Comment I, for one, welcome, our new (Score 1) 44
emphasis on fidelity over complexity in neural simulation.
emphasis on fidelity over complexity in neural simulation.
I'm not sure I've seen any independent study which investigates such questions satisfactorily. (You may interpret that as [citation needed].)
Bessner and Davelaar, 1982. "Basic processes in reading: Two phonological codes."
And looking in from the opposite direction, I've also yet to see someone build a 4-simple-english-word rainbow table to directly attack the claim of security.
You don't need to count to 10^5^4 to know that it's a big number, far greater than the search spaces currently achievable with rainbow tables. Barring a monumental flaw in the hash function, the decreased per-character entropy shouldn't make a difference. (Though I guess it depends on how many "simple english words" you consider there to be.)
Certainly, in the field of memory, I am prepared to believe I am far from the norm. I have an exceptionally poor memory for almost everything. During my academic career I could never remember high level theories or identities, and had to repeatedly derive them from basic principles before using them.
With you on that one.
Forget "brain-on-a-chip" neural simulation FPGAs... The new hotness is "gut feeling"-based reasoning.
You left out a few words there. What you meant to say is "Each time they'd have the engineers look at the problem, and then the managers decided it was really probably OK."
According to their investigation, when reactor parts fail or systems fall out of compliance with the rules, studies are conducted by the industry and government. The studies conclude that existing standards are 'unnecessarily conservative.' Regulations are loosened, and the reactors are back in compliance.
I hate to come down on the side of Big Industry, but this is exactly how things should go. First of all, of course compliance problems spur studies on whether the standards are too conservative. If there's no difficulty in complying with the standards, why bother to do a study? Secondly, of course standards are going to be loosened over time. When you make the first nuclear reactor, you want to have incredible safeguards, even where they seem conservative, just in case. Then, once you've got fifty years of nuclear power under your belt and you have a more informed idea of what's important and what's not, you revise the standards.
That's not to say that nothing smells fishy here, of course. If these "studies" were performed in a biased and unscientific manner, and/or without enough transparency to determine how much bias affected the outcome, then that in itself is the problem. And when a single standards-loosening turns out to have been unwise, that should properly throw doubt on the conclusions of many related safety studies. But the framing of this story seems to be "science can prove anything". No, just bad science... and the solution is not to stop doing science.
here is the actual press release, which (unlike that article) doesn't skip over what they actually did.
Take that, Street View! Your lies about "streets without mailboxes on them" will not be tolerated!
Okay. The density of helium is about one sixth that of air, so one kilo of helium can lift about five kilos of not-helium. They mention lifting 150-ton loads, which would require 30 tons of helium. Worldwide production of helium is about 30 million tons a year. I think it'll be okay.
I can't tell whether you're actually espousing the frequentist interpretation of probability, or are parodying it. My hat is off to you.
You seem to have this weird idea that lambdas == self-modifying code. It's just the opposite -- lambdas are the alternative to self-modifying code. Lambdas represent the ability for code to create other code, without having to modify any existing code. In fact
IP laws
Ba-dum-cshhhh.
The solution he requested is stupid, because I don't have the solution he requested.
Fixed that for you.
Understanding is always the understanding of a smaller problem in relation to a bigger problem. -- P.D. Ouspensky