2026-03-24 04:38:45 <tpbsd       > wow, starlink now has over 20 million users and 10,000 satellites
2026-03-24 06:27:05 <veltas      > tpbsd: https://www.scientificamerican.com/article/as-starlink-and-other-satellites-proliferate-astronomers-learn-to-manage/
2026-03-24 06:27:25 <veltas      > I link for the interesting photos but I guess the article is a bit relevant too
2026-03-24 06:28:17 <veltas      > What can you do though, starlink is genuinely useful/valuable
2026-03-24 06:28:18 <tpbsd       > anyone who thinks that earthbound astronomy has a future is sadly mistaken I think
2026-03-24 06:28:43 <veltas      > I think it's got a future but there's just going to be more noise in the data
2026-03-24 06:29:31 <tpbsd       > I can give you about 100 trillion reasons it has no future
2026-03-24 06:29:41 <tpbsd       > reason$
2026-03-24 06:30:44 <veltas      > I mean as I understand it there's no reason they can't "remove" the streaks, it just will take longer to get the data they need
2026-03-24 06:31:53 <tpbsd       > yeah but musk wants another million satellites as well for a orbital AI
2026-03-24 06:32:02 <veltas      > I was saying the other day I managed to look at Jupiter's largest moon with my binoculars, a satellite didn't break into my house and beat me up to prevent me from doing so
2026-03-24 06:32:13 <veltas      > The orbital AI thing makes no sense, I don't think that will happen
2026-03-24 06:32:34 <tpbsd       > I think it will
2026-03-24 06:32:40 <tpbsd       > Im sure it will
2026-03-24 06:32:49 <veltas      > Why?
2026-03-24 06:32:58 <tpbsd       > unless musk goes broke or dies
2026-03-24 06:33:06 <tpbsd       > 1) power
2026-03-24 06:33:17 <tpbsd       > 2) room
2026-03-24 06:34:05 <tpbsd       > 3) exclusiveness
2026-03-24 06:34:30 <tpbsd       > he gets power for free, and unlimited once the gear is made
2026-03-24 06:34:53 <tpbsd       > a million sats barely takes up the available space
2026-03-24 06:35:02 <veltas      > Downsides: cooling, cost to launch, latency, hardware gets outdated and needs relaunching constantly
2026-03-24 06:35:39 <tpbsd       > he has dealt will all those issues with starlink no ?
2026-03-24 06:35:46 <veltas      > Not at all
2026-03-24 06:35:52 <veltas      > Starlink isn't a datacenter
2026-03-24 06:35:55 <tpbsd       > and they only tlk among themselves with lasers
2026-03-24 06:36:09 <tpbsd       > this isnt either, it's AI
2026-03-24 06:36:09 <veltas      > The power benefits in orbit are very minimal compared to the massive launch costs and cooling will prevent you from using the power fully anyway
2026-03-24 06:36:27 <veltas      > Yeah AI is very power intensive and cooling requirements are off the charts
2026-03-24 06:37:04 <tpbsd       > yeah, but these are only engineering issues
2026-03-24 06:37:24 <tpbsd       > and space has plenty of solar power and room
2026-03-24 06:37:31 <veltas      > For a million reasons I'd say this isn't a good idea, frankly also I predict that the cost for every day LLM usage will get lower and lower until most personal computers can do it
2026-03-24 06:38:03 <tpbsd       > maybe enough of them could do double duty and project shade onto overheated population areas ?
2026-03-24 06:38:06 <veltas      > One of the many reasons the over-optimistic investment in AI datacenters is a bit insane
2026-03-24 06:38:40 <tpbsd       > I agree, I think all the AI hype is insane
2026-03-24 06:39:12 <tpbsd       > do you remember ho many laughed at mucks starlink idea ?
2026-03-24 06:39:34 <tpbsd       > I was a late early adopter
2026-03-24 06:39:42 <tpbsd       > mainly because of the cost
2026-03-24 06:39:45 <veltas      > I'm sure lots did, however if you had explained Starlink to me I would have thought it was at least plausible
2026-03-24 06:39:54 <veltas      > The space AI datacenter idea is not
2026-03-24 06:39:56 <tpbsd       > now he gets 20,000 new users a day
2026-03-24 06:40:27 <veltas      > I mean it's physically possible, but just has no chance of being cost effective or valuable
2026-03-24 06:40:50 <tpbsd       > I had been the project manager for a 8 metre Aussat steerable dish groundlink in the 1980's so I know that field fairly well
2026-03-24 06:41:17 <tpbsd       > does it need to be cost effective ?
2026-03-24 06:41:40 <veltas      > Yes unless there is some kind of incredible value proposition I'm not thinking of
2026-03-24 06:41:45 <tpbsd       > if starlink subs are funding it and hes using existing starlink resources to build it ?
2026-03-24 06:42:17 <tpbsd       > starship is designed to launch 100 sats at a time
2026-03-24 06:42:43 <veltas      > I mean I'd ask whether Starlink is meant to do cool sci fi stuff or make money
2026-03-24 06:42:49 <tpbsd       > muck owns the rockets, the tech, he has the people the resources
2026-03-24 06:43:11 <veltas      > If Musk owns Starlink and wants to do it then I guess he can
2026-03-24 06:43:16 <tpbsd       > I think starlink went cash positive in early 2025
2026-03-24 06:43:50 <tpbsd       > mush is also building a 2nm fab in the USA for millions of AI chips
2026-03-24 06:44:16 <tpbsd       > he says that TSMC is only capable of producing 2% of the chips he needs
2026-03-24 06:44:31 <tpbsd       > so he has to build his own fab
2026-03-24 06:44:48 <veltas      > The main thing stopping him is probably the next government might not be so pleased with the idea of filling up even more of the night sky with clutter and making space harder to navigate than it already is
2026-03-24 06:44:49 <tpbsd       > now *thats* nuts in my opinion
2026-03-24 06:45:28 <tpbsd       > yeah, tho they would be more upset with muskc personal oligarchy power imho
2026-03-24 06:46:11 <tpbsd       > problam is that starlink is now a dependency of the USA military, they cant survive without it
2026-03-24 06:46:42 <tpbsd       > all the military vehicles, ships, cruise missiles etc have starlinks fitted
2026-03-24 06:46:50 <veltas      > Making a fab is a good idea, but there's a risk of over-predicting demand with the current AI bubble
2026-03-24 06:47:26 <tpbsd       > yeah it's a good idea, but a 2nm fan by a rocket and eev guy ?
2026-03-24 06:47:53 <tpbsd       > muck has zero fab expertise
2026-03-24 06:48:05 <veltas      > He'll learn what he needs to know and hire a lot of people who do know
2026-03-24 06:48:17 <tpbsd       > and that kind of thing can bankrupt entire countries
2026-03-24 06:48:31 <tpbsd       > yeah, he is determined
2026-03-24 06:48:59 <tpbsd       > and he brought me $5 USD a month unlimited internet!
2026-03-24 06:49:15 <veltas      > Musk is now maybe an expert in any of the things he runs a business in, but when he started he wasn't.  He can learn it, he's smart and yes determined
2026-03-24 06:49:34 <veltas      > People say he's no genius... well that might be true but he's a business genius, certainly compared to most CEOs
2026-03-24 06:51:04 <tpbsd       > well, starlink technology is so hi tech, it's almost alien
2026-03-24 06:51:32 <tpbsd       > I've been utterly blown away by it, and Im used to hitech wireless
2026-03-24 06:51:39 <veltas      > It's cutting edge, not 'alien' though
2026-03-24 06:51:44 <tpbsd       > I mean seriously hitech wireless
2026-03-24 06:51:56 <tpbsd       > it's more like military tech
2026-03-24 06:52:17 <tpbsd       > 'alien' was a bit expansive I admit
2026-03-24 06:52:41 <tpbsd       > Id never seen commercial active phased arrays before starlink
2026-03-24 06:52:58 <tpbsd       > they just didnt exist outside the military
2026-03-24 06:53:17 <tpbsd       > and thats what the starlink customer dishes are
2026-03-24 06:54:13 <tpbsd       > they belong in the nose of a fighterplane
2026-03-24 06:54:48 <tpbsd       > the only possible upgrade would be gallium nitride in those units I guess
2026-03-24 06:55:04 <tpbsd       > so theyd be smaller, etc
2026-03-24 06:55:29 <tpbsd       > and have GB datarates
2026-03-24 06:56:56 <veltas      > When Sky contracted Amstrad to make satellite dishes Sir Alan Sugar started with dustbin lids
2026-03-24 07:13:56 <tpbsd       > haha
2026-03-24 07:28:52 <veltas      > Well rather I think they contracted a company that made dustbin lids, it's a good idea really if you just need a metal 'dish' stamped out thousands of times
2026-03-24 07:29:27 <tpbsd       > the tech has moved on a lot since then
2026-03-24 07:30:03 <tpbsd       > tho I hear that the usaf could use a supply of dustbin lids for the f35 which have no radars
2026-03-24 09:34:09 <cleobuline  > Allonge toi sous la couette , ecoute radio LaByNet ,  l'azur l'azur l'azur LaByNet ça assure !
2026-03-24 12:25:50 <cleobuline  > I foresee psychiatrists for artificial intelligence within ten years
2026-03-24 12:26:06 <cleobuline  > may also be many humans victims of artificial intelligence interned in psychiatric hospitals
2026-03-24 13:22:00 <KipIngram   > Using a dustbin lid company for such a purpose - that was fairly brilliant.
2026-03-24 13:35:15 <iv4nshm4k0v > I vaguely recall a story of NASA looking for a writing implement suitable for zero gravity conditions, I think even having a press conference to call for suggestions.  Where one of the journalists asked, "how about a pencil?"  No idea if that's indeed happened; http://airandspace.si.edu/stories/editorial/saga-writing-space doesn't mention it, for instance.
2026-03-24 14:01:52 <veltas      > https://www.scientificamerican.com/article/fact-or-fiction-nasa-spen/
2026-03-24 14:03:09 <veltas      > I've heard the same thing, but as the above myth, which is not true.  But what you said is essentially true, it's just missing the facts that NASA had legit reservations about pencils
2026-03-24 14:03:56 <veltas      > So basically they were indeed criticised for over-spending on pens, when "a pencil would work", but there were good reasons to not use pencils so eventually having the "space pen" was worth it
2026-03-24 14:04:59 <veltas      > I had just heard it as "NASA spent millions on a space pen when soviets just used a pencil" but I think if I was half as wise as I am now I would have realised that was probably not true when I heard it
2026-03-24 14:06:32 <veltas      > I wouldn't be surprised if it was laughed about in soviet media / propaganda for a bit, and then when soviet-friendly people in the west heard it it sort of got bundled into this nasa vs soviet thing
2026-03-24 14:06:54 <veltas      > Or maybe the soviets added that bit too, not beneath them
2026-03-24 14:15:31 <iv4nshm4k0v > Having mistaken knowledge about 'potential enemy' (as the US was referred in the SU in the Cold War era) could backfire badly, so I doubt it's something that could've originated from official propaganda.  Soviet comedians seem a more likely source to me.
2026-03-24 14:15:53 <veltas      > Makes sense
2026-03-24 14:16:09 <veltas      > I could imagine it being comedians on either side of the iron curtain
2026-03-24 14:17:59 <iv4nshm4k0v > Also, my experience with felt-tip pens - even the cheaper ones - is that they /are/ usable upside-down.  No idea if that was the case back then.
