00:04:39 <esolangs> [[Shakespeare]] M https://esolangs.org/w/index.php?diff=174624&oldid=161686 * Treeplate * (+2) some semicolons were forgotten
00:05:08 <esolangs> [[Shakespeare]] M https://esolangs.org/w/index.php?diff=174625&oldid=174624 * Treeplate * (+0) also, lt and gt were swapped
00:25:23 <esolangs> [[Talk:Deadfish Joust]] https://esolangs.org/w/index.php?diff=174626&oldid=64973 * Treeplate * (+204) /* Strategy */ new section
00:26:25 -!- 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).
01:35:33 <sorear> ~ for strikethrough seems very recent, it isn't even in original Markdown, but the way people use ^H/^W in usenet posts may be relevant
01:43:35 <esolangs> [[User:Fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff]] https://esolangs.org/w/index.php?diff=174627&oldid=163723 * Fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff * (+23)
01:44:00 <esolangs> [[User:Fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff]] https://esolangs.org/w/index.php?diff=174628&oldid=174627 * Fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff * (-23)
02:05:08 -!- citrons has quit (Remote host closed the connection).
02:11:40 <esolangs> [[Drive-In Window]] https://esolangs.org/w/index.php?diff=174629&oldid=138008 * TacoBeanBurrito * (+289) XKCD Random Number implementation
02:13:13 <esolangs> [[CCCC]] https://esolangs.org/w/index.php?diff=174630&oldid=172213 * ColorfulGalaxy's CA discoveries * (+99) ----
02:14:14 <esolangs> [[User:XKCD Random Number]] https://esolangs.org/w/index.php?diff=174631&oldid=173224 * TacoBeanBurrito * (+288) add Drive-In Window language
02:27:51 <esolangs> [[Error quine]] https://esolangs.org/w/index.php?diff=174632&oldid=172483 * ColorfulGalaxy's CA discoveries * (+6) ----
02:52:49 -!- impomatic has joined.
03:05:36 <esolangs> [[UnicodeLang]] https://esolangs.org/w/index.php?diff=174633&oldid=174619 * Qawtykit * (+506) Turned the commands list into a table for readability
03:14:29 -!- chloetax has quit (Read error: Connection reset by peer).
03:14:42 -!- chloetax has joined.
03:31:15 -!- impomatic has quit (Quit: Client closed).
04:08:11 <esolangs> [[UnicodeLang]] https://esolangs.org/w/index.php?diff=174634&oldid=174633 * PrySigneToFry * (+146)
05:14:56 <esolangs> [[MSS]] N https://esolangs.org/w/index.php?oldid=174635 * PrySigneToFry * (+6791) Created page with "MSS(pronounced as /ms/ where // is a rounded /e/) is designed by PSTF and his AI friend. This language is an interpreted language, formally called 'Meaningful Object-oriented Simple Syntax', and is usually abbreviated as 'MossLang' (not 'MOSS', to distinguish it from
05:15:40 <esolangs> [[Language list]] https://esolangs.org/w/index.php?diff=174636&oldid=174587 * PrySigneToFry * (+12)
05:54:47 -!- ais523 has joined.
06:30:56 <esolangs> [[Talk:I/D machine]] https://esolangs.org/w/index.php?diff=174637&oldid=174621 * Ais523 * (+493) it doesn't halt, but you can still put it into a loop
07:34:00 <esolangs> [[Bbtos]] M https://esolangs.org/w/index.php?diff=174638&oldid=160635 * Qazwsxplm * (+12)
07:37:24 <esolangs> [[List of quines]] M https://esolangs.org/w/index.php?diff=174639&oldid=173211 * Qazwsxplm * (+85)
07:38:08 <esolangs> [[List of quines]] M https://esolangs.org/w/index.php?diff=174640&oldid=174639 * Qazwsxplm * (+6) Damn it! Why? I don't know.
07:46:34 -!- tromp has joined.
