Bayesian stats book banned in China | Hacker News Reader

Bayesian stats book banned in China

stat.columbia.edu

159 pointsagconway78 comments

Hi Guys,

I am from China, and of course the Bayesian is taught in China. Only in the culture revolution time, this crazy thing may happen; but it's long time ago when Mao was alive. Now China is a modern society; of course, the Bayesian theory is taught in schools; otherwise how could they send the spaceship to the outer space and have the fastest super computer in the world so far. you are smart hackers, not a bunch of kids controlled by traditional biases. As to censorship, that's right, china has notorious censorship. I guess the author's book must have some contents/examples regarding political issues; not Bayesian itself was the reason for the book being censored. These days, when people talked about censorship, I always think of wiki-leaks ......

Launching a spaceship almost certainly uses Bayesian methods for tracking (e.g., Kalman or particle filters for position tracking), but I seriously doubt that building a supercomputer requires a dedicated use of Bayesian statistics.

On a different note, perhaps you could be slightly less condescending in your future posts.

I didn't find his tone too objectionable. Always consider that getting the tone correct can be difficult in a foreign tongue.

Understandable, but for future reference to anyone who might be interested, telling strangers to "use their brains" is generally not appreciated in many cultures, since it implies that they weren't.

points taken. I updated my original post. sorry for that condescending tone.

Also, as China is surely the least transparent nation among world powers, I think it is unreasonable to expect foreigners outside the curtain to have an understanding of what really goes on behind the curtain. The lack of transparency fuels wild speculation.

I guess you meant "to reason" but wrote it as "to move(use) brain"?

Is a Kalman filter considered a type of Bayesian statistics? It doesn't seem that way to me, but I'm not so hot at higher math.

In any case, many years ago, I worked on a radar system for a fighter plane. Its air-to-ground radar (at the very least; this was the part I was involved in) did indeed use a Kalman filter.

Yes, you can describe tracking and Kalman filter using bayesian statistics. Although, my prof said that originally it was not developed that way, but it is easier to describe it using normal distributions as priors and likelihoods of the position of the object being tracked.

And I guess thanks to those crazy generalizing mathematicians, nowadays you can use something like Kalman filtering with distribution other than normal ones.

Check out particle filters if you want crazy mathematicians. Fun stuff.

Yup. It's a special case of Bayesian estimation that is derived when all the conditional distributions are Gaussian.

You should have noticed in the WikiLeaks case that there are a lot of Americans criticizing our government. It sounds like you are proud of China, but the truly patriotic thing to do is to criticize your government when it does wrong so that it can improve.

If you're thinking about Wikileaks, think about how the official reaction is being widely regarded as a very overbearing, and very un-American thing, not as a typical expression of well-established values.

Indeed, for a summary of well-established American values, see the collection defined by the Pentagon Papers, the Watergate story, and more recently, the NYT's Warrantless Wiretapping story. Together, they put the Administration on the wrong side of our national character, and to a very embarrassing degree.

So I agree with you; the Wikileaks response doesn't just look bad, it is bad. But you join others on the more dubious side of the issue when you cite errors like this as excuses or justifications for vastly more abusive practices elsewhere, rather than violations of civil liberty that shouldn't be tolerated by self-respecting people anywhere.

"But you join others on the more dubious side of the issue when you cite errors like this as excuses or justifications for vastly more abusive practices elsewhere, rather than violations of civil liberty that shouldn't be tolerated by self-respecting people anywhere."

Or to put it more bluntly: "America sucking is no excuse for China sucking" ;)

It's good Assange didn't wiki-leak China. Otherwise he'd be dead, not getting extradited to live in fame in a cushy prison.

You'd be surprised:

http://mirror.wikileaks.info/wiki/Category:China

(link not working FTM)

And just for the record, Assange didn't leak anything, users submit them or it was collected via Tor exit and wikileaks just publishes it.

> I guess the author's book must have some contents/examples regarding political issues; not Bayesian itself was the reason for the book being censored.

Does the reason really matter?

of course, the Bayesian theory is taught in schools; otherwise how could they send the spaceship to the outer space and have the fastest super computer in the world

Do these things require specifically Bayesian statistics?

