## Monday, January 25, 2016

### Polchinski, Bayes, Woit, Weinberg, and the multiverse

Joe Polchinski was the only participant of the December 2015 Munich workshop on testability who didn't physically attend the meeting (his spirit and world view was represented by David Gross, an actor). He was also the only defender of the multiverse over there. But his talk has made a bigger impact than the talks of anyone else, anyway.

Polchinski focused both on the multiverse and string theory that has been a tremendous success so far. Today, he released a continuation of his essay on the arXiv:
Why trust a theory? Some further remarks (part 1)
He offers a new perspective on his previous claims.

First of all, Polchinski emphasizes that the Bayesian calculation of the "probability of the multiverse" – he ended up with the result 94% – isn't really the key point here. The physics – the physical observations that enter the "Bayesian calculation" – are the essence. The numbers he uses in the Bayesian inference are in no way "precise numbers" whose detailed values matter.

But his calculation was meant to be a quantitative description of some arguments that are really qualitative and whose strength can't be quantified too accurately. And these arguments – revolving around the tiny, positive but nonzero cosmological constant that seems much more likely to be explained by the multiverse than by a non-multiverse theory (and the multiverse explanations seem compatible with string theory) – are the things that matter.

Polchinski paints himself as a physicist who hasn't "planned" to become the anthropic believer but he was forced to adopt these viewpoints by the evidence that was emerging, particularly in his own research as well. Steven Weinberg is praised as a man who knows how to look at generally overlooked possibilities and directions and identify the most general conclusions that follow from a paradigm while making the minimum set of assumptions.

We also learn how Raphael Bousso and Polchinski cooperated on their 2000 paper showing that the "discretuum" of the flux vacua seems to be a way how string theory generates the kind of a "landscape of vacua" that are needed for Weinberg's predictions of the cosmological constant to work. Polchinski tried to slow Raphael Bousso who was too anthropic. Later, Joe Polchinski had to visit a psychiatrist because the realizations about the multiverse turned out disappointing for Polchinski – and maybe for other, non-multiverse-related reasons, too. Several similar human things about Polchinski's relationships with the multiverse and the anthropic lack of principles are outlined in an intelligent way.

In the last, fourth section, Polchinski discusses some particular critics.

He begins with Ellis and Silk. They're wrong because they don't appreciate the relative importance of theoretical arguments in quantum gravity. The problem is hard which is why these arguments are powerful. Joe focuses on Ellis' and Silk's promotion of "unimodular gravity" which tries to erase the equations (and predictions) for the "scalar" degree of freedom in gravity, one influenced by the cosmological constant. Joe says that if Ellis and Silk were using their own definitions properly, they would have to agree that the multiverse is "more scientific" than the unimodular gravity because the former predicts something about the cosmological constant while the latter doesn't. Ellis and Silk don't want to see this fact, probably because they have fooled themselves by prejudices.

Polchinski shares the irritation by the phrase "non-empirical confirmation" with Silk and Ellis – and so do I. However, he also notes that Richard Dawid who uses this problematic phrase doesn't really mean anything terrible – his opinions are not too far from Joe's (or mine, for that matter).

Update: After Joe was harassed and blackmailed by Mr Voits, he erased the section on Mr Voits in V2. You may get V1 of Joe's paper on the arXiv or look at Joe's website for the full version.

The final subsection is dedicated to Peter "Woit" (Mr Voits uses this distorted surname to pretend that his family had no responsibility for the Holocaust), a canonical example of an aßhole (Joe uses the term "cynic"). Polchinski points out that "Woit" is just a pile of stinky hypocritical anti-scientific crap. Or, using some of Joe's words,
It is disturbing to me that my colleagues in other fields might believe this of Weinberg, or for that matter believe that Woit has anything useful to say about science. His writing shows no respect for science, for scientists, and ultimately no respect for himself, if this is the best that he thinks he can do.
Right, if "Woit" had at least some respect for himself, he wouldn't be forcing himself to swim in the cesspool of superficial, stinky, vitriolic lies every other day.

A particular comment is copied in which "Woit" described Weinberg as a "joiner" who likes to jump on bandwagons. Polchinski points out that Weinberg is among the top people who don't deserve this description because he is capable of starting the sensible research of overlooked possibilities – while it is really the bloggers who are characteristic by their desires to talk about every fashionable topic. I completely agree with that. Polchinski has also mentioned the 2005 opinion by a colleague, Mr X (I may know who Mr X is), who said that Weinberg was senile. Weinberg's output just in the last 10 years is enough to disprove the claims about the senility.