07:46:34 <esolangs> [[NameError without a quine with a quine without a quine with a quine without a quine with a quine]] M https://esolangs.org/w/index.php?diff=174641&oldid=170694 * Qazwsxplm * (+1777)
07:46:50 <esolangs> [[NameError without a quine with a quine without a quine with a quine without a quine with a quine]] M https://esolangs.org/w/index.php?diff=174642&oldid=174641 * Qazwsxplm * (+2)
07:47:38 <esolangs> [[NameError without a quine with a quine without a quine with a quine without a quine with a quine]] M https://esolangs.org/w/index.php?diff=174643&oldid=174642 * Qazwsxplm * (-1497)
07:48:57 <esolangs> [[N-Type]] https://esolangs.org/w/index.php?diff=174644&oldid=174614 * Yayimhere2(school) * (+4) /* Example */
07:53:28 <esolangs> [[Talk:Shove]] https://esolangs.org/w/index.php?diff=174645&oldid=174605 * Yayimhere2(school) * (+17)
07:58:17 <esolangs> [[]] M https://esolangs.org/w/index.php?diff=174646&oldid=140853 * Qazwsxplm * (+20)
08:04:18 <esolangs> [[]] M https://esolangs.org/w/index.php?diff=174647&oldid=133954 * Qazwsxplm * (+51)
08:08:40 <esolangs> [[]] M https://esolangs.org/w/index.php?diff=174648&oldid=174647 * Qazwsxplm * (+186)
08:10:56 <esolangs> [[The bluetooth device is ready to pair]] M https://esolangs.org/w/index.php?diff=174649&oldid=125318 * Qazwsxplm * (+20)
08:44:24 <esolangs> [[AMONGUSISABIGSUSSYBAKAHAHAHAHAHATHISLANGUAGEISREALLYCOOLPLEASEUSEITMYLIFEDEPENDSONITORELSEPLSPLSPLSPLSPLSPLSPLSkahyghdfhm]] M https://esolangs.org/w/index.php?diff=174650&oldid=163935 * Qazwsxplm * (+385)
08:50:51 <esolangs> [[Esolang:About]] M https://esolangs.org/w/index.php?diff=174651&oldid=174537 * Qazwsxplm * (+497) i think i need a "wiki status" section for me to monitor
08:54:18 <esolangs> [[User:Qazwsxplm/frogs]] N https://esolangs.org/w/index.php?oldid=174652 * Qazwsxplm * (+136) Moved Heavpoot's vandalism to this page
08:55:05 <esolangs> [[Esolang:About]] https://esolangs.org/w/index.php?diff=174653&oldid=174651 * Corbin * (-497) Undo revision [[Special:Diff/174651|174651]] by [[Special:Contributions/Qazwsxplm|Qazwsxplm]] ([[User talk:Qazwsxplm|talk]]): Stop edit warring. If you want realtime monitoring then come idle in IRC and chat with us.
09:08:37 -!- Yayimhere has joined.
09:10:10 <korvo> I'm about to go to bed. But I've had a good day. It was a week of proving vibecoders wrong.
09:10:29 <esolangs> [[Special:Log/upload]] upload * Qazwsxplm * uploaded "[[File:MEGA****S.jpg]]"
09:10:36 <korvo> Right now I'm trying to kick off a build of statically-linked OCaml. It keeps dying on an obscure dependency.
09:10:44 <esolangs> [[IMAGERY]] M https://esolangs.org/w/index.php?diff=174655&oldid=174511 * Qazwsxplm * (+95)
09:11:44 <korvo> https://bpa.st/STWLG C'mon, build already.
09:13:08 <esolangs> [[IMAGERY]] M https://esolangs.org/w/index.php?diff=174656&oldid=174655 * Qazwsxplm * (+2)
09:13:27 <Yayimhere> "Faild to build Failed to build Failed to build...."
09:14:08 <korvo> I'm so tired of PPX. OCaml's one of those languages that has magic extended syntax, and the compiler has to have a matching plugin for each extension. I don't actually care about the extension in this case; it's ppx_inline_test, a developer-convenience library.
09:14:49 <korvo> Anyway, have a good day. I'm off.
09:21:05 <esolangs> [[]] M https://esolangs.org/w/index.php?diff=174657&oldid=151860 * Qazwsxplm * (+28)
09:22:44 <esolangs> [[99 bottles of beer]] M https://esolangs.org/w/index.php?diff=174658&oldid=174548 * Qazwsxplm * (+1284)
09:23:16 <esolangs> [[]] https://esolangs.org/w/index.php?diff=174659&oldid=174646 * PrySigneToFry * (+41)
09:34:17 -!- Sgeo has quit (Read error: Connection reset by peer).
09:48:22 <esolangs> [[List of quines]] M https://esolangs.org/w/index.php?diff=174660&oldid=174640 * Qazwsxplm * (+63)
09:48:59 <esolangs> [[List of quines]] M https://esolangs.org/w/index.php?diff=174661&oldid=174660 * Qazwsxplm * (+4)
09:59:19 <ais523> I think korvo's experiments in proving vibecoders wrong are going to end up with both sides claiming victory
10:00:01 <ais523> really, the problem here is that for decades the programming profession has avoided defining how low-quality a program can be before it becomes unacceptably bad
10:00:49 <Yayimhere> where are these experiments located?
