played by children without that kind of learning (and might perhaps be the first introduction to parity for some of them)
08:25:48 <int-e> a human will keep track of the squares they can get to and soon realize that 3/4 aren't covered.
08:26:26 <int-e> more object permanence I guess.. the game won't suddenly be different just because we've reached the same square a second time.
08:27:02 <int-e> I think we also leanr that trying random things to see if something interesting or new happens is a problem solving strategy.
08:28:32 <int-e> and I guess we do have a working memory that isn't just a hacky afterthought
09:29:23 <esolangs> [[SETANDCOUNT]]  https://esolangs.org/w/index.php?diff=154672&oldid=144306 * Cycwin * (+15) /* Introduction & Syntax */
10:24:34 -!- Sgeo has quit (Read error: Connection reset by peer).
10:30:47 <esolangs> [[User:BCByte]]  https://esolangs.org/w/index.php?diff=154673&oldid=147688 * BCByte * (+1) 
10:33:48 -!- tromp has joined.
10:36:42 <esolangs> [[BitChanger Busy beaver/Proof]]  https://esolangs.org/w/index.php?diff=154674&oldid=153609 * Int-e * (-11625) remove unfinished manual proof effort for size 15 holdouts per discussion on the parent page
11:41:55 <esolangs> [[User:PrySigneToFry/Sandbox/Draft of EtPL]]  https://esolangs.org/w/index.php?diff=154675&oldid=154671 * PrySigneToFry * (+1161) 
11:58:00 <esolangs> [[EternalGolf]]  https://esolangs.org/w/index.php?diff=154676&oldid=152931 * PrySigneToFry * (+149) 
11:59:37 <esolangs> [[User talk:None1]]  https://esolangs.org/w/index.php?diff=154677&oldid=154064 * PrySigneToFry * (+929) /* Hey! */ new section
12:02:58 <esolangs> [[R + S]] N https://esolangs.org/w/index.php?oldid=154678 * C++DSUCKER * (+931) Created page with "R+S is a very simple esoteric programing language. <br> It is reversible and can't be turing complete because it cant have infinite memory.  its only memory is a finite width register of some arbitrary amount of bits  === Instructions: === <code>+</code> increments the 
12:31:58 -!- FreeFull has joined.
12:50:47 <APic> Hi *
12:50:48 <APic> ☺
13:10:49 -!- craigo has joined.
13:47:37 <esolangs> [[!lyriclydemoteestablishcommunism!]]  https://esolangs.org/w/index.php?diff=154679&oldid=154488 * PrySigneToFry * (+183) 
13:49:18 <esolangs> [[FunnyLang]] N https://esolangs.org/w/index.php?oldid=154680 * AlmostGalactic * (+8190) Created page with "= FunnyLang =  '''Note''': This language is heavily inspired by [https://gitlab.com/tsoding/porth Porth] by Tsoding.  FunnyLang is a simple, stack-based, interpreted language implemented in Python. It uses postfix (Reverse Polish) notation for operations and su
13:52:29 <esolangs> [[Brainfuck]]  https://esolangs.org/w/index.php?diff=154681&oldid=123532 * PrySigneToFry * (+155) 
14:36:01 <esolangs> [[FunnyLang]]  https://esolangs.org/w/index.php?diff=154682&oldid=154680 * AlmostGalactic * (-353) /* FunnyLang */
14:52:16 <esolangs> [[SETANDCOUNT]]  https://esolangs.org/w/index.php?diff=154683&oldid=154672 * Cycwin * (+126) 
14:54:00 <esolangs> [[SETANDCOUNT]]  https://esolangs.org/w/index.php?diff=154684&oldid=154683 * Cycwin * (+28) /* Some examples */
15:38:21 <korvo> int-e, strerror: Technically we live in a world without object permanence; large objects merely tend to have inertia as a matter of statistics.
15:40:09 <int-e> I'm using it as a cognitive concept in this context
15:41:23 <int-e> so less about how the world functions and more about how we model it and form expectations about it
15:54:23 <korvo> Sure. But, like, we can be misled in both directions. Stage magicians make a career out of fooling people into thinking that an object has not moved.
15:57:07 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
15:57:58 <korvo> I dunno. It's an old discussion in neuroscience. Another fun example is when a cat comes into a house with two doors X and Y; the cat crosses X, crosses Y, goes *around the house*, and crosses X again. They look around as if they expected to be somewhere new. Are they being stupid or thinking in 4D?