2026-03-24 14:29:29 <vms14       > veltas: i've killed the rpn lang so no documentation will come :D
2026-03-24 14:29:39 <vms14       > i'm betraying you with lisp
2026-03-24 14:29:50 <vms14       > muahahahaha
2026-03-24 14:30:36 <vms14       > i keep oscillating between two, so i might rewrite it later
2026-03-24 14:30:51 <veltas      > "In 2003, Spanish astronaut Pedro Duque confirmed that a standard ballpoint pen worked perfectly for writing in his diary while aboard the Soyuz"
2026-03-24 14:31:15 <veltas      > This isn't surprising to me because while they don't work upside-down, there is no 'upside-down' in space
2026-03-24 14:31:25 <veltas      > So if the capillary action etc is enough then you can write
2026-03-24 14:31:40 <veltas      > So felt tip would probably work too
2026-03-24 14:33:56 <iv4nshm4k0v > /Some/ pens require gravity to function.  If the pen fails to work upside-down, it /may/ depend on standard gravity, and as such, may also fail in zero-gravity.
2026-03-24 14:33:56 <veltas      > vms14: I feel like Rust will be next
2026-03-24 14:34:22 <veltas      > iv4nshm4k0v: I agree but at least we can say just because it fails inverted doesn't mean it will fail in 0G
2026-03-24 14:39:31 <veltas      > vms14: If you oscillate a lot between ideas there can be many reasons, often it's because there's a push/pull between creating something valuable and something fun
2026-03-24 14:39:59 <veltas      > You realise after a while what you're doing doesn't make sense, or at least not enough to justify carrying on after the initial fun disappears
2026-03-24 14:40:32 <veltas      > But it doesn't have to go on forever, eventually you might actually narrow down something that makes enough sense to see through
2026-03-24 14:52:32 <vms14       > it's that the rpn lang is the coolest thing for extending it and implementing, but not for using it
2026-03-24 14:53:11 <vms14       > and lisp requires stuff the rpn lang does not care about, i can't optimize it so it will be extremely slow which worries me
2026-03-24 14:53:38 <vms14       > one lang can handle a big program but would be a pain to write it in, the other could express a big program but choke executing it xd
2026-03-24 14:54:30 <vms14       > actually i have made experiments mixing both, having the stack based lang by default and triggering the lisp interpreter with a "(" immediate word
2026-03-24 14:54:45 <vms14       > but i guess is best to have one than half baked two
2026-03-24 15:01:08 <veltas      > I might recommend checking out SectorLisp and seeing how that works
2026-03-24 15:01:20 <veltas      > Because it might be interesting to you, it's a very minimal lisp implementation
2026-03-24 15:01:48 <veltas      > Has the usual kind of niceties you like like garbage collection etc
2026-03-24 15:06:06 <vdupras     > vms14: implementing a "cons" structure in Forth and have it garbage collected doesn't imply much complexity and goes a long way towards giving a lispy feeling to the "(" hack you mentioned
2026-03-24 15:29:44 <vms14       > actually i just saw cobol and feels quite like what i always wanted, some sort of english like lang
2026-03-24 15:30:07 <vms14       > i also like that aspect from forth where code can look like an english sentence, but rpn kind of fights it
2026-03-24 15:41:58 <veltas      > The problem is just that in reality code's easier to read when it's less like english, it's something that cobol really solidified for the people that worked with it
2026-03-24 15:42:19 <veltas      > There's a reason why algebra and maths tends to use short notation
2026-03-24 15:49:59 <vms14       > funny that you say that because perl is one of the easiest langs for me to read someone else's code
2026-03-24 15:50:40 <vms14       > unless they start doing clever tricks which perl encourages
2026-03-24 16:58:31 <veltas      > Perl is succinct yes (?)
2026-03-24 17:00:41 <iv4nshm4k0v > More succinct than Cobol, so far as I can tell.
2026-03-24 17:06:26 <tpbsd       > vms14, have you read lua lang ?
2026-03-24 17:06:58 <tpbsd       > I know Perl and have written it for decades, but I found lua really easy
2026-03-24 17:08:40 <iv4nshm4k0v > To me, the killer feature of Perl is: die ("Fatally") unless (survival_possible ()); .  Makes me choose Perl pretty much whenever possible.
2026-03-24 17:59:47 <vms14       > tpbsd: yeah i learned the basics of lua which i forgot
2026-03-24 22:27:47 <veltas      > I've got a soft spot for Lua too tpbsd
2026-03-24 22:28:21 <veltas      > I just like simple stuff, Lua's pretty small