Review Crazynomicz

Minggu, 10 Juni 2012

Crazynomics

by Ian Ayres, Dian Basuki (Translator), Ainia (Editor)

Mungkinkah secara pribadi kita dapat MENERKA DENGAN TEPAT harga tiket pesawat minggu depan, ramalan cuaca bulan depan, peran guru tahun depan, atau kisaran jumlah konsumen produk Anda lima tahun mendatang? Atau dapat meramalkan solusi tepat bagi masalah kesehatan, keuangan, usaha, kemiskinan, kasus penyimpangan dana, bahkan masa depan Anda yang belum jelas? Mungkin saja! BERHIPOTESISLAH! pesan Crazynomics. Lewat uji coba acak terhadap data-data di masa lalu, kita dapat membuat prediksi yang luar biasa akurat mengenai peristiwa apapun di masa depan. Hubungan gaib antara hal-hal yang tampaknya tidak terkait ternyata dapat memprediksi perilaku manusia dengan hasil yang mengejutkan. Metode pengambilan keputusan berbasis data ini membantu Anda membuat sebuah keputusan yang lebih fokus. Temukan segera keajaiban data yang disajikan dalam bentuk cerita-cerita nyata unik di buku ini!! MASA DEPAN ANDA ADALAH SEKARANG!

B First 2010

Garrett: Be prepared to encounter the words "supercrunch" (used as any part of speech) and "nano-" (used indiscriminately as a prefix) approximately one billion times in a mere 272 pages. Dr. Ayres wants to write the next
Freakonomics
, and makes his professional association with Steven Levitt known frequently. What comes out is a repetitive book on applied mathematics fleshed out with anecdotes and descriptions of research. It's okay, but nothing groundbreaking.

According to Ayres, supercrunching involves running statistical analyses on large data sets. Supercrunching has better predictive power than experts do. Supercrunching is changing the way everyone and everything does anything. Supercrunching will pick you up if you're having car trouble. Supercrunching tops its pizza with ham and pineapple, just like you do. Supercrunching is attracted to you, respects that you're in a relationship, and wonders if you have a sister. Et cetera. Ayres, in his descriptions of supercrunching, vacillates from paranoid/creepy to smitten/exuberant.

Supercrunchers was fun at first. Then it got blah, seasoned with dabs of fun here and there. Luckily it is short enough that it didn't have time to grate. Of note, the book never mentions Grape-Nuts, not even once. Also of note, this review includes the word "supercrunch*" 999,999,889 fewer times than the book does. Supercrunch that, Ayres!

Ms: What a book! Shockingly good! I heard about this book while listening to The World is Flat by Thomas Freidman. Immediately, I made a mental note to find and read this book about the impact of computing power on everyday lives. Algorithms, formulas, yikes! (I have a bit of a math phobia.) Thank goodness this book breaks down complex ideas into understandable and applicable explanations. A wide variety of stories about how the computer is enabling huge changes in our schools, businesses, purchases, relationships and government. Inspirational and somewhat frightening, the book explains the high level of sophistication now practiced out 'there' in information gathering and "crunching" of the data. This new data is available to all levels of society, as long as they have access to the internet. In turn, this is leading us all towards new discoveries, more efficiency, greater understanding and of course, profits! The book was so great, I listened to it three times!

Trevor: The best of this one is his discussion of the 100,000 lives campaign (http://www.ihi.org/IHI/Programs/Campa...) which I didn’t really know about until Jim put me onto this book. A previous book I had read said that hospitals were trying to do something to improve their safety record in line with that of airlines, but the previous book didn’t mention this campaign as what was being proposed. I particularly like their slogan, ‘Some is not a number, soon is not a time’. The discussion of this campaign alone makes this book worthwhile.

I’ve been becoming interested in the overlapping fields of economics, human decision making, cognitive errors and statistics for quite some time now. One of the first books I read that started me off down this path was Freakonomics: A Rogue Economist Explores the Hidden Side of Everything. There were parts of that book that I found truly fascinating. I really like the idea of people looking for bizarre correlations between sets of data that tell us something new we did not know before. For example, Freakonomics showed that there was a relationship between abortion rates in one decade and dropping crime rates a couple of decades later. I’m a sucker for that sort of thing.

The writer of this one has done research with the guy that Freakonomics was written about. And this book fits neatly right into that school of thought. Let me try to sum it up for you. Humans aren’t really all that great at working out the best options available to them. We think we learn from experience and that we get better at picking what will be the right outcome when it comes along by our hard won intuitions (see Blink, for example) but when push comes to shove even around things we believe require substantial human discernment (like picking potentially great sports people) are actually generally better understood and the answer better stated by applying a simple mathematical algorithm. The most interesting example given of this is deciding if a vintage of wine will be exceptionally good or not. It has been assumed until recently that this isn’t something that can be decided until long after the grapes have been harvested, crushed, fermented and aged in oak. So, when someone said they could tell if a vintage was going to be good, not by tasting the wine, but by plugging in some values into the variables related to the climate while the grapes were growing in some formula – well, things were bound to get ugly.