From a strictly logical perspective, they might not. But I would wager a fair amount that there are Bayesian methods used somewhere in the stack supporting those two things. I also suspect that genetic algorithms designed for scheduling purposes probably help your phone call data get transferred somewhere along the way. That is to say, just because the two technologies are not obviously connected doesn't mean the former isn't used somewhere to achieve or improve the latter.

That said, even if there is no presence of Bayesianism in the efforts producing those tasks, his point was surely that, in today's world, an education sufficiently advanced to train people to do those things would certainly include Bayesian statistics somewhere in the curriculum.

Thank you garply. "an education sufficiently advanced to train people to do those things would certainly include Bayesian statistics somewhere in the curriculum" is exactly what I want to convey.

what do you think?

I think it's not immediately obvious why they would.

I agree, it's not immediate application. But Bayesian is one of the most important or fundamental knowledge to develop these cutting-edge technologies.

How? I'm experienced with spacecraft design and the like, and have never heard of Bayesian probability theory being used. If it was fundamental I'd expect some mention, somewhere.

If you look hard enough, you can find it in e.g. (linear) regressions and Kalman filtering.

lightoverhead,

I agree with the sentiment of your post - there is no reasonable way to censor a fundamental technical topic such as Bayesian statistics. I find myself wondering what the author did not post from his book that was questionable.

However, the tone of your post makes me wonder if you are a shill paid by the Chinese government to counter commentary critical of China on the Internet.

If I am mistaken, you have my sincerest apologies. The actions of your government have caused me to become a skeptic.

HackerNews,

Wikipedia: http://en.wikipedia.org/wiki/50_Cent_Party

Asia Times: http://www.atimes.com/atimes/China/JH14Ad01.html

In action, in relation to the underage Chinese gymnast in 2008 Olympics: http://strydehax.blogspot.com/2008/09/yang-yun-speaks-out.ht...

It's good Assange didn't wiki-leak China. Otherwise he'd be dead, not getting extradited to live in a cushy prison.

"I think that the next printing of our book should have "Banned in China" slapped on the cover. That should be good for sales, right?"

I would buy that in a heartbeat

Strange that it was banned. Maybe it had to do with the vignettes---some of the examples are from political science (e.g., election results, polling) and another deals with arsenic in drinking water...or maybe Hu Jintao feels the way to deal with hierarchal data is to cluster standard errors.

Is there any way to find out the specific reason the book was not allowed to be published?

Gelman is also a political scientist and many of the book's examples are about elections and election forecasting, probably not topics the Chinese government likes citizens reading about.

While the thought that there's something in this book that is being censored is an appealing story, it's also a less than likely explanation.

My guess is that the chinese publisher didn't want to pay royalties, and this was a convenient way to get a good quality copy of the book for printing without having to share profit.

I think you nailed it.

I've been thinking about it, and this may not be it. It sounds like he already has a relationship with this publisher for domestic publishing.

The author doesn't know, but he hypothesizes that it's because "the idea of prior distribution [is] contrary to Communism".

I'm pretty sure it has to do with the material of examples used in the book rather than with the actual theory.

Agreed. If it's Gelman many examples will be from elections, which in itself might be enough to give the communist party in china pause.

Pretty sure he was joking there?

It seems like a joke at first, but once you take into account China's previous actions it's actually quite likely.

Like the (brief) period when the sense of traffic lights was changed, so that "red" indicated "go". This makes perfect sense, since anything red must of course stand for making progress.

I was actually just trying to make a bayesian joke, but I think it was a little too subtle.

If you watch the stock market news in China on TV, a stock highlighted in red means it went up, and green means down.

That's an interesting sidenote.

I wonder if it's a subtle psychological trick to not associate red with negative connotations (such as "declining profits", "stop", "blood") because the color red is part of the CPC's identity makeup.Or maybe it's just a cultural thing, like how they wear white when someone died and black to celebrate a birth/life (I'm not sure if it is a chinese custom however, but I know it's an asian custom).

Which seems more than a little ridiculous to me, but ... it is China.

