To get an SLS-equivalent payload to the lunar surface, it will take 8-16 Starship launches

You're extremely confused. SLS cannot land on the moon in the way that the (lunar variant) Starship can. It can only launch Orion to the moon. Orion is 8 meters tall and 5 meters in diameter. Starship is 52 meters tall and 9 meters in diameter. These are not the same thing.

SLS/Orion missions are expected to cost approximately $4,2B each. If you fully disposed of every Starship, the cost for 8-16 launches would be $720M-$1,44B. But of course the entire point is to not dispose of them; the goal is to get it down to where, like airplanes, most of the cost is propellant. The propellant for a single launch is $900k. Even if they don't get anywhere near propellant costs, you're still looking at orders of magnitude cheaper than a single SLS/Orion mission.

By far, most of SpaceX's launches are for Starlink, which is self-funded.
Nextmost is commercial launches. SpaceX does the lion's share of global commercial launches.
Government launches are a tiny piece of the pie. They don't "subsidize" anything, they're just yet another minor revenue stream.

The best you can say is that they charge more for government launches, but everyone charges more for government launches than commercial launches. You can argue over whether that's justified or not (launch providers have to do a lot of extra work for government launches - the DoD usually has a lot of special requirements, NASA usually demands extra safety precautions, government launches in general are more likely to want special trajectories, fully expended boosters, etc), but overall, the government is a bit player in terms of launch purchases.

In my experience it is, how effective it is is directly proportional to preexisting project complexity when the commands are run. The bigger the project, and the more parts that are interfacing together, the worse it performs. But for small, simple projects and creating frameworks, it can be amazing.

I'm not sure what "Building the Metaverse" is supposed to even mean anymore. Is he still obsessed with Ready Player One fantasies?

I mean, if he's just talking about generating 3d assets and the like, then maybe? AI 3d model generation is pretty useful if you don't care about every tiny detail matching up to some specific form. For example, I used an AI tool to make an image of an ancient mug with cave-art scrawled around its edges. It got the broad shapes of the model right, but had trouble with the fine engravings, making a lot of them part of the texture rather than the shape, but overall it was good enough that I just left off the engravings, had it generate a mug without them, then re-applied them with a displacement map. It got all the cracks and weathering and such on the mug really nice, and the print came out great after post-processing (cold-cast bronze + patina & polishing).

(I ended up switching from cave art to Linear A, because I also plan to at some point make a Linear B mug so that I can randomly offer guests one of the two mugs, have them rate it, and thus conduct Linear A-B Testing)

It's a cute "attack", but not practical in the real world. The "poison" stands out like a sore thumb on your loss graphs.

It's so easy to get tempted into feature bloat these days. You need a microcontroller for some simple set of features, like doing PWM control on a fan and handling a rotary switch, so you get something like a Seeed Studio XIAO ESP32S3 that's the size of a thumbnail and costs like $10, but then all of the sudden you have way more processing, memory capacity, pins, etc than you need, and oh hey, you now have USB, Bluetooth, and WiFi, and surely you should at least do SOMETHING with them, right? But the hey, for just a little bit of extra cost you could upgrade to a XIAO ESP32S3 Sense, and now you have a camera, microphone, and SD card, so you can do live video streaming, voice activation, gesture recognition... .... it really creeps up on you, because there's so much functionality in cheap, small packages today.

The irony though is that nobody really seems to bundle together everything one needs. Like, could we maybe have such a controller that also has builtin MOSFETs, USB + USB PD charging, BMS (1S-6S) functionality, and maybe a couple thermocouple sensors? Because most small devices need all these basic features, and it's way more cost, space, weight and effort to integrate separate components for all of them. The best I've found is a (bit overbuilt) card that has USB + USB PD (actually 2 of each, and reverse charging support), BMS support (1-5S), one thermocouple sensor, and a small charging display - but no processor or MOSFETs.

Because your great-grandfather who hated loud noises, family gatherings, and was at his happiest tilling the fields in extremely straight rows day after day.

When he retired, he built ships in jars and organized all of his stamps in his spare time. Of course he was autistic.

We just didn't say that, we said, "ah yeah, Grandpa Joe was eccentric. Helluva farmer, though."

I was a programming lead, and this is pretty close to how I worked. We scheduled my time as 50% management and 50% programming. If I started getting over 50% management, it was usually an indication of something else significant happening. Sometimes that was fine and sometimes it wasn't, and we'd do something about it.

