Does P=NP? | Po-Shen Loh and Lex Fridman

Please support this podcast by checking out our sponsors:

GUEST BIO:
Po-Shen Loh is a mathematician at CMU and coach of the USA International Math Olympiad team.

PODCAST INFO:

SOCIAL:

[blakemcalevey-scurr]

Lex has the probabilities reversed. P != NP is the more widely accepted possibility. P = NP means there's a polynomial algorithm for everything that can be checked in polynomial time, which means things can be encrypted about as fast as they're decrypted etc.

[appa]

I feel like probabilities are the wrong language for analyzing this. 3% chance that P=NP is not really a meaningful statement. There is no way you could calculate that. That's literally just gut feeling.

[kvelez]

1:57
I've noticed that too.
*The definite number affirmation that many make.

[deltauniformtangocharlieho]

Is this the mathematical Mandela effect? Lex, go back and check out what your UT-Austin guest said. He said that there is a 2-3ish percent chance that P = NP.

[LightworkingWanderer]

You're a beast brother, I really appreciate the info you've freely given us, I have benefited so much from it! Thank you! 🙇🏽‍♂️🙏🏽 Not to mention the laughs. ☺

[JikeWimblik]

If validity checking a possible number in a sudoku grid is in p time then solving sudoku is in p time how can this issue not have been spotted by now.

[quosswimblik]

You can relatively space network graphs and do Omni symmetric in circled analysis. You can do in circled analysis on sudoku grids. You can also transform the graph into 2 graphs of lower numbers and different rules each with less variation than the original grid which you can convert to .all to find out how to blague via definitive lying in solving the problems in ways that quickly reveal vital clues. But if you lie more than twice in any one test then you may get badly diminishing returns.
That's what the NSA should be seeing if photonic ai and or quantum computing can't get their teeth into.

[AtheistJohnMathers]

We can never perfectly predict the future; not with statistics or totalitarian control or AI/machine learning. We can make approximations and hope that overlooked asymmetries don't bloom into catastrophic storms of chaos. I like some of what Mandelbrot had to say about the flaws in statistical analysis. Gardens of asymmetrical fractal-esque forms can't be truly controlled by humanity or the numerous things that humanity will eventually evolve into provided we don't go extinct. No matter how many pesticides, salt fertilizers, GMOs, slave labor, and Nazi-esque cold corporate control you throw at the gardens in infinite existence you will never be the master. Perfection or a perfect machine or perfect society could only exist as perfect equilibrium; perfect equilibrium being the existence of unchanging truly balanced symmetry free of any internal and external asymmetrical or chaotic forces. Impossible. True equilibrium can only be described as "Nothingness" and Nothingness can't exist because it is a paradox. Either Existence is infinite in every scope and scale and direction or Nothingness is infinite. The fact that things exist shows that Nothingness does not, never has, and will never exist. Existence and Nothingness could not coexist because Nothingness must be absolutely free of asymmetries and chaos in order to genuinely exist. Existence is infinite. The future is uncertain. People and systems and principles of physics are rarities.

[Yamikaiba]

So, isn't it obvious that we already have problems that can be verified more quickly than they can be solved?

Take any example of a computer trying to predict the state of a system that is more complex than it. Like old Go algorithms. The algorithm might take along time to solve a Go problem, but can then easily check against a solution provided by a human.

Am I not understanding the problem?

[thoyo]

"3%..? So you're telling me there's a chance!"

[joshua]

so that's what the simulation hypothesis is all about, "Decades of searching have not yielded a fast solution to any of these problems", Aliens-"Bitch PLease"

[imensonspionrona]

P = NP maybe in a mind that is good at guessing?

[Brandon_Me]

Half way thru, what?
Finished video, oh yeah P=NP

[brucewilson]

This cannot be correct . Something cannot equal itself times something else.

[chrisdooley]

What a silly question Lex. Of course It does

[skarpheinnorsson]

Before I I've seen the entire video:
P=NP
P/P=(NP)/P
1=1/1
P=N
Am I right?