Hacker News Reader

Comment by empath75 | Hacker News Reader

I do wonder if we're just a bunch of drunks looking under a lamppost for our keys because that's where the light is.

We've discovered a bunch of phenomena that happen to be able to be described by the mathematical tools we have because those are the only way we can understand them.

There's no reason to believe that any of them are fundamental to nature. There may be an uncountably infinite number of phenomena out there which are fundamentally unable to be formulated in any way which we can understand or predict them.

Our minds aren't constructed to understand reality, they're constructed to help our bodies survive in a savannah a few hundreds of thousands of years ago.

View on HN

>Our minds aren't constructed to understand reality, they're constructed to help our bodies survive in a savannah a few hundreds of thousands of years ago.

I agree with your main point, however I don't find this a compelling argument. Regardless of what our brains are constructed for, with pen and paper and a set of rules we can roughly approximate a turing machine. Everything we as humanity know about computation suggests that at minimum this is equivalent to any computation possible using all known physical properties & interactions with some amount of overhead (even quantum can be simulated classically).

Thus, while it's possible there is an unknown kind of interaction that we cannot understand or simulate classically, the fact that our brain is more or less designed around survival and reproduction is irrelevant; change or optimize our DNA and the new super-human will still be at best roughly equivalent to a turing machine in terms of computation.

There are computation devices more powerful than the Turing machine, which is why we have complexity classes. (P, #P, #BQP are some examples described by different kinds of Turing-like machines.)

So calling humans equivalent to standard Turing machine is ultimately wrong if you even start to assume quantum processes are important in our thinking. (which is actually unproven as of yet)

Then you need to start with a more intricate parallel quantum Turing machine, high end mathematics required already. It is already non-deterministic.

And no, quantum effects cannot be simulated efficiently classically. That is the difference between P and #P complexity classes. Add parallelism, you get #BQP. You would be able to tell something about the nature of reality by building physical instances of those problems and timing them. Timing attack on the structure of reality, structure of time itself, anyone? (Of course the required energies would be ridiculous.)

To falsify any Turing model, you need to answer the question: what cannot be done by any given machine? Or by any machine? It is possible that there's a Goedel trap in this question.

A parallel would be to find and answer to what cannot be computed efficiently by best human geniuses ever. Hard introspective question might I add. Answer probably requires building or finding something more than human.

#P is still decidable by Turing machines.

And, I’m pretty sure that #BQP is too.

(Just with a large slowdown.)

Now, yes, there are models of computation which cannot be simulated by a Turing machine, but I don’t know of any substantial evidence that a physical process uncomputable.

Any computation which has a finite number of possible inputs is computable, using a lookup table, and therefore using a Turing machine.

So, for any physical process to be uncomputable, it must be such that there are infinitely many possible states of it, for it to proceed from.

And, the Bekenstein bound, if correct, would seem to suggest that no structure like that can exist within a bounded region of space.

>There are computation devices more powerful than the Turing machine, which is why we have complexity classes

Equivalent meaning computability, not complexity. The two machines can simulate each other (even w/ exponential slowdown) then it's equivalent.

> cannot be simulated efficiently classically

Efficiency isn't really the point. The OP is suggesting there's a literal wall which makes something impossible to understand. This would require a computational behavior that cannot be programmed as a turing machine.

> To falsify any Turing model, you need to answer the question: what cannot be done by any given machine?

Correct. The answer is that a turing machine can't decide a turing-complete decision problem, so hypercomputation is achieved by equipping a turing machine with an oracle for a turing-complete problem. This actually leads to a hierarchy that is each strictly more powerful than the previous ones by using oracles for problems that are "more undecidable" so to speak (see https://en.wikipedia.org/wiki/Arithmetical_hierarchy )

Either way, we have no evidence that suggests theres a natural process that lives higher up on the arithmetical hierarchy. An example of a physical process that would fulfill this criteria: suppose there were a type of particle that only sometimes collide with each other, and it turns out that they only collide not just by running into each other but also when the sum of their velocities encode halting turing machines, otherwise the particles pass right through each other. A classical turing machine cannot decide the halting problem, so it's impossible to simulate a system with these particles because you never know if they will collide or not.

I agree with your point, but I'm not quite so pessimistic.

Yes we're pretty dumb. I'm continually amazed at how stupid I am, and even other 'clever' people. The way I see it, we're only just barely sentient.

As you say, we've only just managed to get beyond the savannah. Behavioural modernity and the advanced cultures and technologies we have developed arose in insignificantly recent evolutionary time. Therefore we can't have much more than roughly the minimum necessary intellectual capacity to develop a technological civilization on a planet like ours. If we had less, we wouldn't have achieved what we have. If we had much more, well, we'd have developed it sooner and clearly we didn't.

It's possible there were some environmental factors that held back our technological development despite our intellectual capacity, but it's hard to see what those could have been. Our planet seems pretty abundant in resources and environments for us to colonise. It's also possible that once a species develops the required intellectual capacity for technology, it takes long enough to develop the basics to build on that we evolved significantly grater than the minimum intelligence by the time we actually developed a technological society.

This starts touching on theology. The reasons you explained are one of the main reasons I am a fundamental believer in agnosticism. That is, I believe we see reality through a human lens, therefore we're fundamentally incapable of understanding the reality of our universe unclouded by our own limitations (or assumptions). That's why I think the safest wager is not Pascal's, but to say, "Yes, there may be some order to things, but no, I don't have any idea what governs my reality, nor could I."

I don't say this to incite a religious war or discussion (I've purposely omitted the use of the word "God" here), so if we could keep things philosophical that would be appropriate for such a site. I'm curious if this school of thought has another name beyond agnosticism.

Our minds are general purpose information processors, with significant technical debt, shortcuts, and hardwiring between software and hardware. It's optimized for adapting to a number of very different interactions and environments.

To say they're not constructed to understand reality may or may not be true, but is beside the point.

That said, I often share similar thoughts about progress towards the nature of reality.

But the cognitive revolution gave us abstract thinking that allowed us to create mathematics. Since then we can express anything in the mathematical form. It's almost impossible for us to imagine even a 4-D space, not speaking about the N-D space — but we can write the equations for it and solve them.

Reply on news.ycombinator.com