It is much more difficult than that. Starlink is essentially IT infrastructure, which certainly produces heat but nothing on the level of pure compute. Ejecting heat in space is a difficult problem that is currently solved on the ISS with large IR radiators which take up weight and space. The size, weight, power, cost tradeoffs lean heavily in favor of ground-based compute.
I recently bought a hybrid, and had been trying to wrap my brain around how it had an Atkinson engine in it. Was imagining a solenoid-driven crankshaft linkage or similar. Loved learning from this video that the secret is in valve timing and not fancy linkages.
But. The other points made felt muddled or even contradictory. I either didn't follow the TC guy's explanation closely enough, or the script could have used another pass.
> A deal between the two tech titans would give a boost to SpaceX’s business ahead of a historic public listing.
That tagline says it all. This is just pure market manipulation. There's no current path to space-based datacenters being feasible. Launch costs, unrepairable, enormous amounts of solar panels (and thermal rejectors) required. The concept doesn't survive even casual contact with reality.
> Do you think being alive is necessarily, rather than merely historically, required for consciousness?
Ah, I suppose I don't. My comment was poorly worded. However I feel confident that consciousness does not exist anywhere near the current level of fancy linear algebra and statistics that make LLMs work.
It's interesting to see the Turing test flipped and pointed back at ourselves. Can the computer trick you into believing it is alive all while you know it's a computer?
> Setup Guides: Step-by-step instructions for the “regular person” to get these models running on their laptop without going into the weeds. Coming up in the next post.
Looking forward to this. I was 'born yesterday' and haven't had any AI interactions outside of reading some google search result summaries. But I'm looking to pivot a proxmox server to try out self hosting ai tools, and have found the landscape to be very cluttered and difficult to find a starting point for someone with no context.
In my experience Copilot is such garbage. But I've only used it in the context of being forced to for my job, where it only ever answers prompts by telling me steps to go do myself, and then most of the time the steps are wrong anyway.
It's literally a scam in the same tune of "full self driving next year."
Launch costs, unrepairable once in-orbit, more expensive hardware to withstand space and radiation, the engineering problem of dissipating heat into space (check out how ISS does it), latency and cost to get data sent back Earth. "May" not be commercially viable? None of it is commercially viable from first principles.
> Previous administrations squandered our military advantages and the lives, goodwill, and resources of our people in grandiose nation-building projects and self-congratulatory pledges to uphold cloud-castle abstractions like the rules-based international order. These past leaders neglected and often actively undermined our warfighters' warrior ethos and our military's core, irreplaceable role -- fighting, winning, and thereby deterring the wars that really matter to our people.
At a previous job I had a hiccup onboarding to Atlassian. The startup founder sent initial invite to college email address, and then when we got our own domain and corporate email addresses everyone migrated their accounts over. Except me. Whenever I would try the process would fail. Eventually a human at Atlassian manually migrated me and lamented that their codebase was so labyrinthine and crusty that the bug would never be fixed.
Initially I wanted to write this comment saying buying them would be a bad idea just for the technical debt... But maybe they have the perfect "already vibe coded by humans" software to have an AI company take over.
I felt a knot in my stomach while watching the livestream coverage and hearing "They're doing so much science!" and I wanted to reach into the monitor and ask "Tell me! Give me details!" but instead felt I was watching the tiktok influence bear out on scientific coverage.
It's encouraging to see healthy conversation taking place around the open web's future, and the response to "Endgame to the open web."
Also glad to have discovered Julien Genestoux's article through this one.
As an aside, for a couple of years now I've stalled out on creating/contributing online through a personal blog or my peertube instance. In 2024 my grandfather received a phishing phone call of someone using approximately my voice to ask for bail money. We got lucky that he decided to call me directly to confirm, but that's not a guaranteed outcome and I'm afraid of contributing to making the next phishing attack more potent.
I suppose what I'm saying is that I feel pressured to keep quiet nowadays out of concern for how the web is changing and being taken advantage of.