Is not that possible? An ID sets the wheels in motion and the rest is evolution. After all we are on the verge of becoming ID ourselves... Well, the "I" part is debatable, but the "D" is not out of question. If we can (almost) do it, why should not it have happened before?
Its turtles all the way down, young man...
That's great! Google strikes again... We can now "predict" a flu season when it is already happening. The timing of a flu season is more or less regular for a particular place (one of the mysteries of flu) and the big problem is not to predict when it will happen, but what will be the strain of choice for this year. This is the effort of CDC/WHO and it is a tricky problem because the flu virus mutates a lot (hence the need to be vaccinated every year). The mutation sometimes is such that an avian and a human virus combine to produce a new human virus. That's when bad things happen (predicted vaccines fail, 1918 pandemic, etc.). Google's idea is at best very interesting as a tool for monitoring the spread of flu amongst people with access to internet, but I fail to see its predictive power in general.
This is my feeling about quantum machines as well. Although they will certainly be a nice next step, they will not be miracle workers by far. To the layman there is certainly a lot of mystery about them which includes zombie cats, Einstein, Feinman, cool words like "quantum", breaking codes ("OMG! You mean the Da Vinci Code!"), and so on. The truth is much simpler (and at the same time more complex), but the "simple truth" does not attract investors...
Quantum computation is probabilistic in nature, except that the "probabilities" are complex numbers, and so "destructive" influence is possible. Such influence does not seem to be possible with classical probabilities. In fact the same all-states-possible-until-you-look (a.k.a the zombie cat) phenomenon exist in classical probabilities as well -- all random outcomes are possible, before you pick one. In both randomized and quantum algorithms the problem is how to make it so that most random choices work, and the "destructive" influence of complex valued probabilities over real valued ones gives some advantage, but nothing spectacular (and I do not believe that complex probabilities are the same as non-deterministic choice, so quantum computers probably will not solve efficiently NP-complete problems).
Randomized algorithms have existed for a long time and they certainly give a huge algorithmic advantage -- until recently (and actually still, from a practical point of view) the only way of checking whether a given number is prime was with a randomized algorithm. In fact, there are many good randomized algorithms, whereas pretty much the only quantum algorithm that is known today is for factoring numbers. And as the primality story shows, maybe somebody will show how to factor efficiently on a classical machine.
Randomized algorithms can be run efficiently on a classical machine with access to random bits, so you do not need a billion dollar investment to run them (since classical machines already exist and altough obtaining really random bits is hard, usually "dirty" random bits work). The problem with them is that you cannot tell a user -- "I have this really fast solution to your problem, but once in billion billions you will lose $10M". It does not matter that you explain to them that once in billion billions the sun will fall on earth. Quantum machines will have the same problem, but the zombie cat is way cooler.
So, yeah, quantum computers will bring peace and prosperity for everyone, they will take care of smelly feet, and bad breath, and according to a prof at the Univ of BS, women will find guys that own quantum computers much more attractive. Amen!
In the meantime: the factors of 6 are 2 and 3. $100M, please.
Slashdot Mirror
Is not that possible? An ID sets the wheels in motion and the rest is evolution. After all we are on the verge of becoming ID ourselves... Well, the "I" part is debatable, but the "D" is not out of question. If we can (almost) do it, why should not it have happened before? Its turtles all the way down, young man...
That's great! Google strikes again... We can now "predict" a flu season when it is already happening. The timing of a flu season is more or less regular for a particular place (one of the mysteries of flu) and the big problem is not to predict when it will happen, but what will be the strain of choice for this year. This is the effort of CDC/WHO and it is a tricky problem because the flu virus mutates a lot (hence the need to be vaccinated every year). The mutation sometimes is such that an avian and a human virus combine to produce a new human virus. That's when bad things happen (predicted vaccines fail, 1918 pandemic, etc.). Google's idea is at best very interesting as a tool for monitoring the spread of flu amongst people with access to internet, but I fail to see its predictive power in general.
This is my feeling about quantum machines as well. Although they will certainly be a nice next
step, they will not be miracle workers by far. To the layman there is certainly a lot of mystery about them which includes zombie cats, Einstein, Feinman, cool words like "quantum", breaking codes
("OMG! You mean the Da Vinci Code!"), and so on. The truth is much simpler (and at the same time
more complex), but the "simple truth" does not attract investors...
Quantum computation is probabilistic in nature, except that the "probabilities" are complex
numbers, and so "destructive" influence is possible. Such influence does not seem to be
possible with classical probabilities. In fact the same all-states-possible-until-you-look (a.k.a the
zombie cat) phenomenon exist in classical probabilities as well -- all random outcomes are
possible, before you pick one. In both randomized and quantum algorithms the problem is how to make it so that most random choices work, and the "destructive" influence of complex valued
probabilities over real valued ones gives some advantage, but nothing spectacular (and I do not
believe that complex probabilities are the same as non-deterministic choice, so quantum computers probably will not solve efficiently NP-complete problems).
Randomized algorithms have existed for a long time and they certainly give a huge algorithmic
advantage -- until recently (and actually still, from a practical point of view) the only way of checking whether a given number is prime was with a randomized algorithm. In fact, there
are many good randomized algorithms, whereas pretty much the only quantum algorithm that is
known today is for factoring numbers. And as the primality story shows, maybe somebody will show
how to factor efficiently on a classical machine.
Randomized algorithms can be run efficiently on a classical machine with access to random bits, so
you do not need a billion dollar investment to run them (since classical machines already exist and
altough obtaining really random bits is hard, usually "dirty" random bits work). The problem
with them is that you cannot tell a user -- "I have this really fast solution to your problem,
but once in billion billions you will lose $10M". It does not matter that you explain to them that
once in billion billions the sun will fall on earth. Quantum machines will have the same
problem, but the zombie cat is way cooler.
So, yeah, quantum computers will bring peace and prosperity for everyone, they will take care of
smelly feet, and bad breath, and according to a prof at the Univ of BS, women will find guys that
own quantum computers much more attractive. Amen!
In the meantime: the factors of 6 are 2 and 3. $100M, please.