I don't share Polchinski's and Weinberg's feelings of certainty or beliefs in various claims about the multiverse – but especially the more "radical" claims about the anthropic principle (Polchinski says that the experimental physics has become basically worthless because the experimenters only measure random numbers – well, there's an element of randomness but the "kind" of results they will get will influence the thinking about the fundamental theory and the right stringy vacuum, anyway, because not every "random distribution" is the same) – and I also think that Weinberg's asymptotic safety, Polchinski's firewalls, and Weinberg's recent cynicism about the foundations of conventional quantum mechanics aren't promising ways to think, especially because they're demonstrably wrong in many respects.

But I think it's obvious that Joe's texts such as this one are vastly more scientific and intelligent than the texts by virtually all string theory or multiverse critics. He talks about the big questions as well – but the actual relevant physical arguments, and not cheap demagogic talking points, play the most important role in his reasoning.

These preprints by Polchinski are non-technical to the extent that they could have been posted as blog posts on a blog instead of the arXiv. But I think that it's right for researchers of Polchinski's caliber to contribute "general" preprints of this kind and he (and others) should have posted many of those already a decade ago (or more). However, I also hope that the arXiv won't become the home of similar philosophical essays by lots of authors who are not comparable to Polchinski.

BTW I predict that Joe will also submit a 2-author 29-page technical paper on a different theme to the arXiv as soon as tomorrow.

Off-topic: Casey Neistat showed how cool it is when you may still find free and courageous people in America. During the snowstorm, the New York nanny-ish mayor unsurprisingly banned car-powered snowboarding in the streets but some people don't care much and when it appeared, NYPD was kind enough. Music by Sinatra, 6+ million views in a day. Behind the scenes.

Super-strata

I also want to mention the first hep-th preprint today,
Momentum Fractionation on Superstrata,
in which Bena, Martinec, Turton, and Warner worked out something about the "bulk" description of the superstrata. The superstrata are bound states of D1-branes, D5-branes, and momenta (in the space where the backreaction is neglected) that are highly numerous because their identity depends on whole functions of at least two variables. Note that these are the same charges as those carried by the Strominger-Vafa black hole.

These superstrata should be distinguished from fuzzballs which are solutions of gravity (where backreaction is essential).

The superstrata implicitly know about the ability of D-branes to divide themselves into "fractional D-branes". Many things have been known about the superstrata (that basically live in the "boundary theory") and the fuzzballs (that live in the "bulk" and also depend on whole functions).

But these four authors have done quite some work in connecting these two descriptions.

They have employed some "solution-generating techniques" to construct the solutions of supergravity (i.e. some objects in the bulk) out of the configuration of the D1-brane and D5-brane charges. It turns out that the bulk solutions dual to the superstrata (D1-branes and D5-branes twisted according to some full functions) are basically smooth, horizonless geometries. But they differ from Mathur-et-al. fuzzballs by having some (well-understood) orbifold singularities, too.

I do tend to think that this whole program of visualizing individual black hole microstates is ultimately correct and may be completed to a "full theory" – but it's not the "only correct theory". Note that there are at least two new descriptions of the microstates – superstrata and fuzzball-like solutions, if you wish.

My eclectic view that I have promoted for years is that string theory actually allows lots of these descriptions to co-exist and they don't contradict the other descriptions of black holes that do envision the horizon. The idea is that you may choose basically any background you want – and string theory finds a way to describe the Hilbert space of microstates in terms of excitations of that background or those backgrounds. However, if you insist on a "few low-energy fields" on top of the nearly unique horizonful black hole solutions, the available degrees of freedom won't be able to distinguish all the microstates from each other i.e. to parameterize all of them. This is the case of the Raju-Papadodimas formalism, I think. But that limitation doesn't make the theory "incomplete" because the infalling observer doesn't have enough time to measure a complete set of observables (and pinpoint the exact microstate) before he dies, anyway.

Superstrata as well as fuzzballs involve an exponentially large number of "backgrounds". But there also exists a map relating these particular microstate geometries or configurations with the local excitations of a smooth Schwarzschild-like geometry. Much of the detailed knowledge of string/M-theory then reduces to the analysis of all similar dictionaries. The smooth horizonful black hole is obtained by tracing over some microscopic degrees of freedom of the fuzzballs or the superstrata, for example. The transition between these two descriptions is sort of analogous to the ER-EPR correspondence as well as various stringy dualities.

By now, because of all the dualities, correspondences, and complementary descriptions, we should have gotten used to the fact that for a given microstate or a collection of microstates, the "right background geometry" is in no way uniquely determined.

You know, my broader point is that there are many "paradigms" how to describe the black hole microstates and these paradigms look sufficiently different and cannot be reconciled into "one and only description". But it's not a problem because they may still be complementary to each other. They are different ways to "geometrically visualize" the Hilbert space in terms of some quasi-local fields and similar operators, assuming possibly very different background geometries. T- and U-dualities in string theory (plus mirror symmetry, string-string duality) as well as holography may be interpreted as examples of these "complementarities", too.