Hacker News Reader

Comment by Steuard | Hacker News Reader

I haven't read the original paper, but (speaking as a physicist) I'm always skeptical when physicists leap into extremely different disciplines and claim to have a grand insight that its specialists have never considered before.

It could always be true: physics really does have a lot to contribute in many areas! But the burden of proof is very, very much on the physicists to justify the relevance of their approach.

In this case, based only on this little summary, I have to wonder whether they are really explaining intelligence or whether they are identifying some broader natural principle for which intelligent behavior is just one manifestation. There may be something quite deep here, but I'm holding off on deciding what exactly it is.

View on HN

Welcome to modern physics, where physicists do unfalsifiable and imprecise things based on bad philosophy. Anything goes as long as it gets you attention anywhere.

https://news.ycombinator.com/item?id=5562156

HNers need to more critical of Science articles. This is where we need good criticism not on some poor hacker's pet project.

Edit 1

"Additionally, a company he founded is exploring commercial applications of the research in areas such as robotics, economics and defense."

That is a major red flag when people posit a universal AI theory and try to sell to the hapless government tech based on it.

http://en.wikipedia.org/wiki/Thinking_Machines_Corporation

> Welcome to modern physics, where physicists do unfalsifiable and imprecise things based on bad philosophy. Anything goes as long as it gets you attention anywhere.

Thank you! I've been saying that for a long time, but much less eloquently.

Your comments seem strangely insensitive, given that I already said that I'm a physicist (and presumably a modern one).

As I've already indicated, there's reason to be skeptical when physicists stretch past the usual boundaries of the discipline. (Though it still could make important contributions.) But modern physics as modern physics is, by and large, solid science.

I am also a physicist doing AI. The comment was not aimed at you! I did not mean modern physics (QM,Relativity etc)! That is rock solid!

What I meant was physics which is closer to the Time Cube theory than rock solid mathematics on a spectrum of math and science.

"As I've already indicated, there's reason to be skeptical when physicists stretch past the usual boundaries of the discipline."

There is reason to be skeptical when anybody proposes a grand unifying theory for something as grand and difficult as intelligence!! Extraordinary claims require extraordinary evidence!

This has been going since 1957 with so many young lives lost (sometimes literally) due to techniques bordering on charlatanism. Sorry to be harsh, I love AI and want it to succeed. It won't succeed if people mix pursuing knowledge with pursuing fame and $.

http://www.wired.com/techbiz/people/magazine/16-02/ff_aimyst...

It's turtles all the way down, until you reach Philosophy

... At which point it's still turtles, but nobody can really agree on what a turtle is, or what it means for something to be turtles, or what it means for something to be anything, including (but not limited to) turtles, or what it means for something to mean something. By the time the subject is settled -- if it ever is settled -- the turtle is dead. And so is God.

(For some definitions of "turtle", "dead", "God", and most especially "is".)

Wasn't arguing about the definition of "is" a topic when the Linguists got bored and tried to invade the Philosophy Departments?

>I haven't read the original paper, but (speaking as a physicist) I'm always skeptical when physicists leap into extremely different disciplines and claim to have a grand insight that its specialists have never considered before.

That's bad, because the most interesting breakthroughs came from cross-disciplinary work.

I'm curious as to the basis of your statement. I believe the opposite to be true. Gene theory, the general theory, anti-biotics, the structure of DNA, the function of neurons, and the transistor all came from specialists. Is it not the rare breakthrough that comes from cross-disciplinary work?

The cross-disciplinary breakthrough is often characterized as unifying, and hence rightfully celebrated. But they are neither the norm nor the most influential.

the structure of DNA, the function of neurons

Crick, Hodgkin and Huxley were all physicists poaching in other fields. In fact, relatively few of the people who laid the foundations in molecular biology were trained as biologists.

> Crick, Hodgkin and Huxley were all physicists

This recollection is incorrect across the board.

Crick had only a bachelors in physics (which does not a physicist make), and his postgraduate training was in medical research. Huxley, an autodidact, had his background in medicine. I'm not sure about Hodgkin but I suspect he was also not a physicist.

Crick was perhaps halfway into a Ph.D. in physics when the war broke out, during which he basically did very applied physics for the British. After the war, he did a Ph.D. under Max Perutz on X-ray diffraction. Crystallography is very useful in biology, but it is itself very far from a biological topic. Crick wrote of experiencing extreme impedance mismatch when he switched from physics to biology and, perhaps most significantly, Watson often referred to Crick as 'the physicist' or to 'having him do the physics' (sorry, I don't have the full text in front of me) in The Double Helix, suggesting that his approach to biological problems was primarily a physical one. It is true that Crick did work under the Medical Research Council, but it is somewhat misleading to say that he was doing medical research in the conventional sense. He was studying the biophysics of cytoplasm, which is not something that people trained in medicine do.

I admit that Hodgkin and Huxley are harder cases: both had early inclinations to physics and then got sidetracked into physiology. Both of them spent the war doing things you'd have a physicist doing during a war, and afterwards did their their famous work on the action potential. I read the Hodgkin and Huxley papers, but I read them in a course in the math department-- they contain some of history's finest examples of mathematical modeling, relying heavily on dynamical systems and circuit theory. It is a regrettably rare biologist who wants to go near those papers. It is manifestly not the work of traditionally trained biologists.

Disciplinary boundaries in mid-century Britain were somewhat different then, and I realize the danger of playing No True Biologist. But the mainstream of biology is only very recently acknowledging the importance of the physical approach: Hodgkin and Huxley could have been Hodgkin and Huxley without a lick of genetics or ethology, but they couldn't have been Hodgkin and Huxley without the cable equation.

