Okay, let's try to simplify it even more. Quantum computers have gates just like ordinary computers. There's a difference, though: instead of acting only on a single value, they act on a whole bundle of them at once (that's the probability distribution). You can in effect calculate a function on many inputs at once, but you can't reach behind the curtain and pick all the answers (or any answer you want) from the result. When you do ask the computer to reveal an answer, you get a random answer from the entire bundle according to its probability. The clever part about quantum algorithm design is then to first alter the input bundle in the way you want, then selectively amplify the answer that you're interested in so you'll get it very often and get other answers rarely if at all.

More or less. I Am Not A Quantum Physicist, grain of salt, etcetera.

The process of assembling this quote is done by software at the full discretion of the owner of the computer.

So if you want to play that movie or clip on mplayer instead of Windows Media Player, you're of course free not to divulge any information that you're using mplayer - but if the site doesn't get a signed "I'm using Windows Media Player" token, it won't play the video clip in question. That freedom, like the freedom to work or not to work, is worthless if there's an imbalance of power - if all the video sites only want WMP for instance, or if you want to leak evidence of illicit goings-on at your company and the company server will only serve the data to an MSOffice version that can be trusted to stick the document in sealed storage.

How about this solution: put a switch somewhere on the computer. Switch up? Then you can't forge TPM keys. Switch down? Then you can do so at your discretion. If you want protection that is truly for your own good, you'll just leave the switch up (and no virus can forge anything). If the programs you're running start professing allegiance to their owners in preference to you, flip that switch down and bypass them. Sure, lusers may be socially engineered, but if they're that gullible, the malware could do a DoS by instructing them to stick pieces of metal into the closest outlet, too.
If Trusted Computing is truly about empowering the user, then this switch solution should only empower the user more and so there should be no opposition to it beyond the social engineering complaint.

I've consulted for a bank, and here's the dream : full offline money. If you have a TPM they will manage your account in your laptop (or phone, or ...) and have full offline payments. Because the TPM will only give their program access to the data, they can still prevent you from simply adding money in your own account, while allowing fully disconnected payments to occur which the bank will only find out about weeks after the fact (and so can you on other's computers of course).

That sounds like a pretty big incentive to find someone with an electron microscope, or if you're part of organized crime, getting your own. It's not like TPM hacks haven't been pulled off before.

O(n^3), not O(log(n)). If it'd been log(n), the demo quantum computers would already be good enough. As for symmetric crypto, any black box search can be sped up to take the square root of the time it would otherwise take, so a QC would have about as hard a time cracking 256-bit AES as an otherwise comparable classical computer would have cracking 128-bit AES.

It is dead, dead, dead. It thought it was hot, guess what, it's not!

As a culture, we find that the most appropriate treatment of people who have a criminal psychosis is to isolate them and help them, forcibly. We also find that they are not "wrong" and don't need to be punished, but require help. I don't readily see how an act of violence in this case is a critical point where we force help on the unwilling. So, why not force it earlier and prevent the violent acts?

To be a devil's advocate right back (angel's advocate?):

The danger is that preemptive treatment of, say, schizophrenia, can turn into preemptive treatment of sluggishly progressing schizophrenia, were the government to become sufficiently corrupt.

There's also the usual Bayesian argument: if the pre-crime test has 0.1% false positive and negative rate, and there are 10 terrorists in the US, the test is useless, even though 0.1% sounds really impressive and could convince lawmakers. For ordinary crimes, it would still claim a lot of innocents to be suspect.

Wasn't the League of Nations the first real attempt to get global consensus?

Hi, Dotzler! Good to see you here. You don't have to be anonymous, you know :)

From there, I want to expand the do not call registry. Right now, it only covers telemarketers. I want to also ban solicitations from charities, surveys, and political groups. I want a "leave me the fuck alone" registry. If you aren't one of my friends, customers, or suppliers, or someone I've given permission to call, then I don't want to talk to you.