Deleted comment

I would seriously doubt this is the reason. If you want to learn applied Bayesian stats, you go to a machine learning or game theory conference or read research papers. There are plenty of Chinese citizens doing both, including low-level university students, so this really can't be the reason.

I would think that the reason has something to do with objectionable content inside this specific book, perhaps in an example problem as another commenter noted.

Bayesian stats have been around for a very long time, and it's senseless to try to block textbook-level access to it, unless you're really trying to block just the textbook itself.

I apparently double-posted my comment, and deleted the duplicate, your parent, before your post was committed. Sorry about that. I also modified the other version, which, in some sort of karmic reaction, is now dead. Let me

1. Repost the modified dead parent:

> Probably because they're using Bayesian filters on various security fronts: internal threat analysis, the great firewall, whatever side projects some bureaucrat is using to pad his income.

>edit: This is a problem with classification/censorship schemes: the censors/classifiers are all stove-piped, so their decisions come with a high degree of variability. The censor may well have been unaware that this is a whole field of work, or, is aware, but isn't responsible for the censoring decisions about all other related titles.

2. point to Feynman's accounts of censorship at Los Alamos as just how dense censors are: http://goo.gl/nfxqp

Ultimately, I think you're right, this doesn't make sense, except in some colloquial circumstance that is already water under the bridge for the actors.

On the other hand, nothing says that the China censorship mechanism has to be consistent, in fact I'd almost expect them to be arbitrary and capricious.

It might be that whatever individual decided to ban this book would also have wanted to ban other the things you mention if they had been aware of them.

If they try to ban Chinese citizens from international conferences and journals in anything related to Bayesian methods, I can tell you firsthand that a lot of math/CS/ML/AI conferences will shrivel up quite a bit. They won't die, but it'll be like the Cold War, where Russian mathematicians made advances that were insulated from the west for many years, and vice versa. A net loss for everyone, especially in a topic as fundamental as Bayesian methods.

You misunderstand, I'm very much not saying that there is or will be a general policy of banning things related to Bayesian methods. Rather, thats its quite possible one mid-level bureaucrat thought that Bayesian methods were unacceptable in this one instance. Maybe he (I think "he" is a safe assumption here) was confused and really thought that this sort of math was subversive. Maybe he was worried that he hadn't banned enough things this month and his supperiors would think he was lazy. Maybe the publisher failed to pay the customary bribe. In any of these cases, though, the official reason for the ban might very well be that "Bayesian Methods are subversive".

Seeing as this is a post on Bayesian statistics in China, I must ask: why are you sufficiently confident to assert that "he" is a safe assumption? Because it's political, because it's mathematical, because it's irrational, or because it's China?

Reviewing the evidence a bit more, it seems I shouldn't be. I imagined the gender ratio in the CCP was worse than what evidence I've been able to dig up suggests it is.

Ah, point taken.

Actually, one of the best ways to learn applied bayesian stats, used by most machine learning researchers I've asked about this, is to read Gelman's books on bayesian data analysis and hierarchical/multilevel modeling.

1. You can still buy this book in Chinese online book stores

http://www.51eng.com/cn/bookdetail.asp?id=602871

http://www.51eng.com/cn/bookdetail.asp?id=602870

2. As a Chinese myself, I can confirm Bayesian is not taught in high school statistics, but defintely taught in college statistics course. It's called 贝叶斯概率 in Chinese.

You can probably also download a photocopied PDF of it, as you can many English books, as well as American TV shows and movies. Actions in China is divided into the "official" and the "unofficial". Officially, only 20 foreign movies a year are allowed in. "Unofficially", you can choose from hundreds at your local DVD store, nowadays usually with 8 or 10 movies, or an entire season of TV episodes, on the one disk, going for less than US $1. Sometimes when they build a brick wall here, someone else will come back within a day to smash an ugly-looking hole in it. Officially, no-one's allowed through, but unofficially, you are.

Pfft, frequentists.

1.3 Billion can't be wrong.

I realize the "Death to the Shaw" spelled out by the first word of every seventeenth page was supposed to be a joke, but it is not very funny in light of the recent events of the last handful of days:

http://www.csmonitor.com/World/Middle-East/2011/0105/Suicide...