But doing that "individual contributor" work was the only way I could ACTUALLY manage. If I'm not reading and working on the code, I have no way to assess the programmers I'm supervising, and no way to know if as a team we're actually achieving our goals.

Management for the sake of management is honestly a bit insane to me. I understand that not everywhere has a structure like a game company (fortunately), but I firmly believe if all you do is manage, you'll always be easy to manipulate and lie to, and you'll never actually know what's going on. It makes you a worse manager.

None of the woo "hurt him in terms of accelerating cancer". The problem is that he did that instead of actual treatment, and let a highly treatable cancer turn into a nearly-untreatable one.

Oh, it's because people with success, people that are demonstrably skilled at something—they start to think that their knowledge in one field means that they're smart at EVERYTHING. You see it with physicists all the time. Doctors, too.

They start believing their own hype and think they know better than actual experts. Then they end up dead, like Jobs. It's sad, but it feels like maybe it was inevitable.

Actually, drinking hot coffee or tea is really common among Asians on hot days. It makes you sweat, and that sweat cools you off. My Chinese mother was forever giving me tea to drink during the summer.

(The hot bath part is definitely true.)

Anyway, it is FOR SURE true that corporations don't want to pay us AND AI at the same time, but the reality is that they'll need programmers to oversee all this stuff. The current fantasy that AI can do it all is absurd. The summary says that 70-something percent of coding problems can be solved by AI, but that's such a low level of complexity. I'm a gameplay programmer, and yeah, I use a lot of simple programming elements composed together to make a big system, but I still have to understand the big system, and then the bigger system that it lives inside, and then the whole game. Junior programmers are often lost in the complexity, or take shortcuts that cause bugs. AI is no different.

I have a whole diatribe about natural language programming, but Dijkstra said it first, and he said it better, so I'll just leave the link here. tl;dr Natural language programming (like vibe coding) is a stupid waste of time that provides no gains at all.

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.cs.utexas.edu%2F~EWD...

When you mentioned "third partner" who cashed out early, I thought for a minute you were going to be talking about Ronald Wayne - what a life of bad decisions he made ;)

For those not familiar:

He got 10% of the original Apple stock (drew the first Apple Logo, made the partnership documents, wrote the Apple I manual, etc).
Twelve days later, he sold it for $800.
Okay, but he could still try to claim rights in court... nah, a year later he signed a contract with the company to forfeit any potential future claims against the company for $1500.
Okay, well, it's not like he had an opportunity to rethink... nah, Jobs and Wozniak spent two years trying to get him back, to no avail.
Okay, but he still had, like memorabilia he could hawk from the early days, like his signed contract. Nah, he sold that for $500 in 2016.
And that contract went on later to be sold for $1,6 million.
Okay, well, I'm sure he went on to do great things... nah, he ended up running a tiny postage stamp shop.
Which he ended up having to move into his Florida home because of repeated break-ins.
Which he then had to sell after an inside-job heist bankrupted him.

1 cent is just what they're buying the shares at, not what they're worth.

Jobs committed suicide-by-woo. He didn't "turn away from traditional therapy because it can't keep up with rapidly advancing metastasis", he turned away from treatment for a perfectly treatable form of cancer for nine months to try things like a vegan diet, acupuncture and herbal remedies, and that killed him.

Steve Jobs had islet cell neuroendocrine tumor. It's much less aggressive than normal pancreatic adenocarcinoma. The five-year survival rate is 95% with surgical intervention. Jobs was specifically told that he had one of the 5% of pancreatic cancers "that can be cured", and there was no evidence at the time of his diagnosis that it had spread. Jobs instead turned to woo. Eight months later, there was signs on CT scans that his cancer had grown and possibly spread, and then he finally underwent surgery, it was confirmed that there were now secondary tumors on his liver. His odds of a five-year survival at this point were now 23%. And he did not roll that 23%.

Jobs himself regretted his decision to delay conventional medical intervention.

I did some proof of concept tests with both Pointer Lock and PointerEvents, but both failed because you don't get *any* data if you're not moving the mouse, and only get (heavily rounded) datapoints when you do move the mouse. You'd need raw access to data coming from the mouse, before even the mouse driver, to do what they did.

You *might* be able to pull off a statistical attack, collecting noise in the fluctuations of movement positions and timing in the data you receive when the mouse *is* moving. But I can't see how that could possibly have the fidelity to recover audio, except for *maybe* really deep bass. And again, it'd only apply for when the mouse is actually moving.

Neat attack, but not really practical in the browser.