I would add the fun fact that the applied physics work that Crick was doing was experimental work in fluid mechanics when a bomb fell through the roof of his laboratory and destroyed his experiment! This was certainly a factor in prompting his to switch fields.

I would also add Max Delbruck to your list of turn of the last century physicists cum molecular biologists and the physicist members of the RNA tie club: http://en.wikipedia.org/wiki/RNA_Tie_Club.

I also have a question for the GP. What exactly makes a physicist so? I have "merely" a masters degree in physics, and although I consider myself to be a theoretical biologist, I am often called a physicist by my collaborators (some of whom are physicists themselves). What about a degree actually matters in this descriptive title?

> Crick, Hodgkin and Huxley were all physicists poaching in other fields.

And Fisher was a statistician.

And Mendel was a monk.

Most likely, an unifying framework of nature would come from cross-disciplinary work. That's almost a tautology. Unfortunately we're stuck in a world where isolated people playing with narrow albeit very detailed views of nature, try to capture the big picture all on their own.

I'm glad physicists can try their luck at AI. I'd be glad other disciplines, even remote, do too. Foreign concepts often bring with them new outlooks on problems that are very fertile and rejuvenating. A little bit more humility wouldn't hurt though.

> That's bad, because the most interesting breakthroughs came from cross-disciplinary work.

True, but I think the signal-to-noise ratio is also exceptionally low. The most amazing breakthroughs come from insights that span disciplines, but it's also easy to think you have an insight when what you really have is a mistake. I'm not weighing in on this one in particular, but as a general principle I think caution is appropriate in cases like this.

Physicists are doing quite well in economics, according to economist Steve Keen in his book Debunking Economics.

Agreed

"My proposal is that intelligence is a process that attempts to capture future histories"

'future histories' is just ... the future! .... so his great proposal is 'intelligence predicts the future'? Well duh!

Errr no. His proposal is that intelligence is a process that maximizes the number of available future histories. Just capturing future histories is meaningless. The example he gives of an upright rod clearly illustrates the difference. Really what he's saying is that intelligence seeks to maintain maximum possibilities for the future.

He just set a goal of keeping the potential energy maximized for the rod, just like a big blue set a goal of winning at chess 30 years ago, both by 'predicting the future'.

I don't see anything new here except the phrase 'future histories' which is not even defined because it just means 'the future'.

As soon as I saw it claiming it can win at the stock market I knew it was BS.

It's not "predicting" the future, it's "maximizing" the number of potential futures. In chess, the more moves you have available the better you are doing, generally. But another measure (probably better although I'm not a chess expert) is how few moves your opponent has available. Strictly speaking it sounds like this strategy would emphasize keeping future moves available for both you and your opponent. This is a more idealistic future for many real-world situations but I'm not sure I'd call it the definition of "intelligence".

Yeah, seems to me their idea can be summarized as "in the absence of a specific goal, keep options open".

Which is generally good advice, but only "in the absence of a specific goal".

I think that's what is different between machine and life. Machine always has specific goal. Life never has.

Well, most life has a tendency to reproduce (inherited from ancestors who all reproduced). Aka evolution.

@klipt: Reproduction is not a goal. It's requirement of the physical system. There would be no evolution without reproduction.

It sounds like crazy nonsense but there's just enough in it to sound crazy awesome.

I've got like, an undergraduate understanding of how this works. Does anyone have a link to an ungated copy?

Yes: http://alexwg.org/

Whelp, that's completely unreadable to me. Undergraduate computer science fail.

I got as far as "the model of this physical process of entropy can act as a search function for lower entropic states, which given some poking, can resemble problem solution states".

It's the poking part that is problematic.

>To better understand this classical-thermal form (11) of causal entropic forcingwe simulated its effect [12,13] on the evolution of the causal macrostates of a variety of simple mechanical systems: (i) a particle in a box, (ii) a cart and pole system, (iii) a tool use puzzle, and (iv) a social cooperation puzzle

I am wildly unqualified to comment on this paper, but it's suspicious that you would jump to a "tool use puzzle" and a "social cooperation puzzle". It yields a really fun result and a fun way of talking about cognition, but it seems unsurprising to say that once you describe a problem in the physical terms of this framework, your framework can then solve the problem, as in,

>We modeled an animal as a disk (disk I) undergoing causal entropic forcing, a potential tool as a smaller disk (disk II) outside of a tube too narrow for disk I to enter, and an object of interest as a second smaller disk (disk III) resting inside the tube

The notion that currently understood physical models can act as a learning system, given the right parameters, is really cool and exciting, and there's a great simplicity and metaphoric appeal to it.

phill, this is hacker news, and if someone finds you're wildly unqualified to comment on something, don't worry, he'll be vocal about it and explain exactly why he thinks you're unqualified to comment about it. That's why I love this site. On any similar site aggregator, I'd get cat pictures and people telling me my opinions are stupid because my face is stupid.

More towards the topic, I'd say these kind of articles have to be taken with a grain of something, preferably salt or cynism, depending on taste. I'm usually more of a material science/electronics technology buff, and news articles in that area will not hestitate to present a relatively minor paper as a breakthrough in battery technology that will solve all our energy problems in the next 5 years. But then you read on, compare it against the existing technologies and you realize that they just did something funny to the diodes that increases power density at the cost of usuage time. Somewhat handy in some applications, but not a game changer.

PS: You're certainly not the youngest here.

Then there's lesswrong.com, where I've never felt qualified to even comment at all.

Reply on news.ycombinator.com