How about a "do call" registry? Ban solicitations from said marketers, charities, surveys, political groups, et cetera to any number not on the list, and have all new phone numbers start off-list.

You're not hacking at the BIOS level, you're hacking at the bootloader level. But even if the malware were to try to tinker with the BIOS, that vaunted TPM could just deny it access to that part of the CMOS (or flash or what it is that holds the key/hash). So the UEFI would work like this: in either its own flash or in CMOS, there's a region that stores a hash or set of hashes. The computer comes with the hash for well-known bootloaders and, on boot, checks if the loader hashes to one of them. If not, it checks if it hashes to the user override. If neither, then up goes the warning -- and if the user presses "accept", the new hash gets loaded into the user override field. In any event, after this has been done, the TPM (or UEFI, or whatever is responsible for it) locks access to that part of flash or CMOS so that it can't be attacked from the outside.

Yup, the SSH way. If the bootloader changes, pop up a "WARNING: key has changed!" text, with a revert-loader option. If the UEFI is passworded, require the user to input the password if he wants to proceed without reversion. This means that corporate desktops won't be compromised this way (since the end user doesn't have the password and Mitnick doesn't either), and should give other users some reason for concern. If the other users press "go on ahead" anyway, well, there's only so much you can do against a dancing bunnies attack. As another poster pointed out, if users are that gullible, the malware could just as easily do a denial of service attack by asking the user to stick a fork into the nearest outlet.

Let's try that again. The original poster's point is that if you want a Magical Zero-Emissions Hydrogen Storage System, you just take the hydrogen you were to use, combine it with carbon (from CO2 from the air), then ship your hydrocarbon around. The guy at the other end then burns the stuff and the CO2 you used is released back into the air. Voila, zero emissions.

(I still think sodium or lithium borohydride would be a better reversible energy carrier, as it has a greater energy density than gasoline and can be easily used in direct borohydride fuel cells, but first they have to get the "recharging" working at better than 10% efficiency.)

Hm, why do you think we haven't evolved with perfect memory? Could there be a good reason?

The standard answers to this kind of question are:

1. Local optimum: Evolution is a black box optimization process. No black box optimization process that progresses in reasonable time can cover all of the search space, so they will get stuck in local optima. The eye is a good example of a local optimum. The nerve fibers are on the wrong side of the eye, so you get a large blind spot where the fibers go "out of the eye" and into the brain. However, to fix this suboptimal design would require a rather large trek through solution space, and some of the intermediate solutions would be so bad that evolution would weed them out. Therefore, the eye remains as it is. The same may be the case for memory structures in the brain.

2. Shifting tradeoffs: Evolution is also quite slow. Therefore, its tradeoffs are based on an environment in which we no longer reside, where food was scarce and one had to save energy lest one died. This sort of wrong-tradeoff optimization is evident everywhere you have obesity, for instance.
In this case, perhaps near-perfect memory was too heavy a weight to bear in conditions of scarce resources. If it was, evolution would not have optimized in that direction, because the losses (due to having to expend more energy) would be greater than the gains (of greater planning or what have you). Now, on the other hand, we do have much more energy available, but evolution hasn't caught up.

So to sum all of that up: to the degree that the evolutionary process visits close to optimal solutions, it is slow, and to the degree that the evolutionary process is fast, it skips certain solutions that might be very good. In both of these cases, technology can cover what evolution does not - when evolution is blind, technology can be less so; and when evolution has its blind spots, technology have others so they can make up for each other's partial coverage.

Given their rather long latencies from speed-of-light delays, they might be using forward error correction to fix errors ahead of time. This works more or less like PAR files in that if less than a given fraction of the bits are corrupted, the data can still be recovered without problem, and that would seem to be much quicker than having to do a SYN/ACK type "Hey, these don't CRC out right, give me them again" scheme.

He must have been digging into Moussa Ibrahim's secret stash.