--- Log opened Thu Feb 23 00:00:52 2023
00:45 -!- HumanG33k [~HumanG33k@dau94-2-82-66-65-160.fbx.proxad.net] has joined #hplusroadmap
00:47 -!- stipa [~stipa@user/stipa] has quit [Ping timeout: 252 seconds]
00:49 -!- stipa [~stipa@user/stipa] has joined #hplusroadmap
02:41 < hprmbridge> PlanC> Gm
04:35 < hprmbridge> kanzure> "ok, so what should we do about it?" https://twitter.com/elonmusk/status/1628086895686168613
04:49 < kanzure> brlcad summer of code is back this year https://opencax.github.io/ (there is some speculation that this might be the last year because of some structural changes at google)
05:01 < kanzure> here is my tweet thread overview https://twitter.com/kanzure/status/1628741755683192833
05:04 < kanzure> this opencascade emscripten javascript version would be interesting to use in a weekend project sometime https://ocjs.org/
05:09 < kanzure> sam zeloof's atomicsemi.com company discussed here https://news.ycombinator.com/item?id=34907885   the e-beam lithography stuff being slow doesn't seem like a problem to me, wasn't someone going to mass produce high-parallel e-beam MEMS chips?
05:10 < kanzure> so the speed wouldn't matter if you could just have a bunch of cheap electron beam scanners
06:11 < muurkha> .t https://twitter.com/elonmusk/status/1628086895686168613
06:12 < muurkha> .tw https://twitter.com/elonmusk/status/1628086895686168613
06:13 < muurkha> fenn: a random access to SDRAM still costs close to 100ns typically
06:15 < muurkha> 04:32 < fenn> if you ran a CPU at 300Hz it wouldn't generate that much heat
06:16 < muurkha> it depends on how it's implemented.  currently popular CMOS spends over 1% on leakage
06:58 < docl> interesting thing about biological brains is they seem a lot more energy efficient. I don't expect this to hold true for the indefinite future, but beating the big server farms to market is kind of appealing
07:04 < superkuh> It is not clear to me how you'd have optics for every different electron beam operating at the same time.
07:06 < superkuh> I guess they could literally be separated by conductive baffles with the actual electrical field and magnetic field stuff built into the inside of each rectangular "tube" that goes all that way to the chip and each column operates entirely separately and only does it's own part of the chip. But that's definitely not off the shelf or cheap.
08:03 < jrayhawk> there are a bunch of ReRAM NVDIMM products in the works: fujitsu, Micron, Sony...
08:03 < jrayhawk> I think Intel only sells the motherboards to select partners under NDA at this point, though.
08:08 < kanzure> well for manufacturing chips i was thinking separate e-beam units that are independent; the highly parallel e-beam on a single chip i guess is unnecessary.
08:09 < kanzure> the question would be can you scale the production of e-beam vs scaling the production of high precision photomask photosteppers
08:10 < kanzure> docl: is that true actually, like the cost per query of chatgpt (and maybe include quality) vs human cost per query or quality of the human response, in terms of energy cost
08:11 < lsneff> I saw a startup from a few years ago building massively parallel proton beams for chip fab
08:20 < lsneff> I wonder if a very large array of stm probes, which can not move independently, but can be biased independently, could be slowly scanned across a surface to expose the resist very precisely
08:22 < lsneff> Effectively millions of e-beams that only point exactly in +z
08:25 < lsneff> You’d put billions on a single silicon wafer probably
08:27 < kanzure> https://techcrunch.com/2023/01/10/openai-in-talks-to-back-zeloof-and-chip-legend-kellers-startup-at-100-million-valuation/
08:28 < kanzure> https://openai.fund/
08:33 < kanzure> wasn't halcyon molecular's whole thing "we are going to make electron beam machines extremely cheap by mass producing e-beam machines"
08:35 < lsneff> Better if they just have to be biased electrodes
09:06 < docl> kanzure: I'm thinking probably human is more expensive in the sense of per capita energy consumption. our bodies use 5x as much as our brains alone (10 MJ/day vs 2 MJ/day), and global energy consumption (200 MJ/day/person) is 100x. higher if you narrow to industrialized countries. but it might be possible to run a fruit fly or pygmy frog farm a lot closer to the theoretical optimum, even if their brain 
09:06 < docl> to body ratio isn't as good
09:22 < kanzure> bee breeding https://www.youtube.com/watch?v=R8-9DgXcrfI&t=15m30s
09:28 < kanzure> https://www.youtube.com/watch?v=vDemV5JVvWk&t=6m
09:54 -!- cthlolo [~lorogue@77.33.23.154.dhcp.fibianet.dk] has joined #hplusroadmap
10:47 < muurkha> .t https://www.ims.co.at/en/products/
10:48 < muurkha> (multiple electron beams)
11:28 < kanzure> hacker news discusses embryo editing https://news.ycombinator.com/item?id=34897065
12:06 -!- HumanG33k [~HumanG33k@dau94-2-82-66-65-160.fbx.proxad.net] has quit [Quit: WeeChat 3.0]
12:51 -!- HumanG33k [~HumanG33k@dau94-2-82-66-65-160.fbx.proxad.net] has joined #hplusroadmap
12:54 -!- cthlolo [~lorogue@77.33.23.154.dhcp.fibianet.dk] has quit [Quit: Leaving]
13:09 < hprmbridge> nmz787> https://heidelberg-instruments.com/core-technologies/thermal-scanning-probe-lithography/
13:29 < kanzure> "Multimodal monitoring of human cortical organoids implanted in mice reveal functional connection with visual cortex" https://www.nature.com/articles/s41467-022-35536-3
13:29 < hprmbridge> nmz787> My guess is atomicsemi is aiming to produce chips in rapid prototype quantities.
