Hacker News Reader

Comment by jeffreyrogers | Hacker News Reader

Whenever this topic comes up I think of Philip Anderson's[0] paper "More is Different": https://cse-robotics.engr.tamu.edu/dshell/cs689/papers/ander...

It looks like most of the big ideas have been discovered (although physicists were saying this in the 1890s as well), but there are still many things whose collective behavior is not well understood.

[0]: https://en.wikipedia.org/wiki/Philip_W._Anderson

View on HN

> It looks like most of the big ideas have been discovered (although physicists were saying this in the 1890s as well)

Are many people really saying that now? I'm emphatically not an expert, but my impression was that it's generally agreed that there are clear theoretical signs (quantum physics vs. general relativity) as well as experimental ones (dark energy, dark matter etc.) that our understanding of fundamental physics remains highly incomplete, and that completing it will likely require a lot of very novel physics. There are some hot takes that there may simply be no grand theory that could explain all the relevant observations, but AFAICT they're far from being the general consensus. The more common pessimism is that humanity has been stuck in roughly the same place for several decades now, and there's no clear sign of any experimental or theoretical line of attack that (still) seems likely to get us out of it. This situation doesn't really resemble the complacent belief in the near-completeness of fundamental physics that supposedly existed in the 1890s. (And I'm not sure if that complacency was actually as strong as it's often claimed to be, at least among actual research physicists of the time as opposed to onlookers and popularisers.)

In some respects it seems like you're arguing P=NP. It may not be the case however. It's certainly a possibility that the system cannot be perfectly determined from within the system.

I'm not taking any position here myself on what's possible. Similarly I can't say whether finding a theory to unify general relativity and quantum mechanics really depends on P being equal to NP, but I can say that it certainly doesn't seem to be generally accepted that it does.

What does P=NP have to do with that?

Did you mean to allude to Godel's incompleteness theorem instead?

It's very possible, likely even, that enormous areas of physics still have plenty of room for major discoveries and subsequent applications to new discoveries. Furthermore, tying even increments of these major possible discoveries to the existing field of knowledge and application would by itself lead to a great deal of new exploration. But what if we haven't advanced much in several decades because we now suffer from or soon will suffer from the Dog Problem?

Basically: no matter how hard you try, you couldn't teach a dog how to perform engine mechanics (or any number of other higher cognition things), simply because its cranial biology isn't up to the task of fundamentally understanding such things. I have a sneaking (layman's) fear that we might be close to our own version of the same thing in certain fields.

It would be nice to argue that AI augmentation of human-led scientific advancement will help us along considerably, but it still seems deeply hand wavy to say that. Sure, computers help enormously with vast amounts of technical number crunching, and enable the practical application of physics (among other fields) in ways that we could never achieve without them, but fundamental, self-directed discovery is still way beyond the scope of any AI I know of or foresee for the near future.

I agree with the sentiment, but would say it slightly differently: I think most big ideas about fundamental physical laws that are testable by present-day technology have likely been discovered. This spells the "end of physics" only if one defines physics narrowly (IMO as a non-physicist) as only the search for fundamental physical laws.

Edit: emphasis on "only" added

Reply on news.ycombinator.com