Tuesday, January 24, 2006

When Krauss and Susskind are right

Lenny Susskind, a co-founder of string theory and one of the most original thinkers of the several past decades, has written a book that has made many physicists - including Lenny's fellow anthropic believers - upset.

The less problematic part of the book presents some ideas about cosmology and string theory that the readers have seen elsewhere. The more problematic part of the book promotes the idea of the anthropic principle in such a way that makes it clear that the basic belief in a huge number of other Universes is as religious as any other religion, although Susskind himself denies this correspondence.

After having explained that the arguments based on fine-tuning are essentially identical to those that have been used by the promoters of Intelligent Design, he also argues that this equivalence does not exist either although the explanation is not quite transparent.

Lawrence Krauss has made a comparable and possibly even more serious "sin" by having written his own book. He advocates the idea that all models and theories that require some thinking, explanations based on mathematics, or new concepts such as extra dimensions are a priori known to be probably wrong. Lawrence Krauss believes that extra dimensions are a priori ridiculous and that he can prove it without writing any papers that actually present any physical arguments: pure thought and a popular book addressed to the laymen is apparently enough. Sometimes you wonder whether Lawrence Krauss also believes that a priori it is clear that particles must have well-defined positions and velocities and that "now" must have an invariant meaning.

As you can see, one half of Lenny's book and essentially the whole book by Krauss are scientifically flawed, so you may ask: can it happen that your humble correspondent would endorse whole articles written either by Krauss or by Susskind in 2006?

The answer is, of course, a resounding Yes. It happens whenever Krauss and Susskind exchange their ideas with someone whose opinions are sillier than their own at least by an order of magnitude. Given the high intelligence of Krauss and Susskind, the number of such people is very large, of course, and one of them is called John Horgan. In this particular case, Susskind and Krauss even wrote their answer together.
Who is John Horgan? John Horgan is a leading science-hater. As a completely average human with common sense that is based on the 14th century agriculture, he has no understanding of modern science itself but he has been given numerous opportunities to speak with top scientists and misinterpret their words. He has also written a book called "End of Science". In this bizarre book, he argues that science is approaching its end much like the belief in science. He tries to humiliate top scientists (including Edward Witten) and twist their statements to support his basic silly claim. It is not just science that is ending according to Horgan: we are also approaching the end of progress, end of physics, end of cosmology, end of evolutionary biology, end of social science, end of neuroscience, end of science of complex phenomena, and many other ends.

Given the fact that some of these fields are really hot these days and they are expected to revolutionize our lives in the 21st century, you may think that there will be no one who will agree with Horgan's madness. But you would be wrong. And unfortunately, we are not speaking just about the crackpots from Peter Woit's blog in particular and intellectual trash in general.

A more realistic description of the situation is that there have always been people who did not believe in science. All those people who have lived since 1700 may be described by a word that starts with "ID". The year 2006 and John Horgan are not special in this respect: John Horgan is just another spot in a long sequence of billions of his peers.

Even if the progress in some particular fields slowed down - and there are many fields Horgan mentions where the progress is, on the contrary, speeding up - it does not imply and cannot imply that the belief in science as such should be diminishing. What we have learned cannot be unlearned. And as long as we are humans, no one can kill the natural human curiosity.

When I mentioned the crackpots from Peter Woit's blog, they are having a hard time with the fact that Lawrence Krauss, who was believed to have retracted all of modern physics, suddenly argues that theoretical physics is a legitimate science. It would indeed be pretty serious if Krauss or even Susskind endorsed Horgan's general moronic statements. Susskind and Krauss have helped, in a sense, a whole gallery of anti-scientific bigots including personalities such as William Dembski or John Horgan. If Susskind and Krauss were not protesting against the abuses of their books by these people, they would become two of them.

I, for one, believe that there is still a significant gap between these groups of thinkers, despite Krauss' misunderstanding for the basic motivation behind the ideas of modern particle physics and Susskind's unscientific approach to the vacuum selection problem. Susskind and Krauss are famous colleagues of their fellow physicists, Horgan is not.

I guess that the reasonable ones will continue to live in our fantasy realms where scientific arguments matter and where fantasies often become true - and sometimes they are guaranteed to - leaving Mr. Horgan in his intellectual s**tland.


  1. Lubos:

    Most people would agree that anything that has a beginning must also has an ending. Everything we know of nature observes that philosophical law: Everything that has a begin, also has an end. Sciece clearly has a begin, since science started to emerge when human first learned to make tools.