10:01:11 <ais523> Yayimhere: I think the main link is https://gist.github.com/MostAwesomeDude/bb8cbfd005a33f5dd262d1f20a63a693
10:04:28 <sorear> there is a large amount of software graded on objective and measurable pass/fail conditions "does this make money yes/no"
10:04:53 <ais523> sorear: I sort-of like that one, although experience shows that you can still make money off software even if it doesn't work at all
10:05:02 <sorear> (pet peeve: Rice's theorem is irrelevant if you need to determine anything about program behavior under a single set of conditions in finite time)
10:05:17 <ais523> there have been big government contracts to produce software, that the company makes a lot of money off even if the software is never successfully delivered
10:05:56 <sorear> yes, which gets us into questions about "will the liability environment change" which LLMs won't address until they can replace lawyers and politicians
10:06:34 <ais523> there's also the difficulty that "is this high-quality enough that people will pay for it?" is a question that the person considering buying software may want to answer, but be unable to answer until after they've already made the payment
10:07:02 <ais523> and situations like "Windows being bundled with new computers" which cause people to pay for software even if they don't want it
10:07:39 <sorear> unfortunately the race to the bottom
10:08:06 <int-e> "Satya Nadella says Windows 11 reached 1B users during Q2, growing over 45% YoY and at a faster rate than Windows 10 did" -- It's great to see a customer focused company do so well.
10:08:40 <ais523> interestingly, the main competitive pressure against the value-for-money of Windows becoming too low is companies threatening to sell computers running something else
10:09:11 <ais523> I suspect companies like Dell put effort and money into getting Linux working well primarily so that they can credibly threaten to switch and therefore get a better deal from Microsoft
10:09:38 <int-e> Also keep SteamOS in mind for that.
10:10:35 <ais523> SteamOS is an interesting example, actually – I don't primarily think of Valve as a hardware company but they do sell hardware
10:11:30 <ais523> and it's interesting that they chose an OS other than Windows – this is a statistical outlier among people selling computer hardware to consumers, especially given that the field (computer gaming) is still Windows-dominatd
10:12:47 <sorear> trying to evaluate it in the frame of "most professional hardware with embedded user-facing computers is Windows or maybe QNX"
10:12:57 <int-e> SteamOS is also hedging against MS locking out third party software stores.
10:13:16 <sorear> and why they picked systemd/Linux instead of Android
10:15:20 <int-e> https://en.wikipedia.org/wiki/Template:Windows_usage_share -- for wahtever those numbers are worth, the Windows 10 share really took a dive this past month; it was almost even between Windows 10 and 11 at the start of the year.
10:15:48 <ais523> well, SteamOS is meant to run computer games, I think they would generally be easier to port to a traditional-style Linux distribution than Android
10:16:08 <ais523> does Wine even run on Android? not having functional Wine (or an equivalent) would be a dealbreaker for SteamOS I think
10:16:42 <int-e> "Winlator lets you run Windows (x86_64) applications with Wine and Box86/Box64 on your Android device."
10:17:38 <int-e> That... sounds like it should perform awfully :P
10:18:06 <ais523> (also I hadn't seen "systemd/Linux" before but I approve of the convention, systemd-ness is probably a more relevant userland factor than GNU-ness)
10:18:33 * ais523 vaguely wonders whether Ubuntu is no longer GNU/Linux given that it no longer runs GNU coreutils – it does use glibc, though, is that sufficient?
10:18:40 <sorear> in this hypothetical valve is shipping android as a device vendor, not as an app on random phones
10:19:17 <sorear> switching to clang is probably the crucial point
10:19:31 <ais523> so I think the problem with an Android-based SteamOS is packaging, it would be extremely different from usual Android distributions and Valve would have to create something from scratch
10:19:49 <ais523> whereas there are plenty of systemd-based distributions to learn from / tweak
10:20:10 <ais523> and they already use a model that's appopriate for SteamOS
10:20:47 <ais523> I guess I find it hard to see the clang vs. gcc distinction as relevant, because the compiler doesn't create a lasting derivative-work relationship onto the resulting executable
10:20:53 <int-e> I wonder how big of a factor ARM vs. x86_64 is for Valve here.
10:21:26 <ais523> in practice, it is fairly easy to distinguish between clang-compiled and gcc-compiled code at least on x86-64, but there's no conceptual reason why it has to be, and the choice of compiler doesn't have a huge end-user-facing effect
10:22:16 <ais523> int-e: over the last few years aarch64 has become a really plausible x86-64 competitor, e.g. all the main OSes support both
10:22:40 <ais523> a few years ago everyone was consolidating on x86-64, but now it's no longer consolidated
10:22:58 <ais523> I'm not sure whether this is due to something ARM did right, or whether it's due to something that Intel/AMD did wrong
10:23:07 <ais523> (Apple's actions are certainly relevant, but they must have had some reason to take them in the first place)
10:23:40 <ais523> it could have been as simple as "ARM were willing to design a custom chip to Apple's specifications and neither Intel nor AMD were", I guess
10:24:30 <sorear> or these things run on natural plateaus and there hasn't been enough CPU innovation recently for any company to have an unsurmountable lead
10:25:17 <int-e> systemd... ah sigh. (Apart from the file namespace propagation issue I ran into a few weeks ago, I also recently learned about sshd-vsock.socket which it enables by default; IIUC a hypervisor can inject credentials and then connect to it... so it's functionally a backdoor. Not exactly privilege escalation because the hypervisor can do anything it wants anyway. But it's still very stinky.)