13:30 < hprmbridge> nmz787> You don't need to innovate multi beam e-beam systems, or go into EUV
13:30 < kanzure> "... human cortical organoids transplanted into the retrosplenial cortex of adult mice. Two-photon imaging shows vascularization of the transplanted organoid. Visual stimuli evoke electrophysiological responses in the organoid, matching the responses from the surrounding cortex. Increases in multi-unit activity (MUA) and gamma power and phase locking of stimulus-evoked MUA with slow ...
13:30 < kanzure> ...oscillations indicate functional integration between the organoid and the host brain. Immunostaining confirms the presence of human-mouse synapses."
13:30 < hprmbridge> nmz787> HVM manufacturing could be 30k wafers a week for a single product
13:31 < hprmbridge> nmz787> Producing a few chips per week at a cost that patreon could pay for, or whatever significantly lower barrier to entry... Has value
13:32 < kanzure> and:
13:32 < kanzure> ".. Whether brain organoids can compensate for injured brain regions remains unclear. In this issue of Cell Stem Cell, Jgamadze et al. report that human forebrain organoids transplanted into the rat visual system show long-term structural connectivity and the restoration of visual function following lesions."
13:33 < kanzure> "Structural and functional integration of human forebrain organoids with the injured adult rat visual system" https://www.cell.com/cell-stem-cell/fulltext/S1934-5909(23)00004-8
13:35 < kanzure> "Reestablishment of damaged adult motor pathways by grafted embryonic cortical neurons" https://www.nature.com/articles/nn1970
13:42 < kanzure> this is a bizarre one.. human brain organoids in the structure of a transplantable laminar cortical shape https://www.sciencedirect.com/science/article/pii/S2589004219303839
13:57 < docl> I don't know much about e-beams, but if you have many electron beams close together it seems like they affect each others trajectory by repulsion. if that's predictable, you could account for it but my intuition is that it makes noise because electron positions are probablistic?
13:59 < kanzure> "Advances in construction and modeling of functional neural circuits in vitro" https://link.springer.com/article/10.1007/s11064-022-03682-1
14:00 < kanzure> has some interesting things about neurospheroids, "connectoids" (interconnected cortical organoids), and the "campenot chamber" for axonal growth directed by microslits (1977)
14:03 < kanzure> "Physical reservoir computing with FORCE learning in a living neuronal culture" https://aip.scitation.org/doi/full/10.1063/5.0064771
14:03 < kanzure> this abstract claims maze solving with an in vitro living neuronal culture
14:08 < kanzure> they cite a study on using neural reservoir computing for classification of spoken digits https://arxiv.org/abs/2210.02913
14:10 < kanzure> https://en.wikipedia.org/wiki/Reservoir_computing
14:34 < kanzure> from hn: "Germline editing has the chance to relieve the suffering of real people. My entire life would likely to be quite different and much better, were it available back then. And yet a lot of people are just ... counted as part of the cost. Sorry you will die soon, and painfully, but at least we're not the Nazis."
14:42 < fenn> "The medical industry seems to be largely divided into two camps: Camp A would happily sell allergy medication that triples your risk of diabetes, and Camp B would rather ten thousand terminal patients die in agony than kill a hundred of them with medication that cured the rest."
14:42 < fenn> it's the people without allergies that think allergy medication is something trivial and unimportant
14:54 < hprmbridge> kanzure> "Tracking neural activity from the same cells during the entire adult life of mice" https://www.nature.com/articles/s41593-023-01267-x
14:56 < L29Ah> i was prescribed https://en.wikipedia.org/wiki/Mometasone for my allergic rhinitis
14:56 < L29Ah> while it's the most effective of all interventions so far, the side effects are appalling
15:05 < TMA> I have encountered similar efficiency with https://en.wikipedia.org/wiki/Fluticasone_propionate
15:06 < TMA> however the candidiasis that was the side effect was much more severe
15:38 < hprmbridge> kanzure> pmetzger's argument against doom https://mobile.twitter.com/perrymetzger/status/1628885015684608002
15:45 < hprmbridge> kanzure> https://twitter.com/heyorson/status/1628662955821735937
15:48 < hprmbridge> kanzure> development of short sleeper phenotype after brain surgery https://mobile.twitter.com/Willyintheworld/status/1628631029710815234
17:21 < lsneff> oh man, I really want to be a short sleeper
17:45 < hprmbridge> kanzure> landed: https://www.cnn.com/2021/06/22/health/short-sleep-gene-wellness-scn/index.html
17:45 < hprmbridge> kanzure> oops, meant lsneff
17:48 < L29Ah> go irc
17:55 < L29Ah> bought 5 years worth of vitamin D given 4kIU/d, now i wonder if it spoils much at rt
17:57 < fenn> i wonder how short sleepers weren't recognized until the 21st century
17:57 < fenn> i mean this should be pretty obvious just by observation
17:58 < fenn> "short sleepers were ambitious, type A personalities, but also incredibly positive, outgoing and optimistic"
17:58 < fenn> so nancy kress wasn't just making this up for a cool sci-fi story?
18:00 < hprmbridge> kanzure> it was probably wrapped up in protestant work ethic moralizing ("you're just immoral/lazy")
18:02 < L29Ah> i heard about short sleepers in .ru before 21th century
18:02 < fenn> these mormon short sleepers seem to breed like rabbits; how common is the phenotype globally?
18:02 < L29Ah> probably was featured on some magazine or tv show my parents consumed
18:04 < hprmbridge> kanzure> are unusually long sleepers more happy than the average population? or less? my intuition says less.
18:04 < L29Ah> happiness is socially-defined
18:05 < hprmbridge> kanzure> that doesn't explain downs
18:06 -!- Croran [~Croran@c-73-118-187-65.hsd1.wa.comcast.net] has quit [Read error: Connection reset by peer]
18:06 < L29Ah> also doesn't explain `while :; echo im happy; done`
18:07 < L29Ah> still not sure if it's an useful metric
18:07 < fenn> long sleep correlates with depression
18:07 < L29Ah> suicide risk sounds more interesting
18:07 < fenn> chronic depression doesn't have a higher suicide risk afaik
18:08 < L29Ah> > New data on depression that has followed people over long periods of time suggests that about 2 percent of those people ever treated for depression in an outpatient setting will die by suicide.
18:09 < L29Ah> > In the case of depression, studies have shown that, on average, the risk of suicide is about 15 times higher than the average for the general population.
18:10 < fenn> i'm seeing 3.6x
18:10 < fenn> oh wait, that's ADHD
18:10 < fenn> bleh
18:10 < fenn> it's all BS anyway
18:11 < L29Ah> i'm pretty sure suicides aren't BS
18:11 < fenn> the categories are not based on science
18:12 < fenn> anyway, i'd contend that happiness is not completely made up
18:12 < L29Ah> happiness is more made up than clinical depression
18:13 < L29Ah> clinical depression is more made up than suicide
18:14 < fenn> i believe in the existence of subjective reality. there.
18:15  * fenn points at the "no philosophy" sign
18:18 < fenn> why haven't short sleepers outbred the rest of the world since prehistory?
18:18 < fenn> agriculture demands work, and more hours to work means more agriculture gets done, more food means more children
18:26 < hprmbridge> kanzure> maybe it's recent?
18:26 <+gnusha> https://secure.diyhpl.us/cgit/diyhpluswiki/commit/?id=55d7246b fenn: ADRB1 and CNN puff piece on familial short sleepers >> http://diyhpl.us/diyhpluswiki/genetic-modifications/
18:27 < hprmbridge> kanzure> yeah wikipedia lists a number of new short sleeper mutations found since 2009
18:33 < lsneff> god my mom has so much stuff
18:33 < lsneff> we have boxes of books in several storage spaces in several states
18:33 <+gnusha> https://secure.diyhpl.us/cgit/diyhpluswiki/commit/?id=0341709d fenn: try to work around problam parsing URL containing parentheses >> http://diyhpl.us/diyhpluswiki/genetic-modifications/
18:34 < fenn> eyy it worked
18:36 < hprmbridge> kanzure> pretty weird news article, some sort of advertisement for mormon genetic supremacy?