    So I would not think that science would be an exception by having a begin but with no end in sight. But I may not agree with John Horgan in exactly how close we are to the end of science. But maybe he has a point. Big particle accelerator science, for example, is pretty much near its end. LHC will not become functional, due to lack of energy. Even if LHC becomes functional, it is the end. No one can imagine a next generation machine 100 times more powerful than LHC, consuming 100 times more electricity, lying in a tunnel 100 times bigger in diameter, and cost 100 times more money to built. It will not happen. 100 times LHC power consumption is 20, 000 mega watts. That's more electric power than the whole of France and Swiss consumes.

    Not to meantion that a future nuclear war, possibility of which can not be ruled out, could easily wipe out the whole of human civilization in the next hundred year or so. More terrible weaponries could be invented in the future due to advanced technology. In that sense, judging from long term, I do not see what human race as a whole benefited from modern science. If we lived like monkeys, we could survive the next few million years without problem. But as a nuclear armed race, it's quite questionable even on whether we could survive the next 100 years.

    Lubos said that anything we have learned, for example quantum mechanics, could not be un-learned, we could only learn more in the future. Not true. All human activities will have to obey the second law of thermal dynamics, learning process included. We can continue to gain negative entropy on the price that the nature provided an entropy sink for us so the total entropy do not go down. We are fully contained within a bottle of finite size. Once we reach our limit we can not grow any more and we can not learn more.

    Not only a never ending learning process breaks physics laws. There are also plenty of counter examples to show that learning is never a never-ending process. Things that we learn CAN and has been un-learned, both for us individuals and for society as a whole. As individual, you forget something you knew after you have not used it for a while. As a society, we also forget things that we no longer use.

    One example is the art of Alchemy. No modern human know how it was practiced. You probably can still find some printed material in library and dig the information out. But overtime books become rotten and lost and eventually even the word "Alchemy" could be forgotten. We have many ancient languages that once was studied and spoken by many people. But these languages have long forgotten, and archiologists have a hard time figuring out what they mean.

    There are plenty of example of lost human knowledges. The ancient Mayas were capable of extremely precise calculations of astronomy terms and they knew the precise length of one year to within no more than 7 seconds discrepancy from our modern measurements. With no modern numerical calculators and no telescopes to look at the sky, and not even written languages available to the Indians. How did they do that and how did they figured out their knowledge. It's a riddle lost forever. So that's something the Indians learned but un-learned later.

    Another example being the Polynesians. They spread out to occupy virtually all islands in the Pacific Ocean, within a short period of no more a thousand years, using very primitive tools like canoes. Today, the suboriginal people from New Zealand and from Hawaii could talk in their native lanugaues freely, without an interpreter. You know that. It's like a miracle. They look at the sky and they knew exactly where they are in the ocean, without modern satellite GPS. They put a hand in the water, by sensing the tiny fluctuations of the water waves reflected from remote islands, they can tell the direction and distance of an island a thousand miles away. That's how they discover and occupy all Pacific islands. Those are amazing human knowledges but they were lost.

    There's compeling proofs that knowledges can be acquired, but can also be un-learned and forgotten.

  2. Hi Lubos -

    It may well be that not only is scientific progress not going to slow down and stop, but it could continue to grow exponentially. Science (and the production of all other ideas) is in general an evolutionary process, iterative trial and error, where each new generation builds upon the successes of the previous one. Say the number of memes at some time t is m(t), then to get exponential growth you just need the rate of production of new memes dm/dt to be proportional to m(t). This is quite reasonable - the more ideas there are then the more starting points there are for new ideas, and also the more adaptive niches there are to fill.

    Various examples from biology might lead one to conclude that the exponential growth must level out at some point, but that isn't necessarily the case. The really interesting question is whether we will be able to develop human level artificial intelligence, as it then wouldn't take that much longer - perhaps less than a decade - to develop beings considerably smarter than any unenhanced human being. It is certainly a formidable problem, but researchers are making nice progress in understanding the algorithms that the brain's neural nets use to learn information, and it appears that we will have the necessary computing power to run models with billions of artificial neurons - the largest supercomputer in the world, Blue Gene L, running at 280 teraflops, may already have enough power if simple neural models are sufficient.

    It is certainly a possibility that there is no finite upper bound to what we can learn and do. I suspect that this is the case, since, first, it seems reasonable that all mathematical structures actually exist (call this vast collection the ensemble). It then stands to reason that the unbounded growth scenarios dominate the counting in the ensemble, and thus that is what we statistically expect to find happen. Indeed, I propose this is why we are observers at all (we are, after all, only a specific type of information, as encoded by the neural patterns in our brains): observers can extract information from any other type of structure, and thus observers form the largest class of information in the ensemble.

    Well, we'll see!