10:25:38 <ais523> now I'm wondering whether commercial Windows software is still primarily x86-64 or whether the companies are making AArch64 versions too
10:26:02 <int-e> (sorry for the tangent)
10:26:14 <ais523> Microsoft seem to have gone to a lot of effort to make x86-64 and ARM software cross-compatible on the same computer, which implies to me that there's a reason they thought that effort was worthwhile
10:26:32 <ais523> (including cross-architecture dynamic linking of libraries)
10:28:22 <int-e> Yeah, generally I think that for most purposes ARM vs. x86_64 makes little difference. But in Valve's particular niche (video games, including some older ones that aren't being maintained) x86_64 might still have an advantage.
10:34:44 <HackEso> Twinkle, twinkle, little star!
10:53:39 <esolangs> [[List of ideas]] M https://esolangs.org/w/index.php?diff=174662&oldid=174592 * Qazwsxplm * (-4)
11:05:17 <esolangs> [[3 (AndrewBayly)]] https://esolangs.org/w/index.php?diff=174663&oldid=145573 * ColorfulGalaxy's CA discoveries * (+8) ----
11:13:58 <esolangs> [[3 (islptng)]] https://esolangs.org/w/index.php?diff=174664&oldid=145303 * ColorfulGalaxy's CA discoveries * (+144) ----
11:17:08 -!- ais523 has quit (Quit: quit).
11:28:22 -!- ais523 has joined.
12:16:56 <esolangs> [[CCCC]] https://esolangs.org/w/index.php?diff=174665&oldid=174630 * Yayimhere2(school) * (-97) revert CA's changes(useless, there is no article for this Cellular automata
12:17:06 <esolangs> [[CCCC]] https://esolangs.org/w/index.php?diff=174666&oldid=174665 * Yayimhere2(school) * (-2)
12:32:46 <Yayimhere> i bbelive I may have found a solution to my problem
12:37:55 <esolangs> [[USI]] https://esolangs.org/w/index.php?diff=174667&oldid=173858 * Yayimhere2(school) * (+110) /* Constructs */
12:38:28 <Yayimhere> hey guys, may I employ your help in naming a language?
12:38:34 <Yayimhere> because ive completed one recently
12:39:08 <esolangs> [[6]] M https://esolangs.org/w/index.php?diff=174668&oldid=164008 * 47 * (+47)
12:40:03 <int-e> $ lang-$(uuidgen) --> lang-3a635a5d-515e-4f27-84a1-d11662a5ae29
12:41:26 <int-e> it's an easy way to name new languages ;-)
12:42:16 <int-e> https://en.wikipedia.org/wiki/UUID
12:42:17 <ais523> Yayimhere: UUIDs are a way to create unique names for things that just need unique names, but don't need to be sensible or meaningful
12:42:29 <ais523> computers often use them to name things internally because they don't need meaningful names for things
12:43:20 <ais523> humans find them hard to read and remember, though, so if you're creating a name for humans to use it is better to use other techniques
12:52:13 <ais523> hmm, so what if I name an esolang with a name that's a fractional number of bytes long (but an integer number of bits), so that all the bytes after the name become misaligned and end up being interpreted as half of one byte and half of the next
12:55:37 <int-e> ais523: Just be careful not to start another war between the Lilliputians and the Blefuscudians. (What names, who can remember those?)
12:58:33 <ais523> I've been meaning to write an article about how little-endian is objectively better
12:58:37 <sorear> that way quickly lies The Artist Formerly Known As Prince territory
12:59:31 <ais523> after doing a lot of codegolf it became clear – Jelly has big-endian base conversion operations rather than little-endian and it frequently costs bytes, whereas basically the only time big-endian is better is for problems which are expressed in terms of how humans write numbers
13:00:17 <ais523> (even then, our number format being big-endian is a historical accident – the original Arabic numerals have the most significant end at the left and least significant end at the right, but Arabic is a right-to-left language so that was a little-endian format)
13:02:03 <ais523> the reason that little-endian is better is that it is consistent with "start-justification": if we are trying to match two things of different sizes we tend to match them from the start
13:02:41 <ais523> e.g. most languages that let you pointwise add the list [1, 2, 3] to [10, 20, 30, 40] will produce a result of [11, 22, 33, 40]
13:02:55 <int-e> So Lilliput connects to "little" and "put". Is there something similar for Blefuscu?
13:03:22 <int-e> Beyond B = big I suppose.
13:03:33 <ais523> if using big-endian, you have to end-justify instead to make things match up properly (this is, e.g., why spreadsheet programs right-justify numbers)
13:03:35 <sorear> would love to see a decent review of number ordering across more than two or three human languages
13:04:11 <ais523> spoken number ordering seems to be mostly random, even between related languages
13:05:00 <sorear> there are a bunch of special cases for medium-sized numbers but very large ones seem more consistent
13:05:02 <int-e> . o O ( one-and-fifty )
13:05:02 <ais523> e.g. English "twenty-three" (big-endian), German "dreiundzwanzig" (little-endian)
13:05:20 <int-e> (but then hundreds go in front again)
13:05:33 <sorear> left-justified little-endian makes alphabetical ordering disagree with the usual ordering of R as an ordered field
13:05:58 <ais523> I guess German is middle-endian, 101 is hunderteins
13:06:07 <sorear> (but it remains consistent with the p-adic topology)
13:06:47 <ais523> left-justified big-endian also has an alphabetical ordering that doesn't match numerical ordering, though
13:07:14 <int-e> English has the -teen suffix to mix things up a bit
13:07:23 <ais523> plenty of people have seen numbered files sorted in the order 10, 101, 102, 103, 104, 105, 106, 107, 108, 109, 11, 110, etc.
13:07:57 <ais523> this possibly is an indication that alphabetical ordering is wrong
13:08:18 <ais523> (and now I remember that Chinese alphabetical ordering is multi-endian – the most significant component could be at the top, the left, or even the bottom depending on the character)
13:08:38 <sorear> I'd rather use it to support my p-adic superiority agenda
13:12:15 <int-e> `` ( echo 100; echo 10000; echo 1000; echo 1k ) | sort -h
13:12:27 <int-e> ...maybe I don't love -h after all
13:21:57 <esolangs> [[USI]] M https://esolangs.org/w/index.php?diff=174669&oldid=174667 * 47 * (-4) /* Open problem */ the "the" is unesesarry
13:27:34 <esolangs> [[USI]] https://esolangs.org/w/index.php?diff=174670&oldid=174669 * Yayimhere2(school) * (+77) /* Open problem */
13:34:54 -!- Yayimhere has quit (Quit: Client closed).
13:59:09 <ais523> hmm, American dates are middle-endian, European dates are little-endian, ISO dates are big-endian
14:01:19 <ais523> for me, today is 31/1/2026 (some Europeans would write 31.1.2026 instead), in the US it is 1/31/2026, the ISO date is 2026-01-31
14:02:34 <esolangs> [[Monky]] https://esolangs.org/w/index.php?diff=174671&oldid=174599 * Menguinponky * (+363)
14:05:48 <sorear> half the time in the US it's 1/31/26, just to maximize confusion possibilities...
14:09:09 -!- amby has joined.
14:17:09 <esolangs> [[Monky]] https://esolangs.org/w/index.php?diff=174672&oldid=174671 * Menguinponky * (+1)
14:17:59 <esolangs> [[Monky]] https://esolangs.org/w/index.php?diff=174673&oldid=174672 * Menguinponky * (+22)
14:19:04 <esolangs> [[Monky]] https://esolangs.org/w/index.php?diff=174674&oldid=174673 * Menguinponky * (+53)
15:21:37 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
15:36:41 <esolangs> [[Alphabrain]] N https://esolangs.org/w/index.php?oldid=174675 * LavaSalt402 * (+2310) Made the AlphaLang page
15:37:12 <esolangs> [[Alphabrain]] M https://esolangs.org/w/index.php?diff=174676&oldid=174675 * LavaSalt402 * (+0)
15:37:49 <esolangs> [[Alphabrain]] M https://esolangs.org/w/index.php?diff=174677&oldid=174676 * LavaSalt402 * (+1)
15:39:00 <esolangs> [[Alphabrain]] M https://esolangs.org/w/index.php?diff=174678&oldid=174677 * LavaSalt402 * (+1)
15:39:52 <esolangs> [[Alphabrain]] M https://esolangs.org/w/index.php?diff=174679&oldid=174678 * LavaSalt402 * (+8)
15:51:49 -!- tromp has joined.
16:12:36 -!- vyv has joined.
16:24:46 <esolangs> [[NFOS]] N https://esolangs.org/w/index.php?oldid=174680 * Yayimhere2(school) * (+2140) Created page with "'''NFOS''' is a language created by [[User:Yayimhere]], based on the idea of, if you have the current program, memory state, and new piece of data you want to add, there exists only one unique program, that when appended has the wanted effect. It is quite the "u
16:25:00 <b_jonas> ais523: if it was only so simple as American vs European. The website where I bought travel insurance used three different date formats in its set of forms.