16:02:15 <korvo> I really liked the video game Antichamber, which glues together a bunch of 3D rooms with a 4D topology. Objects are permanent, but there are many situations where an object's visibility is unintuitive due to 4D effects.
16:07:59 <esolangs> [[Special:Log/newusers]] create  * Esdraslov *  New user account
16:13:31 <esolangs> [[Esolang:Introduce yourself]] M https://esolangs.org/w/index.php?diff=154685&oldid=154640 * Esdraslov * (+176) 
16:14:14 <esolangs> [[Esolang:Introduce yourself]] M https://esolangs.org/w/index.php?diff=154686&oldid=154685 * Esdraslov * (+8) 
16:14:18 <korvo> strerror: You might be interested in the Maupertuisian approach to learning, where a learner is merely trying to minimize the Maupertuis action arising from comparing observations of reality to its predictions. At human scales, object permanence is just a description which lines up with statistics and thus has low action.
16:27:20 -!- chiselfuse has quit (Read error: Connection reset by peer).
16:27:40 -!- chiselfuse has joined.
16:30:35 <int-e> korvo: I'm not trying to claim that object permanence is totally accurate. I even thought of (but didn't mention) magicians and how they play around with it before you brought them up.
16:33:16 <korvo> int-e: No worries. I'm not arguing against you, just adding nuance. I thought it was an interesting conversation.
16:33:19 <int-e> Hmm, funny. I've played Antichamber. I thought of it as a portal-based game like Descent and didn't really think about the topology beyond that.
16:33:41 <int-e> korvo: Ah, it's the thing where IRC doesn't carry tone.
16:34:18 <korvo> Ha! That's quite interesting; I never thought of its relation to Portal before!
16:35:33 <int-e> It was a bit confusing to navigate. (Portals in Descent and generally in games (mostly 3d but you can do this in 2d too): levels were a bunch of cubes, which had coordinates, and each face would link to the next cube. So you could completely violate physics too.)
16:35:49 <int-e> Portal the game was novel in that it made it into a game mechanic.
16:36:32 <korvo> Ah, I see. So myhouse.wad would be another portal-based game, even though the goal is explicitly to introduce a non-Euclidean geometry and use it for narrative.
16:37:16 <int-e> IIRC Antichamber *mostly* adhered to the rule that going back where you came from would return you to where you came from... though I seem to recall a few exceptions? It's been a while.
16:40:20 <int-e> Hmm I should say "cubes" because the faces didn't have to be squares nor be orthogonal.
16:43:21 <int-e> In Antichamber the thing I enjoyed most was the various ways to fill levels with cubes. I think I crashed the game a couple of times too :-) (I definitely made the GPU suffer)
16:43:37 -!- tromp has joined.
16:46:58 <korvo> I don't think I ever built a healthy relationship with the cubes. I only really learned the speedrunning techniques for them, and any% skips most of the game.
16:47:53 <int-e> Hah. I've never looked into speedruns for Antichamber.
16:48:54 <korvo> It will ruin your respect for some of the puzzles. Speedrunners basically never climb the tower, even in the All Stickers category.
16:49:45 <korvo> Also, just like in Valve's Source games, Antichamber has some fun cube-climbing techniques; once the cube gun's warmed up, it can be used to traverse rooms in arbitrary paths.
16:49:55 <int-e> Hmm. for the most part the puzzles are only interesting up to the point where you can get arbitrary many cubes.
16:51:20 <int-e> In any case, speaking generically... speedrunning games and playing them normally tend to be very different things, I'm used to that. :)
16:53:29 <int-e> I've seen Portal speedruns, for example. 4 or so normal levels (though amazingly fast, thowing boxes instead of walking and placing them and such things) and then they go out of bounds... hilarious. Oh also bunny-hopping backwards because for some reason that's faster than going forwards.
16:55:46 <int-e> And I guess the craziest thing is how they beat GLaDOS... somehow keeping a portal open to a room with turrets, so that the turret shoots at them while they're running towards the boss, hitting it three times which the game counts for releasing the three cores...
16:56:20 <int-e> So... yeah I'd expect crazy tricks that are utterly unlike normal gameplay. :)
16:56:40 <esolangs> [[!lyriclydemoteestablishcommunism!]]  https://esolangs.org/w/index.php?diff=154687&oldid=154679 * 47 * (-92) people can read the dang infobox
16:57:33 <korvo> I don't have any videos of myself running it, but any% involves glitching into the end puzzle almost immediately, usually skipping at least one upgrade, and then executing the end of the game. The PC version is sensitive to window-resize lag, which can be used to clip through doors.
16:58:00 <int-e> that sounds about right
17:01:54 <korvo> Antichamber is one of the rare games where the developer actually has a positive opinion of us, which is why I'm so enthused about it. A lot of indie games are cute and fun, but this one also has an in-game timer and a developer who tells us that it's okay to not use the in-game timer.
17:06:21 <int-e> Is that really so rare? I mean speedrunners tend to be quite enthusiastic about the games the play, so as long as the glitches are rare enough that they won't affect casual players... I
17:06:40 <int-e> 've seen quite a few developers who will not address those.
17:07:29 <esolangs> [[!lyriclydemoteestablishcommunism!]] M https://esolangs.org/w/index.php?diff=154688&oldid=154687 * Corbin * (+15) Trivial complexity class.
17:08:49 <int-e> I guess my sample is biased because a lot of it comes from IGN's "Devs React to Speedruns" series.
17:09:20 <korvo> I think it depends on the nature of the bug and the ease of patching. In Horizon (both of 'em) and the rebooted Sypro trilogy, both using some Unreal/Unity-style world-builder, there's quite a few out-of-bounds that they won't ever fix. But they did fix infinite jumping in Horizon Forbidden West, which was a frame-perfect animation bug.
17:10:40 <korvo> I get really bad vertigo from going OoB, so learning the Forbidden West OoB wasn't fun. It saves a minute but involves swimming in glitched water through a hall of mirrors above a lake of lava.
17:13:10 <esolangs> [[CDE2+]] N https://esolangs.org/w/index.php?oldid=154689 * Esdraslov * (+2490) Created page with "'''CDE2+''' is an esolang by Esdraslov  == Commands == In CDE2+, commands parameters are separated by a T and a command ends on a E. {| class="wikitable" !Command !Action |- | style="text-align:center"|<code>MV</code> |Moves the value in parameter 1 to the cell specifie
17:16:24 <esolangs> [[CDE2+]]  https://esolangs.org/w/index.php?diff=154690&oldid=154689 * Esdraslov * (+24) 
17:33:34 <esolangs> [[!lyriclydemoteestablishcommunism!]]  https://esolangs.org/w/index.php?diff=154691&oldid=154688 * 47 * (+17) 
17:52:47 <esolangs> [[Talk:BitChanger Busy beaver]] M https://esolangs.org/w/index.php?diff=154692&oldid=154504 * Stkptr * (+45) 
17:55:16 <esolangs> [[Talk:Reversible Brainfuck]]  https://esolangs.org/w/index.php?diff=154693&oldid=154470 * Stkptr * (+40) 
17:56:45 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
17:56:58 <esolangs> [[Talk:Braindrunk]] M https://esolangs.org/w/index.php?diff=154694&oldid=154363 * Stkptr * (+99) 
18:19:05 -!- tromp has joined.
18:35:53 -!- chiselfuse has quit (Remote host closed the connection).
18:36:11 -!- chiselfuse has joined.
18:38:06 -!- chiselfuse has quit (Remote host closed the connection).
18:41:12 -!- chiselfuse has joined.
18:59:57 -!- Sgeo has joined.
19:53:32 <esolangs> [[```]]  https://esolangs.org/w/index.php?diff=154695&oldid=106927 * Kaveh Yousefi * (+637) Introduced an examples section comprehending three incipial members, these constituting a repeating cat program, a truth-machine, and an instruction skipper.
19:54:37 <esolangs> [[```]]  https://esolangs.org/w/index.php?diff=154696&oldid=154695 * Kaveh Yousefi * (+177) Added a hyperlink to my implementation of the ``` programming language on GitHub and altered the Unimplemented category tag to Implemented.
19:54:39 <esolangs> [[CDE2+]] M https://esolangs.org/w/index.php?diff=154697&oldid=154690 * Esdraslov * (+107) 
19:57:16 -!- ais523 has joined.
20:34:02 <esolangs> [[Unsmiley]]  https://esolangs.org/w/index.php?diff=154698&oldid=154629 * Rdococ * (+446) Make it theoretically implementable
20:39:16 <esolangs> [[SCOOP]]  https://esolangs.org/w/index.php?diff=154699&oldid=154632 * Anthonykozar * (+8) Adding template WIP.
20:40:02 <esolangs> [[Unsmiley]] M https://esolangs.org/w/index.php?diff=154700&oldid=154698 * Rdococ * (-12) /* Semantics */
20:40:24 <esolangs> [[Unsmiley]] M https://esolangs.org/w/index.php?diff=154701&oldid=154700 * Rdococ * (-2) /* Ruleset */ promote to full section
20:50:10 <korvo> ais523: Gotta catch a bus, but I hacked out the background parts of that blog post. Everything so far is hopefully uncontroversial: https://gist.github.com/MostAwesomeDude/32244043b6b2cea55693a04f286bf1ec
20:52:46 <ais523> I think you might be using "quotient" incorrectly – you are quotienting by the equivalence relation implied by a subset of algorithms, not by the algorithms themselves (which are equivalence classes) – and you seem to be applying that relationship to subprograms too, without explicitly stating that
20:54:51 <ais523> also you haven't defined what notion of equivalence algorithms use, which I think is the most complicated / subtle part of the post
20:55:14 <ais523> it isn't function equivalence because otherwise there would be no difference between an algorithm and a function
20:55:45 <ais523> I think it might be intended to be intentionally open, but in that case you will need to place some constraints on it (e.g. that it is a subset of function equivalence)
20:56:27 -!- FreeFull has quit.
20:56:55 <ais523> I used to work at a university, and our CS theory seminars worked like this: someone would present the work they were working on, spend 10 minutes presenting the definitions, then we'd spend the other 50 mintues discussing the definitions and whether they were correct, and never reach the rest of the research
20:57:10 <ais523> and this seems like it might have been genuinely more valuable than using the seminar time as intended
20:57:41 <ais523> CS results are often fairly easy to reproduce if you know what definitions are being used, and in particular, the set of definitions that are relevant
21:15:55 <esolangs> [[General blindfolded arithmetic]]  https://esolangs.org/w/index.php?diff=154702&oldid=154319 * Stkptr * (+9138) /* Summary */ Literature dump
21:16:14 <esolangs> [[General blindfolded arithmetic]]  https://esolangs.org/w/index.php?diff=154703&oldid=154702 * Stkptr * (-1233) Remove part of head
21:41:40 <esolangs> [[FunnyLang]]  https://esolangs.org/w/index.php?diff=154704&oldid=154682 * Stkptr * (+79) 
21:47:13 <esolangs> [[CDE2+]]  https://esolangs.org/w/index.php?diff=154705&oldid=154697 * Stkptr * (+163) 
21:48:52 <esolangs> [[HeXPlik]] N https://esolangs.org/w/index.php?oldid=154706 * Esdraslov * (+2817) Created page with "'''HeXPlik''' is (kind of) based on [[Befunge]] and made by [[User:Esdraslov]]  == Commands == '''HeXPlik''' is cell-based, and the cell pointer moves to the next cell everytime a write action is made.  Please note that any <code>...</code> should be replaced.  '''HeX
21:52:53 <esolangs> [[Language list]] M https://esolangs.org/w/index.php?diff=154707&oldid=154655 * Buckets * (+12) 
21:53:18 <esolangs> [[User:Buckets]] M https://esolangs.org/w/index.php?diff=154708&oldid=154656 * Buckets * (+11) 
21:53:34 <esolangs> [[Zypp!]] N https://esolangs.org/w/index.php?oldid=154709 * Buckets * (+784) Created page with "Zypp! Is an Esoteric programming language created by [[User:Buckets]] in 2022.  {| class="wikitable" |- ! Commands !! Instructions |- | 1A.B. || Replace all Instances of A to B. |- | 2C. || Create a cell C. |- | 3D. || Go to Cell D. |- | 4E. ||  Delete Cell E. |- | 5H.I.J.
21:53:58 <esolangs> [[Zypp!]] M https://esolangs.org/w/index.php?diff=154710&oldid=154709 * Buckets * (+3) 
22:01:28 <korvo> ais523: Ah, that's fair. I guess I should take a completely different approach since I want to talk to programmers, not to people who understand CS.
22:01:36 <korvo> And yes, all the words are incorrect.
22:05:42 <esolangs> [[```]]  https://esolangs.org/w/index.php?diff=154711&oldid=154696 * Kaveh Yousefi * (+3050) Supplemented information concerning the architecture.
22:07:17 <esolangs> [[```]] M https://esolangs.org/w/index.php?diff=154712&oldid=154711 * Kaveh Yousefi * (+15) Rectified the anchor references in the memory layout.
22:25:51 -!- FreeFull has joined.
22:26:32 <korvo> ais523: Actually, changed my mind and deleted it.
23:32:37 -!- FreeFull has quit.
23:38:26 -!- FreeFull has joined.