Looking at it, this shouldn’t have caused as much outrage as seems to have been the case (at least, as it is reported here). I say that because wine is a natural product. And as a natural product it would seem its quality is ultimately dependant on the quality of its inputs (that is first and foremost, the grapes). If the summer has not been hot enough it is reasonable to think that the grapes may not become ripe enough, the amount of rain will probably also have an effect on the grapes.

Given that all of the things a wine maker can control probably already is controlled, the ‘acts of God’ that cannot really be controlled are obviously going to be the main variables left to decide whether a vintage is to be exceptional or ordinary. Like I said, if it is true (as is reported in this book) that some wine critics were outraged that someone could say they knew years ahead of time if a vintage was going to be exceptional, it seems very silly of them to do so.

So, what this book is saying is that the best decision are often made not by ‘experts’ (like wine experts) but by objective mathematical algorithms in which you plug in your data and wait for the answer to pop out the other side. Clean, clear, untouched by human hands – we are talking of the truth; unalloyed and pure.

My problem with this is that too many of these examples seemed to depend entirely on there being a right answer to the question at hand – and, in many of them, I wasn’t quite sure I understood what the ‘right’ answer was or how you could verify the answer the formula threw up was actually the right one.

I’ve just finished reading another book, for example, (Why We Make Mistakes: How We Look Without Seeing, Forget Things in Seconds, and Are All Pretty Sure We Are Way Above Average) in which people invariably preferred the bottle of wine labelled $90 over the one labelled as costing $15, even when they were the exact same wine. Having said that, although I can see the logic of knowing how ripe the grapes are when they are harvested and how much rain fell in the weeks leading up to harvest should make a difference to the quality of the wine once it is made– I also know that I’ve been fooled before by stories about subjects I know remarkably little about, particularly when those stories seem to make lots of perverse and twisted logical sense.

A case in point. Years ago I told my daughter (while she was still young enough to believe I was some kind of God) that it was likely that T-Rex didn’t go stomping around ripping other dinosaurs apart, but rather crawled around on his belly and ate mostly carrion. An article I had read somewhere explained this in fascinating detail. You see, T-Rex was cold blooded and big – so he was going to be at a disadvantage in having the energy needed to do the whole Jurassic Park thing. Also, there were those pathetic little forearms (think Charles Atlas just before the sand-in-face moment) which don’t seem to be of any use at all if he walked around upright, but might have made much more sense if he crawled about on all fours. Fi had a plastic T-Rex dinosaur and we used to have it crawling around on its belly as it dragged itself to dolls for a feed. As they say in Mythbusters, the guy who wrote the article had done the maths and the myth of the big, fast, angry T-Rex was busted.

Except, of course, that was before they worked out dinosaurs were less like lizards and more like birds. As a lizard, T-Rex wouldn’t have had the energy to run and fight and kill, but as a bird none of that was a problem.

My point is that the presentation of this stuff as simply objective truth is a bit dangerous – not least because we humans (particularly we boy humans) just love numbers, especially either big numbers or numbers with decimal points. Tell a boy it happened 165 million years ago or it will cost 15 trillion dollars to fix or that you are 43.7 per cent sure of something and, in what is the closest sensation to post-coital bliss, he will become so excited he won’t even know you made this stuff up on the spot. The most obvious reason why creationism is bound to fail is that any idea that stops men (well, boys) saying ‘9.8 billion years ago’ and replaces it with ‘around10,000 years ago and that’s tops’ really doesn’t stand a sporting chance.

Sorry, off topic. There are very interesting discussions in this book on data mining, data warehousing and corporate manipulation of customers that, even though the author did not feel were all that bad, ought to be read so that you have an idea of what is being done to squeeze money out of you. The discussion on how casinos work out your ‘pain point’ so they can stop you from losing beyond that pain point and thus get you to come back again later is disgusting. The most useful message to come from this book is that we should all be afraid when a corporation is being nice to us – it almost invariably means we are paying too much for the service they are providing us.

The most disgusting lesson presented here is that if you are black or female in America and you need a new car then you really ought to think about becoming white and male. And don’t be fooled if the dealer gets you someone to bargain with you of your own sex or race. Become white and become male – even if you have to pay someone both white and male a couple of hundred dollars to be you – it might save you thousands.

I worried while reading this book that a lot of the questions the author felt were settled by crunching numbers were still somewhat open for debate – I struggle with the idea that direct instruction (where teachers spend their days instructing kids literally by reading from a script with no discretion at all) can be the great boon to education that it is presented in this book to be. Figures were given that made direct instruction sound like the only sensible way to teach kids anything – however, even though I’m not a teacher, I still have strong doubts that this kind of rote learning (measurable as it is) will produce kids able to respond to the challenges of the modern world.

This book suffers from the same self-congratulation Freakonomics struggled with. I have a fairly natural aversion to people who pat themselves on the back quite so fulsomely. All the same, there were a number of things I learnt from this book and it was worth reading for its interestingly simply method of explaining regression and standard deviations.