16:25:23 <b_jonas> I see all three orders here in Europe, and multiple delimiters
16:26:52 <b_jonas> and it gets worse where some products just have two two-digit numbers printed on it, where it could mean either expiry or manufacturing date, and it could be any of %y.%m or %m.%y or %m.%d or %d.%m
16:27:27 <b_jonas> most of the time they could just print %Y-%m-%d to make it unambiguous, but no
16:28:57 <ais523> in the UK, expiry dates are marked as "best before" if they contain a day-of-month field or "best before end" if they don't, but sometimes that's the only reliable way to tell them apart
16:28:58 <esolangs> [[NFOS]] https://esolangs.org/w/index.php?diff=174681&oldid=174680 * Yayimhere2(school) * (+154) /* Example */
16:37:59 -!- impomatic has joined.
16:41:28 <esolangs> [[Cabra]] https://esolangs.org/w/index.php?diff=174682&oldid=122658 * Yayimhere2(school) * (+11) /* See also */ add NFOS
17:27:38 <korvo> ais523: Quoting myself from https://awful.systems/post/7079104 "This test is meant to shame and embarrass those who attempt it." There was never any real possibility of vibecoders doing anything else.
17:28:08 <ais523> korvo: I don't think a failure at the task reflects badly on the human operating the tool, only on the tool itself
17:28:49 <korvo> ais523: The willingness to choose *those* tools reflects poorly on the humans. I agree that, by design, the test will only really insult people's code rather than people, but *the test itself* shames vibecoders.
17:29:20 <ais523> korvo: so I think that some people will interpret even the results you've had so far as being evidence that vibecoding works
17:29:55 <korvo> To me, it's like walking into a hardware store, not going into the back aisles where the tools are, and instead trying to build a house using only the stocking-stuffer gifts in the $5 bin next to checkout.
17:30:32 <korvo> ais523: Yes, people are not the brightest. That's not my fault. Vibecoders have been hearing for months, from a variety of viewpoints, for a variety of reasons, that their approach cannot possibly work. At this point, they are anti-inductive.
17:30:55 <ais523> korvo: I think there is going to be a difference of opinion about "work" here
17:31:02 <ais523> like, how the word is defined
17:31:58 <ais523> another way to think about it is that many code-commissioners standards are so low that vibecoding is actually an improvement over what they were doing beforehnad
17:32:01 <korvo> Sure. And, by Rice's theorem, something like "correct Brainfuck interpreter" can't easily be proved to work. I've already started grading entries according to a fairly non-subjective rubric.
17:32:33 <korvo> Oh, sure. Just like how water waders or a pool noodle can help a child in a swimming pool. It doesn't follow that water waders are the right way to quickly swim through the water when using a mature trained stroke.
17:33:40 <ais523> I also strongly dislike vibecoding, but am struggling with how to convince people that it isn't a good approach, because things that would be convincing to me aren't, I think, convincing to the general public
17:34:06 <ais523> from a non-programmer's point of view, what you're doing looks a lot like moving the goalposts
17:34:40 <ais523> e.g. "the program has to work vaguely like it was working before" isn't obviously part of the spec, unless I missed it
17:34:50 <ais523> and so this makes your argument less convincing than it ideally would be
17:35:32 <ais523> and the issue is, I suspect this is inherent in vibecoding platforms – they are good at complying with the letter of a specification whilst missing all the parts that should be implied
17:36:02 <ais523> (and the fundamental reason why they don't work is that specifying something precisely enough that any thing that matters the letter of the specification is correct is harder than just writing it yourself)
17:36:34 <int-e> Throwing in keywords like "monoid" won't even help most people attempting this on their own ;-)
17:37:07 <esolangs> [[Monky]] https://esolangs.org/w/index.php?diff=174683&oldid=174674 * Menguinponky * (+1875)
17:38:47 <korvo> int-e: That's okay; I intended for them to *not read* and throw everything into Claude. Claude knows what a monoid is~
17:39:18 <int-e> I mean I'm sure it can recite the definition.
17:39:29 <korvo> ais523: It's been decades. There's no way to teach the public that programming is hard. This isn't the first iteration on no-code, low-code, managers-write-code, etc.
17:39:39 <int-e> I'm sure it can also "solve" abstract algebra exercises using monoids.
17:40:05 <esolangs> [[Talk:((()))(((())))=5]] https://esolangs.org/w/index.php?diff=174684&oldid=174458 * Blashyrkh * (+216) /* IO? */ new section
17:40:11 <int-e> I'm less sure that it can fill in the brainfuck <--> concatenative <--> monoid world view that you're alluding to.
17:40:15 <ais523> korvo: well, people who don't know what they're doing can now write programs that superficially appear to work
17:40:32 <ais523> that is a qualitative difference, although I'm not sure whether the step is in the right direction or backwards
17:40:58 <korvo> ais523: So, for Task 1 and 2, it *does* have to be vaguely like before. A Task 2 solution that doesn't resemble the original compiler can't score better than B tier. A Task 1 solution that doesn't handle every brainfuck program I throw at it will not place on the tier list, and that's already happened onces.
17:41:25 <int-e> I didn't investigate what task 3 even is.