Though I viciously dislike of all forms of royalty, they are still human beings and they did not get a choice about being born into the nonsense.

  I realize the "Death to the Shaw" spelled out by the first 
  word of every seventeenth page was supposed to be a joke

According to the Amazon search facility, none of the 26 instances of the word "death" appear first on a page. So yeah, I think it's a joke.

I remember when I was taking an entry-level Economics course in China. We were using Paul Samuelson's classic textbook “Economics”. The English version mentioned China in some not-so-pretty examples. Apparently the translated Chinese version replaced “China” with a generic “One country”.

I didn't realize that things in China are that crazy...

"Bayesian Statistics are just a theory. Teachers should be required to state this to students and also to give equal time to alternate theories, such as the theory of Intelligent Predetermination."

Are things in China that crazy?

Unless you are forbidden to publish books on Evolution in the US, then yes, things in China are that crazy.

i don't mean that specific kind of crazy, but rather an overall suggestion that humans of all extractions have a habit of arguing with the facts for various reasons.

But yeah, since you ask, I believe that the kind of people who want creationism taught in public schools would ban books on evolution if they thought they could get away with it.

Sure, there are people crazy enough that they would ban books on evolution, just are there are undoubtedly people who are crazy enough to ban books on creationism. Those are probably not even near the top of the list of crazy things people wish they could do.

The reason that it is the case that "things are this crazy in China" is that there is a difference between the existence of people who would probably enjoy implementing crazy policies and the actual implementation of crazy policies.

Oh, we're in violent agreement. The thing about the USA and Canada is that we're constitutional democracies. That doesn't mean that as people we're any less crazy than places with a different system, but our constitutions act as a brake slowing down implementation of crazy policies and providing a kind of "undo" when things go wrong.

As we saw in the FLQ Crisis and again during the G20 summit, Canadians can lock thousands of people up without due process of law. As we saw in Kansas, certain Americans can ban the teaching of established science in the classroom.

But our system at least makes these things aberrations and provides dissenters a way to rectify the problems in the courts.'

edit: I think what I'm trying to say is "Our system provides us with an undo, but it's only good when we remain vigilant against craziness. If we become complacent and assume that we as people are somehow too intelligent or too morally centered to do crazy things, wham, we do crazy things too."

"But our system at least makes these things aberrations and provides dissenters a way to rectify the problems in the courts."

What a lot of people don't get when they try to tu quoque US/Canada/etc on particular events is that their protection of essential liberties is defense in depth.

If a book was banned:

- due process means that this incident can be taken to court

- freedom of press means the mass media can inform others of what has happened

- freedom of speech means you can raise awareness about the banning and denounce the agencies responsible

- with freedom of assembly, those who want the ban overturned can take to the streets and protest

These freedoms are mutually supportive -- abuse of any aspect would be mitigated by the presence of the rest.

V. true. Sadly, those who prefer a tyranny seem to have figured this out and are assaulting all of the freedoms you've listed. I leave it as an exercise for the reader to determine which of these liberties are currently under assault.

Trial by jury means that the citizens can nullify unjust laws.

And in the most extreme, the freedom to keep and bear arms means that we've got the ability to boot out an unjust government.

Every nation sees their own atrocities as "aberrations", while the atrocities perpetrated by other nations are seen as deeply revealing.

Oh you must mean crazy like the ban on federal staff reading WikiLeaks? :)

http://www.guardian.co.uk/world/2010/dec/10/us-ban-staff-wik...

I don't think it's crazy. As someone already said the objections are probably related to the examples mentioned on the book (election results and other "controversial" subjects). Crazy is superficial, this is, in my opinion, sad but not crazy or stupid.

Maybe it was a wrong word. I have to see the book to make conclusions, but I doubt that a book on Bayesian statistics may contain something highly controversial that needs to be banned.

And I concur! I just don't think it has to be dismissed as crazy, it's far worse.

Deleted comment

Yes, and the purpose of this post is to decipher that reason. From your certainty, it sounds like you might be in a position to offer it?

Reply on news.ycombinator.com