18:38 < fenn> maybe it's true
18:39 < hprmbridge> kanzure> my argument to someone the other day was that this is the simplest and closest thing to human intelligence enhancement and longevity (at least 5-10 more years of wakefulness and applying your brain; plus the reported better memory)
18:39 < fenn> mormons are obsessed with keeping genealogical records, so that's an obvious place to do research on heredity
18:40 < fenn> it's 5-10 years of time-travel-enhanced wakefulness
18:41 < fenn> you get more time in a day than competitors
18:41 < fenn> that's better than the equivalent amount of longevity (probably)
18:42 < hprmbridge> kanzure> oh because end-of-life life extension years would be backloaded?
18:42 < fenn> also it's less overhead for all the daily chores
18:44 < fenn> i dunno, the fact that it's so rare is suspicious
18:44 < hprmbridge> kanzure> wonder if the surgical result is repeatable https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3899325/
18:45 < L29Ah> 03:18:52]<fenn> agriculture demands work, and more hours to work means more agriculture gets done, more food means more children
18:45 < L29Ah> pretty sure it's problematic to do agricultural work during the dark time of day
18:45 < fenn> maybe they fare really poorly during plagues or starvation times
18:45 < L29Ah> 03:33:42]<lsneff> we have boxes of books in several storage spaces in several states
18:45 < L29Ah> i have boxes of boxes
18:45 < fenn> you should throw out your boxes of boxes
18:45 < L29Ah> in multiple countries even, until the recent times
18:46 < fenn> unless there's stuff in the inner boxes
18:46 < L29Ah> fenn: nah i sometimes fill some of these boxes with not-boxes
18:47 < L29Ah> otoh i have 0 books
18:47 < L29Ah> boxes > books
18:47 < hprmbridge> kanzure> maybe email the researcher who found the short sleeper gene and is studying those families. they might have a thought about prevalence.
18:47 < fenn> kanzure: looks like a pretty scary surgery, they have to shove a big tube through your brain to get to the center
18:47 < L29Ah> ok time to sleep
18:48 < fenn> also, sometimes people with drainage shunts just suddenly die for no apparent reason
18:54 < fenn> should we have a wiki page for surgical enhancements
18:55 < fenn> it seems like it could be an endless drama source
18:55 < hprmbridge> kanzure> surgical techniques would be more interesting
18:56 < hprmbridge> kanzure> like the double heart transplants, ovary arm transplant, IVF, xenotransplants, healing, regeneration, etc
18:57 < hprmbridge> kanzure> re: the above hn lament "but at least we're not Nazis" https://twitter.com/perrymetzger/status/1628083695138615314
18:57 < hprmbridge> kanzure> "It is very convenient that the Nazis existed, because without them, it would be hard to find a good way to describe people we disagree with."
18:59 < hprmbridge> kanzure> i think a lot of single cell surgical interventions tend to be overlooked, like injection, electroporation, nuclear transfer, ooplasm/cytoplasmic transfer, somatic cell fusion, etc
18:59 < hprmbridge> kanzure> the xenobot surgical manipulation of individual cells and just pasting/fusing them together...
19:00 < hprmbridge> kanzure> organ transplantation in general
19:01 < hprmbridge> kanzure> engraftment of stem cells into cerebral tissue, those wacky human-mouse synapses from earlier today. CAR-T and blood stuff might count as surgical.
19:03 < hprmbridge> kanzure> the surgical implantation of a functioning third eye in tadpoles (mentioned in earlier 2023 logs)
19:03 < L29Ah> installing more cooling capacity for the brain
19:03 < hprmbridge> kanzure> unless you're typing from a lucid dream this is not sleep
19:04 < hprmbridge> kanzure> https://xkcd.com/269/
19:07 < hprmbridge> kanzure> is there a word for sleep signing like there is for sleep talking (somniloquy)
20:12 < kanzure> this 2019 article has a good review of biological reservoir computing https://www.sciencedirect.com/science/article/pii/S0893608019300784#sec8 (section 8)
21:30 < muurkha> L29Ah: some agricultural work is routinely done during the dark time of day
21:56 < fenn> (this happened a year ago) https://theconversation.com/first-gene-therapy-for-tay-sachs-disease-successfully-given-to-two-children-176870
23:15 < fenn> in japan, 'copyrighted products can be recorded or adapted to the extent necessary without the consent of the copyright holder if it is for the purpose of “data analysis"'  https://storialaw.jp/en/service/bigdata/bigdata-12
23:15 < fenn> '(meaning the extraction, comparison, classification, or other statistical analysis of language, sound, or image data, or other elements of which a large number of works or a large volume of data is composed)'
23:16 < fenn> in the first place, i thought copyright infringement only happens if you redistribute the copyrighted material
23:18 < fenn> i suppose if you want to just buy a training data set, this would matter
--- Log closed Fri Feb 24 00:00:53 2023