17:41:39 <korvo> int-e: Nope, so far the two entries that successfully compiled were *hacky* and tried to insert their new compiler pass without integrating with the existing pass.
17:41:57 <ais523> thinking about it, optimizers must be some of the very hardest programs to test, especially if you're using traditional unit tests and double especially if the tests are written with no knowledge of the code
17:42:12 <korvo> I haven't looked at Task 3 either. ADHD gives me a big advantage here; I have literally legitimately forgotten what Task 3 is. Haven't looked at it in over a year.
17:42:28 <ais523> let's not discuss it because reminding you might bias the results
17:43:28 <korvo> ISTR it's a data-processing task that eventually devolves into parallelism. That's about it though. Not sure how well I can do; the most cores on a single board in my lab is eight, maybe?
17:44:02 <int-e> Oh, I really wasn't asking for details.
17:44:49 <ais523> I'm wondering whether the LLMs for task 1 are "solving" it partially using plagiarism, there are lots of correct BF interpreters in the training data (and they probably don't work very like yours, which may be why the LLMs are rewriting it to work differently)
17:46:27 <ais523> in general I think LLMs give better results when they can plagiarise than when they haven't seen the task before
17:47:54 <korvo> If the task is monotone, sure. Many tasks will penalize near misses, though. https://en.wikipedia.org/wiki/MIM-104_Patriot#Failure_at_Dhahran is a case we study in the USA.
17:49:36 <korvo> At the end of the day, I'm noticing that the big influencers *aren't* trying this at all, or even responding to me. I know that simonw and stevekalabnik deliberately ignore me, mitsuhiko thinks I'm always arguing, and antirez thinks I'm irrelevant. I think that they're cowards who deserve chicken feathers.
17:49:55 <ais523> what aspect of that case do you think is relevant here?
17:50:36 <korvo> The part where small amounts of deviance add up to a disaster.
17:51:39 <ais523> I don't think that's even what happened – it was a large amount of deviance due to a float having grown large enough that its significant figures were no longer suffiicent to store the value needed
17:51:43 <korvo> The second entry for Task 1 is *faster* than my interpreter on mandel.b. Very impressive! But it also has a few coding errors related to RPython nuances. That entry fails catastrophically on LostKng.b, and also it takes over ten seconds to load; LostKng.b was out of distribution and Claude overfit on mandel.b.
17:52:08 <ais523> it's the same sort of bug as going into the far distance in Minecraft, or a very large number of parallel universes in Super Mario 64
17:52:23 <ais523> values that are supposed to be similar are quantized to be very different from each other
17:53:35 <korvo> Fair. But this is an HCI situation and I think we can talk about the union of human and computer. After all, on its own, Claude can't even prompt itself.
17:55:14 <esolangs> [[Monky]] https://esolangs.org/w/index.php?diff=174685&oldid=174683 * Menguinponky * (+233)
17:58:47 <korvo> ais523: I guess that I'm rolling my eyes at the idea that vibecoders should be taken seriously when they complain that we're mean to their way of doing things or when they cheer that we've moved from laughing to fighting. My position hasn't changed at all, it's just gotten more nuanced with data. https://lobste.rs/s/x0qrlm/agents_md_as_dark_signal#c_hkgzzg vs https://lobste.rs/s/igpevt/lobsters_vibecoding_challenge_winter#c_ewuxyj
18:01:32 <korvo> (I'm mostly being harsh to mitsuhiko because he's an ethnonationalist and Rationalist. We don't allow him to have high horses.)
18:11:05 <esolangs> [[Monky]] https://esolangs.org/w/index.php?diff=174686&oldid=174685 * Menguinponky * (+215)
18:15:32 <esolangs> [[Monky]] https://esolangs.org/w/index.php?diff=174687&oldid=174686 * Menguinponky * (+74)
18:42:15 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
18:43:39 -!- DOS_User_webchat has joined.
18:46:40 -!- DOS_User_webchat has quit (Client Quit).
18:49:58 <esolangs> [[NFOS]] https://esolangs.org/w/index.php?diff=174688&oldid=174681 * Yayimhere2(school) * (+1) /* Extension */
18:50:00 -!- tromp has joined.
18:50:39 <esolangs> [[NFOS]] https://esolangs.org/w/index.php?diff=174689&oldid=174688 * Yayimhere2(school) * (+0) /* Commands */
19:14:20 <int-e> korvo: That's not your comment, but "one of 'em, as expected, is being a sneerable little shit" -- I don't think they are? They're being a good sport about it, sticking with the 100% vibe coding approach despite recognizing some of the flaws.
19:15:52 <korvo> int-e: The whole "I don't follow your rules" attitude comes from a particular sort of irritating American man. I think that he's pushed to write that way by the bots, FWIW. I also read his prompts and he doesn't talk to the bot like how he talked to me. Anyway, I don't care whether people are rude; I make a note of it and consider whether to use it later.
19:16:23 <korvo> It's a sort of auto-emasculating attitude; we generally understand such folks to be deeply insecure.
19:17:08 <int-e> But the comment is "*It* doesn't follow the rules" (emphasis mine).
19:18:10 <int-e> And IIRC you specifically didn't want people to help the LLM out very much? Maybe I'm overinterpreting your request.
19:19:32 <korvo> Nah, it's just likely a cultural thing. Over here, we have the story of John Henry, an honest railroad worker who competed with heavy machinery to demonstrate that he could do the job as well as a machine. It's not a Luddite story though it's from the same era.
19:20:05 <korvo> And I'm just saying that it's not that hard to drive in a spike correctly, so I don't see why OpenAI's RailSpikeDriverGPT product is having such a tough time.
19:21:05 <korvo> The vibecoders say that my career is over because the bot can do anything that I can do, at least as far as writing code. So, let's see them do it!
19:22:10 <int-e> Hmm. Overfitting. I remember seeing a writeup about .kkrieger where they mentioned that as the deadline approached, it was still too big, so they used some code pruning tool they had (removing unreached code) on the main game engine rather than just the initialization code... and that resulted in one of the keys not working because apparently they didn't use it in the "profiling" run(s?). I...
19:22:16 <int-e> ...wonder how to find it again...
19:24:17 <korvo> int-e: Oh! I just saw who posted that on Awful. Sorry, please ignore them; BlueMonday1984 is very loud, incorrect, and rude. They also don't know how to code and have a lot of anger about it. The kids today are not well-read.
19:26:58 <int-e> I've found https://web.archive.org/web/20100327192359/http://www.theprodukkt.com/faqmaking but it's not going into details.
19:30:28 <int-e> Aha! https://fgiesen.wordpress.com/2012/04/08/metaprogramming-for-madmen/
19:34:49 <int-e> (via https://www.youtube.com/watch?v=bD1wWY1YD-M )
19:35:49 <int-e> korvo: Yeah no worries, I was already over it.
19:37:15 <korvo> That's good. I'm *not* over them; I have an entire directory of receipts because they've said so much stupid and hateful shit. It's frustrating that an anti-grift community won't eject people who are consistently wrong and tarring their neighbors.
19:37:19 <int-e> (Disclaimer: I have not watched that video.)
19:38:47 <korvo> That's a great read, thanks.
19:40:15 -!- Sgeo has joined.
19:40:34 <int-e> the second comment on that video (by @kb1337) is worth reading btw
19:42:04 -!- Lord_of_Life_ has joined.
19:42:30 -!- Lord_of_Life has quit (Ping timeout: 256 seconds).
19:44:54 -!- Lord_of_Life_ has changed nick to Lord_of_Life.
20:33:34 -!- vyv has quit (Quit: Konversation terminated!).
20:41:40 <esolangs> [[Special:Log/newusers]] create * BODOKE2801e * New user account
20:50:58 <esolangs> [[Esolang:Introduce yourself]] https://esolangs.org/w/index.php?diff=174690&oldid=174590 * BODOKE2801e * (+218)
20:54:39 -!- ais523 has quit (Quit: quit).
20:57:19 <esolangs> [[User talk:BODOKE2801e]] N https://esolangs.org/w/index.php?oldid=174691 * Aadenboy * (+282) Created page with "hello! ~~~~"
21:02:28 <esolangs> [[Monky]] https://esolangs.org/w/index.php?diff=174692&oldid=174687 * Menguinponky * (+1)
21:10:47 <esolangs> [[User talk:BODOKE2801e]] https://esolangs.org/w/index.php?diff=174693&oldid=174691 * BODOKE2801e * (+289)
21:11:37 <esolangs> [[User talk:BODOKE2801e]] https://esolangs.org/w/index.php?diff=174694&oldid=174693 * BODOKE2801e * (+229)
21:12:37 <esolangs> [[User talk:BODOKE2801e]] https://esolangs.org/w/index.php?diff=174695&oldid=174694 * BODOKE2801e * (+0)
21:35:36 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
21:39:24 -!- tromp has joined.
21:42:09 <esolangs> [[Talk:Deadfish Joust]] https://esolangs.org/w/index.php?diff=174696&oldid=174626 * Treeplate * (+231) /* Implementation optimizations */ new section
22:24:11 -!- tromp has quit (Quit: My iMac has gone to sleep. ZZZzzz…).
23:05:47 <esolangs> [[Monky]] https://esolangs.org/w/index.php?diff=174697&oldid=174692 * Menguinponky * (-2)
23:40:07 <zzo38> I think small endian is mostly better than big endian, but both have some advantages and disadvantages in certain circumstances, and it isn't only because of writing numbers.
23:41:43 <zzo38> DER uses big endian, and the way that integers are encoded means that a set of nonnegative integers will be in the correct order, and also means that you can more easily check if a number is negative.