The abstract process of computation as implemented by anything we'd call a computer is equivalent to any possible mechanism that operates on information.
There can never be any mechanism that deals with information that can ever be more general than classical computing.
Data is most general concept we (can) have for anything that can be known or observed. If it can't be known or observed, it can't have any tangible effect, so you can never make a mechanism that operates on it. It's not a cop-out, it's the exact opposite.
I think I see the disconnect here. When I say general, I also mean specific. That is to say, when I say that computers are general, I mean to say that any given, specific, individal computer (including a bird if it was smart enough), given any and every specific task and some amount of memory, can do it if and only if it is possible at all.
In other words, computers are interchangeable, you only need one and it can do any task (with enough memory)
This isn't true for a watch or a wrench or a wheel, but it's true for any computer.
With this definition of general, as you can see, there is nothing vague about the statement at all, and it can't be understood as a cop out.
Sorry for being unclear, I definitely understand why your interpretation of general in my comment is valid.
There are things that computers are inherently incapable of. The problem is that as far as anyone knows, they are things that are simply impossible, as long as we are talking about transforming data. The idea that there are possible things that computers can't do in this wheelhouse is called hypercomputation:
https://en.wikipedia.org/wiki/Hypercomputation
The only contender that isn't either mathematically or physically impossible is to take a regular computer and sling it around a black hole with particular attributes, ie, a normal computer.
I mention this in the comment you're replying to - I say that a computer can do such a task if and only if it is possible.
The brain isn’t a computer. It doesn’t function off of binary logic as far as we know.
Physical systems which rely on continuous processes (ie, neither on nor off) are exactly equivalent to discrete processes (which have a finite number of states) are exactly equivalent to binary processes (having exactly two states) as far as computation is concerned.
This is because continuous physical systems inherently, necessarily have some amount of noise/uncertainty, and there is a rigorous mathematical proof that once this is true, then the two are equivalent.
It is exceedingly unlikely that there is any operation on data that the human brain can do but that computers can't. There is not an inkling of a scientific hint that this could possibly be true. Of course, you can't prove a negative.
There are a lot of assumptions here you’re making, and that the majority of tech bros incorrectly make
The thing I am assuming here is the Church-Turing thesis. It's a thesis that is assumed to be true by every computer scientist and the vast majority of mathematicians who it concerns. If you can show that it's incorrect I'd love to hear it, but no one even has a lead on how it could possibly be untrue. People tried very hard.
Unfortunately, we've proven that all physically known processes can be simulated on a Turing machine, so unless there is some crazy new discovery of physics, there is no kind of computation that can do something Turing machines can't. It's not the only kind, but it's equivalent to any other that is possible we've come up with, and to any we could physically implement with our knowledge of physics. It would be very cool if that wasn't the case, though.
Feynman wrote a paper about it in the 80s, called "Simulating Physics with Computers", in which he lays out how all known laws of physics can be simulated using a computer to sufficient accuracy, but that without a quantum computer it would take exponential time to do so.
I'm not sure I follow. Surely, more people have observed the action of wheels than any of the others.
The abstract process of computation as implemented by anything we'd call a computer is equivalent to any possible mechanism that operates on information.
There can never be any mechanism that deals with information that can ever be more general than classical computing.
deleted by creator
Data is most general concept we (can) have for anything that can be known or observed. If it can't be known or observed, it can't have any tangible effect, so you can never make a mechanism that operates on it. It's not a cop-out, it's the exact opposite.
deleted by creator
I think I see the disconnect here. When I say general, I also mean specific. That is to say, when I say that computers are general, I mean to say that any given, specific, individal computer (including a bird if it was smart enough), given any and every specific task and some amount of memory, can do it if and only if it is possible at all.
In other words, computers are interchangeable, you only need one and it can do any task (with enough memory)
This isn't true for a watch or a wrench or a wheel, but it's true for any computer.
With this definition of general, as you can see, there is nothing vague about the statement at all, and it can't be understood as a cop out.
Sorry for being unclear, I definitely understand why your interpretation of general in my comment is valid.
deleted by creator
deleted by creator
There are things that computers are inherently incapable of. The problem is that as far as anyone knows, they are things that are simply impossible, as long as we are talking about transforming data. The idea that there are possible things that computers can't do in this wheelhouse is called hypercomputation:
https://en.wikipedia.org/wiki/Hypercomputation
The only contender that isn't either mathematically or physically impossible is to take a regular computer and sling it around a black hole with particular attributes, ie, a normal computer.
I mention this in the comment you're replying to - I say that a computer can do such a task if and only if it is possible.
Physical systems which rely on continuous processes (ie, neither on nor off) are exactly equivalent to discrete processes (which have a finite number of states) are exactly equivalent to binary processes (having exactly two states) as far as computation is concerned.
This is because continuous physical systems inherently, necessarily have some amount of noise/uncertainty, and there is a rigorous mathematical proof that once this is true, then the two are equivalent.
It is exceedingly unlikely that there is any operation on data that the human brain can do but that computers can't. There is not an inkling of a scientific hint that this could possibly be true. Of course, you can't prove a negative.
The thing I am assuming here is the Church-Turing thesis. It's a thesis that is assumed to be true by every computer scientist and the vast majority of mathematicians who it concerns. If you can show that it's incorrect I'd love to hear it, but no one even has a lead on how it could possibly be untrue. People tried very hard.
deleted by creator
That is not provedly true. Turing machines are just a form of computation, fairly sure we've not proved they're the only possible kind.
Unfortunately, we've proven that all physically known processes can be simulated on a Turing machine, so unless there is some crazy new discovery of physics, there is no kind of computation that can do something Turing machines can't. It's not the only kind, but it's equivalent to any other that is possible we've come up with, and to any we could physically implement with our knowledge of physics. It would be very cool if that wasn't the case, though.
Have we proven that? If so, super cool.
Feynman wrote a paper about it in the 80s, called "Simulating Physics with Computers", in which he lays out how all known laws of physics can be simulated using a computer to sufficient accuracy, but that without a quantum computer it would take exponential time to do so.