jakub_h

Sounds like an opportunity for applying Ohm/OMeta?

Then why not have standardized APIs? You could just switch implementations, then, if you stop liking your current vendor. (Or maybe the future lies in call-by-meaning architectures and automated adapter generation?)

For fuel delivery to LEO, the delivery vehicle could be quite cheap. Even the Dragon might be an overkill. Perhaps a simple tank with a simple RCS assembly would be enough.

True, but if this additional testing reveals issues not revealed earlier, they'll just iterate again until they won't have to X-ray every plate after every flight, because that is clearly not the MO they have in mind. In the worst case scenario, they might need one extra vehicle generation (BFR?) to actually get there, but I'm quite certain they will eventually get there. Otherwise there's just no point to that company.

Remember that with reusability in play, the guidance system can decide "screw the first stage landing, the payload is more important" and simply switch to an expendable launch profile if it turns out that you have more engine problems on the first stage than you designed for. Then you use the reserve fuel to get the $300M satellite safely to orbit instead of getting the $30M stage safely to the ground. The customer is happy and you've only lost a unit for which you have a dozen of direct substitutes in storage anyway (SpaceX already doesn't have enough storage space, apparently!).

"Engines are incredibly expensive to manufacture"

Well, the Shuttle engines were, on the order of $50M or something like that. But the Merlin-1D, according to a former SpaceX employee on Reddit, is in the ballpark of $1M per unit. They've really streamlined the design for mass production, so even if thrown away after a single use, the Merlins are actually much cheaper than the Shuttle engines even if you take RS-25 reuse into consideration (9 launches per one RS-25 unit in average - so about $5-6M per engine-launch in just amortized manufacturing costs?). They're also much simpler and sturdier, much like the F-1 engines - despite not being designed for it, it turned out that those could have been used for about twenty-thirty times without major problems and without major component replacements. The RS-25s pretty much required general overhaul after every flight.

That's possible. But historically, when you talk about the middle class, the first thing that should come to your mind is executives, surgeons and lawyers, not car assembly line workers.

Exactly. There's a vicious cycle of expensive payloads driving costs of launches up (extra scrutiny, fewer launches, less experience, less cost-reducing hardware innovation - the RL-10 alone is sixty years old and quite expensive to build by hand from thousands of components, for example) in turn again driving costs of payloads up (you want the best bang for the buck if the price per kilogram is high, so you use overengineered spacecraft designs and materials).

Fortunately, there seems to be no shortage of ways out of this - new large markets like massive LEO constellations and fuel depots for more capable exploration missions immediately come to one's mind.

Just 0 kilometers from West Germany, there used to be a country with "strict exit visas", too. You can't get closer than that.

It sounds horrible that we should compromise potential performance (including per watt and per transistor) by giving up on (micro)architectural flexibility. Why should branch delay slots, instruction combinations that do not handle hazards and weird memory models matter in an era in which something like IBM's TIMI could be used for compatibility if you need it, and native ISA accessed if you don't or if you have your own toolchain?

Which is delusional; blue collar is the archetypal working class.

If it covers results, what about the virtual ISAs that allow you to load data from any address of a byte-adressed memory and their implementations include optimizations?

I guess it's all a matter of your idiolect. It works, for some values of "works". Maybe it works for the macroeconomist studying national GDP growth rates? Sadly, there are also barbers and truck drivers.

> Prolog and Lisp were the two languages we were forbidden to use in compiler design project, as it was considered "cheating" versus using something harder like C.

"Cheating"? Smells like a chance for the revenge of the pattern-matching metacompiler!

Sadly, Typed Racket always seemed to me like some kind of shoehorning an ad-hoc system onto an AST. For that purpose, a deductive system seems much more natural. There may be problems with decidability, but the fortuitous thing is that this is not a deal-breaker for a primarily dynamic language. You should always be able to rewrite your modules so that they're easier to check, but even if you can't do that outright, the system could still find faster versions if a few type guards or other invariant checks get inserted here and there, or perhaps some form of assistant in your development environment could hint at the need to insert them manually.

> A set of conditionals is probably the most obvious way to express branching.

Besides object polymorphism and sets of conditionals, there's also generalized predicate dispatch, but that's probably an overkill for many things.

Well, let's not read much into it. The headline says that dissonant tones sound fine to people not raised on Western music. It doesn't say that dissonant tones don't sound fine to people raised on Western music. ;)

If the data structures in memory don't fit the new code, I suspect that's a compiler bug.

"I have a copy of TR 149, Transmission of Information II, as well, but no scanner anymore."

How delightfully ironic! ;)

You could also pack three 20-bit numbers into a 64-bit word, if you needed it. Then again, the C family languages are kind of weird about that.

Very yummy, thanks!

> Did a colleague take at peek at the screen and said, hey I have the same equations?

This would an interesting thing to try - a computer system that would scan all the papers for math and find parallels. I think we already have something like term indexing for deductive systems?

Heh! For some reason, that reminded me of this famous conversation:

"...and that, my liege, is how we know the Earth to be banana-shaped."

"This new learning amazes me, Sir Bedevere. Explain again how sheep's bladders may be employed to prevent earthquakes."

> Andrew Ng's machine learning class

Ah yes, the one I've heard about but still have to take. :) Well, I guess I should give Coursesa a chance. (Somehow I'm not fond of their "timelined" format, it seems redundant if you're communicating with a machine. I hope the future of online learning will avoid it like the plague.)

It seems to me that if we humans are the only ones doing this, it is either trivially natural or trivially unnatural based on whether you include or exclude our creations in/out of the natural world.

Thank you very much. I'm aware of a large number of books that I could read, but if Sturgeon's law applies, or at least if a lot of the reading would be redundant anyway, I'll rather ask for pointers than to waste a lot of time by rediscovering the wheel. Basically, what I was asking for was the NLP's respected equivalent of Norvig and Russell, or similar books from other fields. Those two books seem to fit to bill.

> but I would never have considered Markov processes to be an adequate predictability model for natural language

Isn't it the case that for very short distances (several elements), power decay and exponential decay are (or can be made, with proper constants) more similar? Thus, if predictive models were originally studied only for very short sequences in the past (limited computational resources!), it seems to make sense that this is a mistake that anyone could have made more easily back then.

> This seems like a very important paper

> mark_l_watson

Hah, I'll take your word for it, then. :) Are there any recent comprehensive monographs you'd recommend for state-of-the-art NPL, for someone who has yet to enter the field?

> and Vulkan, the new replacement for OpenGL, defines a vendor-independent bytecode format for representing GPU programs.

Vulcan's SPIR-V, however, is more like a compiler's intermediate format than an ISA to be interpreted by either a program or hardware. It's probably more similar to AS/400's TIMI encoding in how it's supposed to work.

(Also, the mention of SWEET16 and the way of making parts of low-level programs more palatable reminded of the invention of Forth in late 1950s/early 1960s as well.)

> that's not freedom of speech.

I didn't say that, I was talking about the value of anonymous communication in a democracy. Surely voting is a form of communication.