Sunday, December 29, 2013

Three Books

Another day, another book report, but first: It's a rainy day here, but I got up full of pep anyway, resolved to do, among other things some laundry, some yoga and weights, and some dusting & vacuuming -- the vacuum, busted for a few weeks, just got fixed and the general dust/dirt is enough that even I noticed (that's bad).

Anyway... jumped out of bed, tossed in some laundry and got the coffee going, and then Anne, washing some spinach for breakfast, noticed that the water coming out of the tap was brown. Brown! Blech. She called the water department, and they said there was a water main break that they were trying to locate & fix; in the meantime we shouldn't drink or cook with the tap water (not that we would have...). OK, throw out the coffee and tea, start over with some stored water from the basement, and get on with our day, resolving to redo the laundry when the water's fixed.

Some recent reads, in the order I read them:

The Twilight of the Elites, by Chris Hayes
The Victory Lab, by Sasha Issenberg
The Signal and the Noise, by Nate Silver

The first two were sort of recommended by various political blogs; I knew that the likewise blog-celebrated Nate Silver had a book out and bought it without any additional recommendation when I saw it (at Penn Books, during our last foray into Philly).

Long story short: the first was meh, the second maybe even meh-minus, but The Signal and the Noise was (or is -- I'm not finished yet) awesome.

I got Twilight of the Elites (subtitle: "America After Meritocracy") maybe a year ago, expecting to read about the problems caused by our current power structure and how it maintains itself. That's pretty much what the book delivered: problems such as baseball scandals and the housing bubble are analyzed in terms of what happened, who screwed up and why, and who did or didn't get held accountable; the main premise is that our "meritocracy" -- or as I'd rather put it, our "so-called meritocracy" since some of the supposedly meritorious things are really just class markers or the result of unexamined privilege -- leads to an elite that is overly competitive (too ambitious, too envious of the next level of success), prone to cheat (since incentives are based on measurable "performance," which can be gamed), and convinced that they are truly the elite since they won their positions based on "performance." He further points out that existing power structures use their positions to work the system in favor of their own members (think private day-care undermining the level playing field of education), which I guess gets back to my point about privilege.

So far I'm with author (though the problem as he puts it looks more like an example of what's wrong with Management by Objectives than anything else), but somewhere in there he seemed to lose focus, and the second half of the book really lost my interest; I had to push to finish it. Too bad really, because the end of the book had some prescriptions for fixing things, including the idea that more equality would help a meritocracy work as it should -- though my own feeling was that meritocracy itself is fundamentally flawed.

Anyway, I finished the book, which is more than I can say for The Victory Lab. This was the story of how the Obama campaign used new concepts from Big Data (automated polls and constant monitoring of the electorate's mood, microtargeting, etc) and an emphasis on evidence in campaign decisions. All well and good, but my expectations was that it would be about the data and tools, and it turned out to be about the people involved, which consultant came up with what insight that led to the idea of microtargeting in some other campaign etc, with only the vaguest idea of what the actual methods were -- just a few Time Magazine-level examples and oversimplifications, like "algorithms" being defined, essentially, as weighted averages. (Poor old Time Magazine used to hit a nerve all the time with me whenever they had an article touching on something I knew anything about: their descriptions and explanations were so plausibly written, but were oversimplified to the point that their meaning was the opposite of whatever the case really was in the thing they were explaining. Drove me nuts.)

The author also had a maddening habit of introducing new actors into the story, then going through long digressive backstories on the new guy's history and CV. The constant jumping back and forth broke up what should have been the straightforward flow of the narrative, and it also eventually became hard to keep track of all the characters involved, both of which made a fuzzy story even fuzzier -- once again, the author couldn't maintain focus.

I got most of the way through before I started skimming, and finally just managed to "finish" the book by plowing through the last few pages in a row. I guess The Victory Lab would make a good read if you're interested in the Obama campaigns themselves and especially the personalities involved, but that's not my cup of tea, and it wasn't what I was looking for in this book.

Luckily I saved the best for last: I picked up The Signal and the Noise on a whim, and before I was done with the introduction I was hooked. Nate Silver is a very smart, personable guy, with a good grasp of the concepts he's writing about, so it's not a surprise that he could write clearly on his subject -- if poor writing is a sign of poor thinking, the converse is probably also true -- but even above that, he's an accomplished and engaging writer.

The book itself is about predictions and forecasts, why so many go wrong and why some don't (to paraphrase the book jacket), and he breaks the book down into two main parts, the first being a description of the problem (bad predictions) over several chapters, and the second part looking at a way to improve the situation (Bayesian statistics). The problem chapters are case studies of political punditry, baseball -- two subjects in which the author really made a name for himself -- weather forecasting and earthquake forecasting, and so on, all totally fascinating. The start of the Bayesian part is where I am now, and it looks to be at least as good.

I can say that I see the appeal of the first two books, and I can see them being enjoyed by the right personality, but The Signal and the Noise is the one that gets my wholehearted recommendation. Take my advice and read it.

(By the way, I finished Bleeding Edge, and I have to say I'm glad I did.)

No comments: