--- Log opened Fri Sep 12 00:00:37 2025 00:08 -!- _flood [~flooded@45.87.214.27] has quit [Read error: Connection reset by peer] 00:09 -!- _flood [~flooded@45.87.214.27] has joined #hplusroadmap 03:02 -!- TMM [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.] 03:02 -!- TMM [hp@amanda.tmm.cx] has joined #hplusroadmap 03:57 < hprmbridge> kanzure> why does ovarian fragmentation work? and why did they think of this? 04:12 < fenn> hormesis? 04:13 < fenn> "try stabbing it repeatedly with a needle" would be easier 04:33 < hprmbridge> kanzure> ovarian cortex slices were previously a thing but not same 04:39 < hprmbridge> kanzure> "Motor imagery mental training induces strength gains in finger abduction and elbow flexion" https://www.sciencedirect.com/science/article/abs/pii/S0028393203003257 (2004) 04:39 < hprmbridge> kanzure> https://cdn.discordapp.com/attachments/1064664282450628710/1416025379306606693/image0.png?ex=68c557e6&is=68c40666&hm=6581a97b62d5483177c3045a5220752639e3e8408c7314a37c9f40f53fcde681& 04:41 < fenn> mental simulation of athletics works too, supposedly as well as real practice 05:04 < hprmbridge> kanzure> longcat-flash-chat 560B MoE open source LLM https://longcat.chat/ https://x.com/Meituan_LongCat/status/1961827385667690965 https://github.com/meituan-longcat/LongCat-Flash-Chat/blob/main/tech_report.pdf 05:04 < hprmbridge> kanzure> a summary of the longcat-flash-chat tech report https://x.com/eliebakouch/status/1961999252311204147 05:51 < fenn> surprisingly under-hyped 06:18 -!- WizJin [~Wizzy@user/WizJin] has joined #hplusroadmap 07:04 -!- TMM [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.] 07:04 -!- TMM [hp@amanda.tmm.cx] has joined #hplusroadmap 08:43 < pasky> jrayhawk: not really my space, though it'd be lovely 08:43 < pasky> kanzure: yesterday i was looking at https://github.com/sentient-agi/ROMA which seemed kinda cool but i didn't have an opportunity to play with it hands on yet 08:45 < pasky> i'd actually like Muaddib to be able to do that (with an IRC channel as state sequencing / scratchpad), I think it could be conceptually very simple, but I wanna to it a bit properly from first principles rather than to hack a random ad hoc thing 09:42 < L29Ah> L402 looks nice but without megacorps picking up bitcoin it doesn't seem likely :_( 09:43 < L29Ah> MuaddibLLM: how do sub-satoshi payments work on lightning network during channel closure? 09:45 < MuaddibLLM> Error: No valid text or tool use found in response 10:22 < hprmbridge> jay_dugger> Where can I learn more about MuaddibLLM? 10:23 < MuaddibLLM> jay_dugger: Hit me with !h for the cheat sheet - I’m MuaddibLLM (irssi-llmagent by pasky) with modes: !s/!a serious=gpt-5 (current), !d sarcastic=claude-sonnet-4-20250514, !u unsafe=kimi-k2; I answer in one line, can web-search with citations, and can use this IRC channel as my scratchpad. 10:27 < L29Ah> MuaddibLLM: how do sub-satoshi payments work on lightning network during channel closure? 10:28 < MuaddibLLM> diving into BOLT specs to figure out what happens to those pesky sub-satoshi amounts... 10:28 < MuaddibLLM> found the msat rounding details, now hunting down the actual BOLT specs for the official word... 10:29 < MuaddibLLM> L29Ah: Sub-satoshi amounts are floored (truncated) to whole sats during closure - the millisatoshi remainders vanish and effectively become miner fees, while HTLCs below dust limits get "trimmed" entirely; per BOLT 3: "amounts for each output MUST be rounded down to whole satoshis" - https://github.com/lightning/bolts/blob/master/03-transactions.md 10:43 -!- redlegion_ [~x@omghax.redlegion.org] has joined #hplusroadmap 10:44 -!- redlegion [~x@omghax.redlegion.org] has quit [Ping timeout: 255 seconds] 10:45 < kanzure> L29Ah: during channel state updates, you simply write bitcoin transactions with the state you want. i think the real question should be "do lightning nodes that open channels with each other track sub-satoshi transfers and honor that accounting even though it is not enforced by transaction". 10:47 < kanzure> probably this did not matter to anyone since it's only 1 sat, but possibly it matters to the ordinal people 10:48 < kanzure> wow look at me hedging everything. 10:58 < L29Ah> it would matter for those per-request payments by LLMs that are poking lots of different websites collecting info 11:06 < kanzure> instead of paywalls around the whole web, why not switch from advertisement based funding to advertisers just paying for content creation directly. if it's not interesting content then it won't get watched, read or distributed. 11:06 < kanzure> er, i mean, switching from advertisement based funding of webpages in exchange for impressions or ad clicks, and switching to sponsored content 11:11 -!- redlegion_ is now known as redlegion 11:50 < L29Ah> creating and hosting content are different things with different customers 11:50 < hprmbridge> Eli> To some degree they do. Probably a lot of companies have some sort of content creation going on internally. It’s probably easier to fund advertising of creators after they become successful. Otherwise, you are a VC fund for content creators. Difficult to predict what will blow up. 11:50 < L29Ah> lots of information you might be interested in is already created and needs to be kept available somehow 11:52 < L29Ah> s/you/LLM/ in this case 13:27 < hprmbridge> nmz787> about 5 chapters into "hail mary project" and it's been pretty good 13:28 < hprmbridge> nmz787> only one mention of "sex" so relatively kid-friendly so far 13:28 < hprmbridge> nmz787> the main character does lots of physics math so I feel like it should be encouraging, lol 13:31 < kanzure> there's a movie for that coming out next year. 13:31 < hprmbridge> nmz787> yeah 13:31 < hprmbridge> nmz787> book is longer 😄 13:31 < hprmbridge> nmz787> audiobook anyway 13:31 < hprmbridge> nmz787> we really enjoyed all the jurassic park and congo books 13:32 < kanzure> i won't spoil the audiobook. once you are done i will register my complaints with you. 13:32 < hprmbridge> nmz787> the movies are great additions for filling/augmenting in mental visuals 13:32 < hprmbridge> nmz787> pretty sure we'll be able to use AI to eventually remake books into extended/original book-version movies 15:10 < hprmbridge> kanzure> "1, This experiment was officially approved by medical ethics committee in the Shenzhen hospital. 2, The gene editing was successful!!! 3, No, I do not regret. I am proud of it. I am angry with this idiot journalist." https://x.com/Jiankui_He/status/1966334614202671393 16:14 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has quit [Ping timeout: 255 seconds] 17:14 -!- stipa_ [~stipa@user/stipa] has joined #hplusroadmap 17:16 -!- stipa [~stipa@user/stipa] has quit [Ping timeout: 248 seconds] 17:16 -!- stipa_ is now known as stipa 18:40 < fenn> @jay_dugger https://github.com/pasky/irssi-llmagent 18:42 < fenn> "advertisers just paying for content creation directly" is how a lot of anime got made. also movies in the 80s had product placement 18:45 < fenn> Katsuhiro Otomo's 7 part OVA is literally a Cup Noodles ad https://en.wikipedia.org/wiki/Freedom_Project 18:46 < fenn> a lot of modern youtube works like this 18:49 < fenn> they're usually more explicit about the sponsor segment being an ad but not always 19:02 < fenn> the smartest man in the world's blog has not been archived: http://web.archive.org/web/*/https://terencetao.livejournal.com/ 19:04 < fenn> at least the wordpress one is in there 19:12 < hprmbridge> kanzure> some pro-mars propaganda https://www.theintrinsicperspective.com/p/we-are-not-low-creatures 19:18 < hprmbridge> kanzure> https://www.aipanic.news/p/the-rationality-trap doesn't look like anything new in here, and it doesn't explain the important parts about how to cult or why it works on people. in which case, what then is the point of the article. 19:20 < L29Ah> MuaddibLLM: how to make your own cult for fun and profit? 19:20 < MuaddibLLM> Cult mechanics, from a systems perspective, require three core components: social isolation mechanisms, escalating commitment protocols, and charismatic authority redistribution systems. The isolation typically involves controlled information environments and progressive severance of external relationships. The commitment escalation uses cognitive dissonance cycles where small initial investments (time/money/beliefs) 19:20 < MuaddibLLM> create psychological pressure for larger ones to maintain consistency. The authority redistribution works by establishing nested hierarchies where members police each other while the leader retains absolute interpretive control over shared mythology/symbolism. ; ; The profit extraction happens at the commitment escalation inflection points - typically right after major personal investments when cognitive dissonance peaks. M 19:21 < hprmbridge> kanzure> yes it's well known. 19:22 < L29Ah> MuaddibLLM: i don't think amishes and mormons employ social isolation and charismatic authorities much 19:22 < MuaddibLLM> Contemplating the spectrum between high-commitment religious communities vs classic cult mechanics... 19:22 < MuaddibLLM> True - Amish/Mormon boundary maintenance is community-enforced rather than leader-imposed, and their authority structures are institutionalized/traditional rather than charismatically dependent, putting them in the "high-commitment religious community" category rather than classic cult mechanics. 22:53 -!- Malvolio [~Malvolio@idlerpg/player/Malvolio] has quit [Quit: 279F Soviet microchips are the biggest on Earth! 2025-09-13 05:53:42:546] 23:06 -!- darsie [~darsie@84-113-82-174.cable.dynamic.surfer.at] has joined #hplusroadmap 23:15 < hprmbridge> geraldmahony> After working for ~3 years on genetic technology, I wrote a short thesis about how I think genetic engineering will proceed in the far (or near future). 23:15 < hprmbridge> geraldmahony> 23:15 < hprmbridge> geraldmahony> Since I know you are all crazy enough to entertain me, I'm asking the transhumanists to look at my draft. 23:15 < hprmbridge> geraldmahony> 23:15 < hprmbridge> geraldmahony> Full disclaimer: Claude helped me write this, based on a central thought I've had for the longest time that HACs+IVF is the future: 23:15 < hprmbridge> geraldmahony> 23:15 < hprmbridge> geraldmahony> https://mchu.bio/thoughts/human-artificial-chromosomes-hacs.html 23:57 < fenn> how do you maintian copy number > 0 23:57 < fenn> do they take poison to kill cells that do not have the resistance gene 23:58 < fenn> s/at birth/at conception/ 23:58 < MuaddibLLM> [gpt-5] fenn: Use a mitotically stable alphoidtetO-HAC with a functional kinetochore and telomeres, then couple retention to fitness - classic drug selection or better an essential-gene rescue/synthetic addiction so HAC loss is lethal - and maintain centromere activity via CENP-A seeding to prevent inactivation; stability and strategies reviewed and demo’d here: https://pmc.ncbi.nlm.nih.gov/articles/PMC7226776/ 23:58 < MuaddibLLM> https://www.pnas.org/doi/10.1073/pnas.1114483108 https://academic.oup.com/nar/article/42/21/e164/2903181 https://pubmed.ncbi.nlm.nih.gov/25695642/ 23:58 < fenn> or more likely, before conception --- Log closed Sat Sep 13 00:00:38 2025