Mystery of Entropy Finally Solved After 50 Years (Stephen Wolfram) | Hacker News Reader

Mystery of Entropy Finally Solved After 50 Years (Stephen Wolfram)

youtube.com

Can anyone make sense of this? I haven't watched the whole thing (on my phone) but it seems simply wrong to me. Like the part about the disordered state being like encryption that could be reversed. The 2nd law is about the entropy of a closed system. How does one decrypt without producing more entropy?

Another thing in find fascinating is if the universe did run backwards in time, we would never know as our memories only form by a process that produces more heat than order.

I usually don't have the patience for videos, and I'm going to wait for reviews before considering his book, which on Wolfram's site is at: [0]

However, maybe this helps: (1) perfectly reversible systems do not inherently increase entropy, and (2) irreversible systems generate entropy because they throw away information that could otherwise be used to run the system backwards.

The role of information in thermodynamics typically arises in Statistical Mechanics [1]

[0] https://www.wolfram-media.com/products/the-second-law-resolv...

[1] https://en.wikipedia.org/wiki/Statistical_mechanics

I usually don't have patience for videos and this is no exception. It's a bad format to understand a new thing, in my opinion.

I don't know if the fantasy of reversible systems is a new thing. What's the innovation?

You appear to be unhappy with the notion of reversible computation, but that is not the topic here.

Time reversibility of most physics at the subatomic level is standard, mainstream, but also posed one of the biggest physics puzzles of the 20th century. It is still occasionally presented as a mystery, but generally speaking it is entropy at a macroscopic level that gives time a one-directional forward arrow.

Edit: your profile makes it look like you might be a scientist, in which case nothing I said was new to you -- but then I don't know what you're referring to.

Videos might be good for people to learn who don't want to read a book, I suppose, but books are typically far more information-dense.

The closest analogy I can think of is that of quantum decoherence--large things can in theory be in entangled states but they aren't.

I think the topic here though is about the large scale and time reversibility isn't standard/mainstream.

I don't have an opinion on reversible computing.

I don't know what his revelation about entropy or reversibility is supposed to be - I guess I was just fishing for the two sentence summary.

I assume it involves cellular automata.

> (2) irreversible systems generate entropy because they throw away information that could otherwise be used to run the system backwards.

It seems to me that the video is saying there are no irreversible states, only that we treat multiple states similarly in some cases and not others. It also assumes everything to have a definite measureable state and be perfectly deterministic. I would need more convincing to buy into these.

All of that appears to directly contradict things established by experiment, as well as mainstream theory, so I dunno.

Like most of Wolfram's scientific work, it's hard to pin down whether he's a genius or an eccentric crank. He built his own theory of the universe from first principles. It all sounds plausible, but most of the large-scale assertions are only empirically verifiable with the necessary compute power (his law of computational irreproducibility says effectively that). It makes for really cool cellular automata graphics but I'm not convinced that it's necessarily a new kind of science with broad explanatory power at scale. But it might? Maybe I'll try to read NKS again, I "only" got 200 pages into it the first time.

I'm generally impressed with his software work though. Later in the podcast, he talks about training an LLM to write Wolfram code and automatically execute it, with the ability to refine the intermediate code. I've always appreciated Wolfram Alpha as a powerful computation tool and combining it with an LLM is a pretty creative approach. You can express a complex analysis in plain language but, instead of getting a full response likely to contain hallucinations, you get some code in a high-level programming language to then generate the response. Even if you just execute it and accept the result, at least you have a copy of the "logic" used to generate your result. Neat approach to escape the black box problem of LLMs, at least for the kind of questions that you'd normally write in Wolfram.

WA genuinely helped me understand thermodynamics and allowed me to offload painful calculation to focus on bigger stuff doing my PhD

But it seems less clever than it used to be. It's perfectly happy to ignore inputs and return something simplified to pointlessness

It sounds like openAI is more quickly undergoing a similar crippling

Yup, Wolfram Alpha is much worse than it once was. Even simple things like comparisons using "vs" no longer works. Peak WA was a few years after release. It also used to be able to evaluate Mathematica expressions, which it can no longer do.

Reply on news.ycombinator.com