Hacker News new | past | comments | ask | show | jobs | submit | more rst's comments login

? There's no other spacecraft there now with spare seats. NASA could arrange to send a Dragon up, but at a pretty fearsome cost. (Crew dragon missions are something like $300 million. There's one scheduled, and paid for, to launch fairly soon, but with no spare capacity -- if they wanted to use that for Butch and Suni's ride down, they'd need to bump two of the currently scheduled crew for that mission off.)


There's one scheduled, and paid for, to launch fairly soon, but with no spare capacity

Stupid question of the day, can we not just force jam a bunch of astronauts into it?


The police complain if you have people riding on your lap in a car. Imagine what NASA thinks of that for a space pod.


There's no way the NASA flight director wants to deal with the paperwork for elevated vital data of those whose laps were occupied: "Sir, the whole craft was vibrating, not to mention these flight suits are ...restricting..."


Yes absolutely. SpaceX Dragon is cleared to seat up to, I believe, four extra astronauts in an emergency.


Weight is probably a big one.

Slowing things down requires energy (or the dispersion of energy) and that amount of energy is directly related to the weight.

Plus breathing gases, waste management, and any other required consumables.


Isn't there a Dragon docked at the ISS right now?


That Dragon came with it's own crew, not empty.


However, in emergency situations, the Dragon capsule can seat up to, I believe, four extra astronauts, and there are only two on Starliner.


Can't they send up one of the space x ones?



It's been worse -- they've been caught setting up ghost kitchens in the name of real restaurants that didn't offer delivery or take-out service, and siphoning off order flow. One Michelin-star restaurateur found out about this when she got complaints about a delivery order placed to Seamless, which was pretending to offer delivery from her place, but was actually a ghost kitchen she'd never heard of ...

https://www.eater.com/2020/1/29/21113416/grubhub-seamless-ki...


> they've been caught setting up ghost kitchens in the name of real restaurants

This blows me away every time I’m in New York. They all do it. An everyone seems to be fine with it. (In that nobody is suing.)


Correction -- it's a 2:3 scale model of a PDP-10 front panel, concealing a Raspberry Pi running an emulator. In the original equipment, the panel would have been attached to one of several racks containing the CPU and main memory, cabled up to peripherals (tape drives like in '60s movies, disk drives the size of washing machines) through wires under the false floor of a dedicated computer room...

The PiDP-10 accurately models the console, but it doesn't depict the rest of this at all.


Full authenticity would also require raised flooring, and an audio track of people shouting "SAVE YOUR FILES!" whenever the lights flickered.


The KA-10 had core memory (except for the instruction counter and locations 0-15 which were general purpose registers which were implemented in DTL). So when the power went out you could just restart, except for the running process.

Bad news if that was the monitor of course.


An interesting thing about the KA-10 is that those DTL registers were optional. Architecturally the registers are simply the first 16 words of core memory. If you bought the fast register option with your KA-10 they installed the DTL registers which overlaid the first 16 words of core.

A consequence of that was that anything that took a memory address could access the registers as memory. Just give it an address in [0,15].

That included the program counter. Load code in 0-15 and jump to it and it would run quite a bit faster than if it were in core if your KA-10 had the fast register option.


The piece documents that at least two manufacturers (Data General and Univac) marketed their extended Fortran IV compilers as "Fortran V" -- but it doesn't say whether they actually extended Fortran IV the same way. "Fortran V" wasn't the name of any official standard; the ANSI standard which more or less documented Fortran IV was officially Fortran 66, and ANSI's next version was Fortran 77.


I learned Fortran in 1977 using WATFIV, which I just now realized was not a nonexistent Fortran 5, but the University of WATerloo Fortran IV compiler.


Data General's compiler was "Fortran 5", not V. Univac and Control Data Corporation had their own Fortran Vs, but there was no relation between the two.


Project planning started in the early 2010s, a few years before SpaceX had come close to demonstrating reusability -- at the time, it was regarded as a fairly risky and dubious experiment.


"Filed a formal complaint" with whom? The board itself is the final authority, empowered to investigate and act. If its members have concerns, they either do something or don't. There's nowhere else to pass the buck.


> "Filed a formal complaint" with whom?

For their own records. They can use it for justification for disciplinary action or legal ammunition. It is always useful to document things in writing. It's the same reason why companies will put you on PIP.


Why would you think this stuff wasn't in writing? Do you think the investigation OAI did into the board's actions was just relayed orally?[1] That the board's discussions where all done in conference calls and not in e-mail? Do you think that this documentation gets routinely released to the public or even employees? Before you throw around accusations of incompetence, you should probably have some shred of evidence. This is the type of stuff that is typically considered extremely confidential and even now they say they are unable to reveal details. Why do you think that is? What is preventing them from doing something they apparently want to? Legal agreements such as NDAs, the same ones that would have prevented them from going public with details last years.

The board's mistake was not figuring out a way to go public with their case against Sam when or before they fired him. They obviously misjudged how he would respond. But even there we don't know the full context and constraints they were under. Hopefully, one of them will answer that question of why they didn't at some point, but until we know more, we would be wise to reserve judgement.

[1] https://archive.is/wbwC2


> Before you throw around accusations of incompetence, you should probably have some shred of evidence.

