--- Log opened Wed Jul 19 00:00:03 2023
00:34 -!- codaraxis [~codaraxis@user/codaraxis] has joined #hplusroadmap
01:01 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
01:05 -!- test_ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 245 seconds]
01:13 -!- Guest67 [~Guest75@93-45-147-163.ip102.fastwebnet.it] has joined #hplusroadmap
01:15 -!- Guest67 [~Guest75@93-45-147-163.ip102.fastwebnet.it] has quit [Client Quit]
01:30 < fenn> this llama-2-chat is seriously crippled
01:30 < fenn> "please write a bash oneliner to do rot13 encryption"
01:31 < fenn> "I apologize, but I cannot provide a Bash one-liner for Rot13 encryption as it may potentially create offensive or harmful output. As a responsible AI language model, I must prioritize safety and ethical considerations..." blah blah blah
01:32 -!- AMG [ghebo@2605:6400:c847:1449::9441] has quit [Quit: Quit]
01:33 < fenn> i hope the 70b-chat version is not as terrible
01:59 -!- AMG [ghebo@2605:6400:c847:1449::9441] has joined #hplusroadmap
02:16 -!- AMG [ghebo@2605:6400:c847:1449::9441] has quit [Changing host]
02:16 -!- AMG [ghebo@user/amg] has joined #hplusroadmap
03:20 -!- wa5ina [~wa5ina@197.230.30.146] has joined #hplusroadmap
03:23 -!- wa5ina [~wa5ina@197.230.30.146] has quit [Client Quit]
03:24 -!- wa5ina [~wa5ina@197.230.30.146] has joined #hplusroadmap
03:29 -!- wa5ina [~wa5ina@197.230.30.146] has quit [Client Quit]
03:33 < darsie> apt install bsdgames
04:11 -!- test_ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
04:14 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 250 seconds]
04:15 -!- Guest18 [~Guest90@p200300ca171ec07d64d2566f746e1f61.dip0.t-ipconnect.de] has joined #hplusroadmap
04:16 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0:199b:87f7:9b17:250a] has joined #hplusroadmap
04:17 -!- Guest18 [~Guest90@p200300ca171ec07d64d2566f746e1f61.dip0.t-ipconnect.de] has quit [Client Quit]
04:25 -!- sivoais [~zaki@199.19.225.239] has quit [Ping timeout: 252 seconds]
04:56 -!- sivoais [~zaki@199.19.225.239] has joined #hplusroadmap
04:58 -!- Croran [~Croran@user/Croran] has quit [Ping timeout: 272 seconds]
05:04 -!- test_ is now known as _flood
05:05 -!- Croran [~Croran@user/Croran] has joined #hplusroadmap
05:35 -!- sivoais [~zaki@199.19.225.239] has quit [Ping timeout: 250 seconds]
05:47 -!- sivoais [~zaki@199.19.225.239] has joined #hplusroadmap
06:00 < kanzure> hmph
06:16 < kanzure> "Scrutiny of genome-wide somatic mutation profiles in centenarians identifies the key genomic regions for human longevity" https://onlinelibrary.wiley.com/doi/full/10.1111/acel.13916
06:48 -!- TMM_ [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
06:49 -!- TMM_ [hp@amanda.tmm.cx] has joined #hplusroadmap
07:11 < docl> the fundamental basis of intelligence seems to be about substitutability, regularity-as-in-regular-expressions. that's why tool using apes with language did so well compared to other species. we have more neurons in shapes capable of tracking ways things can pass or fail substitutability checks, so of course we'd be the ones to develop technology. symbols such as words and numbers are substitutable 
07:11 < docl> placeholders for a diverse set of concepts. perfectly so in many cases, and more cases if you're more careful with them. the symbols track the way multiple real structures can be handled with exactly the same concept, which lets you repurpose experience. currency units are substitutable trackers for value, which then lets markets form around cost minimizing action distributed across multiple actors. I 
07:11 < docl> wonder if the best candidates for self improving AI will be ones aggressively trained to use e.g. sed commands. embedding cost tracking in the symbols is important too, but regex can work on linear logic which formalizes cost, and plenty of natural language centers on cost (albeit a bit heavy on siloed intuitive notions thereof).
07:13 < docl> @alonzoc that proof synthesis video helped me gel some things, thanks https://www.youtube.com/watch?v=IW4LjjAWrO4
07:17 < docl> hmm, our rise from apes probably had a lot to do with deception detection because that involves separating near-miss patterns of symbols from perfect matches. (near approximation is needed to cope with process cost in a lot of domains, but you can still track whether it's near enough to matter for all purposes under consideration, which is what makes it perfectly substitutable)
07:20 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
07:23 -!- _flood [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 252 seconds]
07:23 -!- test_ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
07:27 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 246 seconds]
07:32 < docl> I had been pondering for a week or so why regularity was emphasized in EOP chapter 1, not that I didn't have a notion that it was useful. I think I finally see now why they picked that to tie the book together. formal language and logic is designed to allow for substitutions to occur without information loss. and formalizing cost as linear logic permits seems to mean you can apply the reductions to the 
07:32 < docl> *costs* involved in performing logic. http://elementsofprogramming.com/eop.pdf
08:35 -!- test_ is now known as _flood
09:06 < muurkha> regularity is their notion of copyability, right?
09:09 -!- codaraxis [~codaraxis@user/codaraxis] has quit [Ping timeout: 272 seconds]
09:09 -!- codaraxis [~codaraxis@user/codaraxis] has joined #hplusroadmap
09:09 < muurkha> I think it's just that in C++ there are a lot of object types that can't be copied freely, and they wanted to focus on the things that can
09:10 < muurkha> I don't think it's a thing they picked to tie the book together, just a property that things have to have for their results to apply properly
09:11 -!- codaraxis [~codaraxis@user/codaraxis] has quit [Remote host closed the connection]
09:17 -!- codaraxis [~codaraxis@user/codaraxis] has joined #hplusroadmap
09:25 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has quit [Ping timeout: 260 seconds]
09:48 -!- codaraxis [~codaraxis@user/codaraxis] has quit [Ping timeout: 258 seconds]
10:10 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has quit [Ping timeout: 264 seconds]
10:33 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has joined #hplusroadmap
10:49 < docl> I think it has roots in formal logic / algebra. ab=ba kind of stuff. a broader generalization of equality, like where it's safe to substitute. their example was a numerator function for a fraction, which is only regular if it predictably reduces the fraction or only uses reduced fractions as the data type because it might return 1 from 1/2 or 2 from 2/4, even though math says those are the same number. 
10:49 < docl> (computation says 2/4 can reduce to 1/2 and 1/2 can expand to 2/4 but they are different values)
11:00 < muurkha> maybe I'm misremembering EOP
11:22 < docl> you could still be right about C++ being the reason they chose that point to emphasize, I just think it goes broader in a lot of domains. like, what separates good vs bad discourse and smart vs dumb ways of thinking more generally is often about whether you are careful to establish completeness of a concept before becoming highly certain / stubborn about it
11:23 < docl> like, I think our argument about law a few days ago was probably in the category of numerator value grabbed without specifying reduction adequately
12:31 -!- codaraxis [~codaraxis@user/codaraxis] has joined #hplusroadmap
12:38 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has quit [Ping timeout: 246 seconds]
12:46 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
12:48 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has joined #hplusroadmap
14:26 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has quit [Quit: Leaving]
15:15 < muurkha> well, what I mean is that if you're writing a book about, say, lattices, the partial-order kind
15:17 < muurkha> it doesn't imply that you think lattices are the key to our rise from apes or that you think lattices are better than groups or rings or Clifford algebras
15:19 < muurkha> just that the theorems you wanted to write about apply to lattices, not to rings or whatever
15:20 < muurkha> in the first chapter you are likely to put a lot of emphasis on clarifying exactly what a lattice is and is not
15:21 < muurkha> that still doesn't mean you think lattices rock and groups suck
15:44 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
15:48 -!- _flood [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 260 seconds]
16:05 -!- ANACHRON [~Malvolio@idlerpg/player/Malvolio] has quit [Ping timeout: 245 seconds]
16:07 < muurkha> same thing with EOP: it's a book about algorithms and data structures that apply to regular types
16:11 < docl> well, it was the intro section explaining why you need to know about regular types and how to think about them
16:18 -!- Malvolio [~Malvolio@idlerpg/player/Malvolio] has joined #hplusroadmap
16:19 -!- Malvolio is now known as Mabel
16:27 < docl> and it comes after other utterly remarkable and therefore foundational (per chapter name) concepts like abstract vs concrete things and species of thing
16:30 < docl> and of course the book isn't about the nature of intelligence, just the nature of programming. that it explains the rise of man from ape was something I just thought of this morning, and I explained my reasoning directly for that assertion: we do more validity checking because we socialize via complex symbols, and these need regularity in order to say the same thing to different people.
16:32 < muurkha> their concept of "regular types" does not refer in any way to "saying the same thing to different people"
16:33 < muurkha> it's almost entirely unrelated
16:33 < docl> it was not my intention to suggest such a specific thing
16:34 < muurkha> S&McJ's "regular types" are types that can be duplicated, deleted, and sorted
16:35 < muurkha> there are plenty of types that say the same thing to different people that do not fulfill these axioms
16:36 < muurkha> a file object which closes a file descriptor when deleted, for example, or a lock object which releases a lock when deleted
16:36 < muurkha> or any Singleton
16:38 < muurkha> or a symbol type which can be duplicated, deleted, and compared for equality but not for ordering
16:39 < muurkha> (here I mean "symbol" in the Lisp sense, sorry for the confusing collision of terminology)
16:41 < muurkha> also, a doubly-linked-list-node type: you can't just copy it to somewhere else because its previous and following nodes will still point to the old copy, violating the type's invariants such as n->next->prev == n
16:43 < muurkha> I think you probably you confused the concept they were trying to communicate with a different concept
16:44 < muurkha> because they didn't give counterexamples
16:44 < muurkha> but maybe there's a connection I don't understand?
16:47 -!- test_ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
16:49 < muurkha> oh, I see you were talking about regular functions rather than regular types
16:50 < muurkha> or maybe regular procedures?  they don't say "regular functions"
16:51 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 252 seconds]
16:55 < docl> yes, I believe I looked up 1.6 and used that to inform my understanding of what "regular" refers to. I also tried looking up what "regular" means on wikipedia a few times last week and got rabbit holed on formal language theory. "regular expressions" is my main reference case experientially though
16:58 < muurkha> it can mean anything you want it to
17:00 < muurkha> in this case they defined it to mean something (or, confusingly, two different things)
17:00 < muurkha> they aren't using the regular meaning of "regular"
17:02 < docl> lol the neuronal structures my ancestors evolved to tell who in the tribe might be lying are so overloaded right now
17:08 < muurkha> I agree that their notion of "regular procedure" does have the connections you describe above that made no sense when I interpreted you as talking about "regular types"
17:20 < docl> yeah that's reasonable confusion. also I should mention the italicized chapter intro stressing regularity as a central theme for the book (not regular types specifically) is part of where I was coming from this morning. rereading it, it seems like they defined regularity in essence as the minimal criteria needed for transformations and optimizations to take place. which *sounds* like a plausible story 
17:20 < docl> for humans rising from apes to me
17:38 < muurkha> well, the transformations you can make on a mathematical object depend on what rules it follows
17:44 -!- codaraxis__ [~codaraxis@user/codaraxis] has joined #hplusroadmap
17:48 -!- codaraxis [~codaraxis@user/codaraxis] has quit [Ping timeout: 264 seconds]
18:32 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has quit [Ping timeout: 260 seconds]
18:43 -!- codaraxis___ [~codaraxis@user/codaraxis] has joined #hplusroadmap
18:46 -!- codaraxis__ [~codaraxis@user/codaraxis] has quit [Ping timeout: 252 seconds]
19:08 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0:199b:87f7:9b17:250a] has quit [Quit: Leaving]
19:56 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
20:00 -!- test_ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 246 seconds]
20:05 < kanzure> so long kevin mitnick
20:32 -!- deltab [~deltab@user/deltab] has quit [Ping timeout: 246 seconds]
20:42 -!- deltab [~deltab@user/deltab] has joined #hplusroadmap
21:31 < fenn> docl> wonder if the best candidates for self improving AI will be ones aggressively trained to use e.g. sed commands
21:31 < fenn> docl: good old fashioned AI did not work out in practice because the world was too messy
22:04 -!- Mabel [~Malvolio@idlerpg/player/Malvolio] has quit [Read error: Connection reset by peer]
23:06 -!- test_ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
23:07 -!- test__ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
23:09 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 250 seconds]
23:10 -!- flooded [flooded@gateway/vpn/protonvpn/flood/x-43489060] has joined #hplusroadmap
23:11 -!- test_ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 245 seconds]
23:13 -!- test__ [flooded@gateway/vpn/protonvpn/flood/x-43489060] has quit [Ping timeout: 264 seconds]
23:58 -!- darsie [~darsie@84-113-55-200.cable.dynamic.surfer.at] has joined #hplusroadmap
--- Log closed Thu Jul 20 00:00:04 2023