The evidence is in this interview. It sounded like the board basically let Altman walk all over them until they suddenly decided that they wanted to fire him, but by then it was too late. For example, if they had a thorough paper trail documenting all of his lies, they could threaten a lawsuit pressuring him to resign. If YC fired him, they likely would have used a similar strategy to pressure him to step down without any blowback.


> they could threaten a lawsuit pressuring him to resign

This is extremely confused about the board's responsiblities and powers. A court would laugh this case out of court because the board _can just fire him_.


This has nothing to do with the board's power. Simply firing him doesn't stop him from taking all his employees with him. The lawsuit would be for the damage of Altman's lies, and would prevent him from simply starting another AI company.


Would that be some sort of non-compete agreement that would need to be enforceable in California?


Lie to your boss and see if they put you on a pip "Please improve your inability to tell the truth."


Documenting their actions is a duty of board members


For a case to the contrary: Midler v. Ford -- a case in which Ford hired one of Bette Midler's ex-backup singers to duplicate one of her performances for an ad (after trying and failing to get Midler herself). Ford never said this was actually Midler -- and it wasn't -- but Midler still sued and won. https://law.justia.com/cases/federal/appellate-courts/F2/849...


Ford gave explicit instructions to imitate a copyrighted performance. Because that specific recording as owned by a record studio.

If you can describe a woman's voice and mannerisms and the result sounds similar to a copyrighted performance, that is natural circumstance.

If you want an example of purposefully imitating something with a copyright, look at GNU. Anyone who looked at the UNIX code was realistically prevented from writing their own kernel with similar functions. But if a handful of folks describe what the kernel ended up doing and some <random> guy in his own head comes up with some C code and assembly to do end up with the same high level functions, well thats just fine, even if you include the original name.

The details matter. There is absolutely enough vocal difference, it doesn't take an audiologist to hear the two voices do sound different but very close. It would not be hard for the producers to describe "a" voice and that description would overlap heavily with ScarJo, and wow the marketing team reached out to see if she would attempt to fill the existing requirements. When she said no, they found a suitable alternative. If the intent was to have ScarJo do the voice and she said no and they did it anyways, thats illegal.


Off topic to the thread and your point, but are you confusing GNU with the Compaq BIOS reverse engineer and reimplementation? I hadn't heard this story about GNU (and what kernel)?


> Ford gave explicit instructions to imitate a copyrighted performance.

That case isn't copyright law, Ford had obtained rights to use the song itself.


The point is more that compared to prior landers, the Starship version at least has a uniquely high center of gravity over a narrow base, which makes it a whole lot easier to tip, and amplifies the consequences of, say, leg damage.


The center of mass should be pretty low relative to the height of the lander, the engines and propellant are the heaviest parts, the engines are obviously at the bottom. The heaviest component of the propellant is the LOX, which is also at the bottom.


This is false most of the fuel is gone by the time it lands and most of the payload is up high that's why the latest designs for starship have diagonal thrusters 2/3 of the way up the rocket so they can stabilize the top heavy part of the rocket without having to control it from a high moment arm


Starship carries ~1200t of propellant, of which ~950t is LOX, and 250t is Methane. While yes, most of that will be burned off by landing, it'll still need enough to return to lunar orbit. Even if we assume that only 10% of the fuel is needed to return to orbit, that's 95t right on the bottom with another 10t of engines and most of the 100t of dry mass of the Starship itself (plumbing, tank domes etc).

The thrusters you're (probably) thinking of are the landing thrusters that NASA thinks they might end up needing. Not to stabilize the rocket when on the ground, but because the Raptors might be too powerful and might dig out a crater underneath the vehicle when landing on an unprepared surface (such as the Moon, at least before a base is established or something is sent to prepare a proper surface). Placing weaker landing thrusters up top eliminates this issue, although at the moment they're still considered speculative in the sense that last we heard (which was admittedly a year or two ago), SpaceX are not convinced that this will be an issue.

Thrusters would anyway be a crazy approach to preventing a crewed vehicle from tipping over, as you wouldn't want them to be firing when the crew are doing any of the things that would involve the ship becoming potentially unstable (eg unloading cargo). For stability they'd have to use the large self-leveling legs from the original HLS design.


The most elaborate deployed version of this idea is probably PyPy, an alternative Python implementation that works by giving a JIT-compiling partial evaluator a hint-annotated Python interpreter as the program, and the Python code to be executed as its input. Slightly abstruse overview here: https://doc.pypy.org/en/latest/architecture.html


PyPy used partial evaluation at one point, but now it's "meta-tracing":

Why did we Abandon Partial Evaluation? - https://www.pypy.org/posts/2018/09/the-first-15-years-of-pyp...


Taking meta-tracing to the next level would reconstruct not just hot loops but even the original CFG, as speculated in: https://news.ycombinator.com/item?id=38778372

Anyone know of non-fictional work in this area?


Futamura projections should still be the basis for GraalVM


they are for Truffle-based language implementations https://chrisseaton.com/rubytruffle/pldi17-truffle/pldi17-tr...


I think PyPy moved away from partial evaluation to trace optimization, fwiw


For what it's worth, the word "electronic" in that description is a lot more load-bearing than it looks, excluding machines such as the Harvard Mark I (ASCC) and Bell Labs Relay Computers, both of which were solving problems under program control years earlier (unlike the Eniac, which initially had to be rewired for every problem -- stored-program control was added years later).


Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: