Pincher Martin's Bibliothèque

Cal,

McWhorter doesn't believe the Celts were killed off in a genocide. He plays off that straw man thesis, however, in writing the first chapter of his book about the "Welshness" of English grammar.

Traditionally, what we would today call a genocide was the preferred scholarly explanation for why there were so few Celtic-speakers found in just a few corners of the the main island of Britain in modern times, and for why speakers of a Germanic tongue (English) dominated Britain. The assumption was that the German invaders killed most of the Celts off when they began invading in the fifth or sixth century, and marginalized those they did not kill.

There was some tenuous historical support for this thesis. One of the first historians of England, Venerable Bede, described the Germanic tribes as overrunning Britain. And, linguistically, there didn't seem to be much of a Celtic influence on English, which suggests the Celts weren't around long enough in large enough numbers to have an impact on the invading tongue.

This traditional thesis of the settling of Britain was once believed by pretty much anyone who mattered. But, recently, the work in mapping DNA has overturned it. In his book, McWhorter argues that the Celts had more of an influence on English than we believe. That section is the one part of the book that PE respects as being solidly argued.

I think it's a very modern coinage for that context. Every invasion and displacement of what is today some marginal ethnic group is a genocide now.

When did it finally become common to speak of the genocide of Native Americans in US history, for example? The 1960s? Maybe not even that early? When I was a kid, my history courses in the 1970s never called it a genocide. I don't think I recall seeing the term applied in US history until the mid-80s. Even then it was still sensationalist to speak that way. It's much less so now.

I got into an argument a few months ago with someone who decried the human rights violations of Genghis Khan.

Perhaps it's not profound, but Silver's book describes and applies statistical concepts in a way I had never seen before. There's a lot of value - at least to me - in having someone show the many practical applications of Bayesian statistics, and then follow up by carefully explaining in clear prose why, say, forecasting in meteorology has improved over the last three decades but forecasting in economics or predicting earthquakes has not.

Clearly that was part of it. But there was a lot more.

I particularly liked Silver's chapter on "The Poker Bubble". Because I play a lot of poker, it helped make me more comfortable with some of his points about changing probabilities over time.

Everyone knows the best starting hand in Texas Hold'em is A-A. Any good player is willing to bet his entire stack of chips on that hand before any other cards are dealt. Because he knows that, heads-up, he is at least a four-to-one favorite over an opponent unless that player also has A-A.

But what if the flop comes and the board shows a pair of face cards (either kings, queens, or jacks)? Or you see three cards all of the same suit and you don't have an ace in that suit? If you put a big bet out and your opponent calls you, I would say you shouldn't feel as confident as you did before the flop.

But were you wrong to feel confident about your two aces before the flop? Obviously not. It's the best hand in poker. If you can't bet big with two aces, you shouldn't play poker.

One of Silver's most valuable points in the book is that the probabilities change depending on the circumstances, and that there is no shame in changing your forecast to take those new circumstances into account. Changing your mind doesn't devalue your earlier forecast. You can be both right to go all-in with a pair of aces before the flop and, if you don't go all-in, right to slow down your betting with that same pair of aces if the board and your opponent give you good reason to.

This seems such an elementary point, but it's one I see people (and poker players) neglecting all the time. Most amateurs feel it is the wise and honest policy to make a forecast and then stick to it. Silver quite rightly says the wise and courageous thing to do would be to change your forecast if you have new evidence which shows it's no longer likely to be accurate.

Putting this lesson in poker terms illustrated it for me in a way that a mere write-up would not have. For every good poker player knows that the ability to throw away a good hand when you know you're most likely beat is what separates the great players from the amateurs. In poker terms, some players fall in love with their good hands and can't let them go even when the odds have moved against them. Some people fall in love with their forecasts.

*****

I also liked Silver's chapter on financial bubbles and the efficient market. He showed a degree of sophistication in his modest claims about we can and can't predict that I don't see in many finance books.

Here's a small grab bag of other valuable points I found in the book.

• It's important to have the right baseline for judging the accuracy of a forecast. In meteorology, for example, how do we know for sure that weather forecasting has gotten better over the last four decades? Because the baseline for weather forecasting is the historical record. If a meteorologist's forecasts can't beat the historical record for any particular day, then they are of little use. If, on the other hand, his forecasts are more accurate than the historical record, then they have value. Over the last few decades, weathermen have extended their mastery over the historical record from three days to one week.

• The inability to make specific short-term predictions about a subject does not prevent you from making useful long-term predictions. Take earthquakes. We can be fairly confident that San Francisco will have at least one major earthquake (6.0 or above) sometime in the next fifty years. I can't tell you if it will happen next week or next decade, but if I were to live long enough, I would feel confident betting someone that the event will happen eventually. Why is it useful to know that San Francisco will have a major earthquake and Rio De Janeiro will not? Because disaster preparation and building codes can be adjusted accordingly.*

• Betting is a useful and indeed necessary activity for checking your forecasts. People who aren't willing to wager on their forecast aren't likely being honest with themselves. As Silver says, "The most practical definition of a Bayesian prior might simply be the odds at which you are willing to place a bet."

* I found this useful in thinking about the growing Hispanic vote. I don't know exactly when it will begin to provide Democrats with more than a small advantage at the polls, any more than a seismologist knows exactly when a major earthquake will next hit the Bay Area. I just know that if it's not checked, it eventually will.

The analogy isn't perfect, but I think it's still useful. You're right, of course, that voters are freer to change course than are the continental plates. Neither of us should need Nate Silver to understand that obvious point. But it's remarkable how steady many of these group voting patterns are over the long term when you study them, and over the very long term it may be the changing political environment that alters these patterns more than it is the voters themselves changing their voting preferences.

Look at U.S. Catholics as an example. They were a reliable Democratic constituency for over a century. Hard data is difficult to come by before the Second World War, but it appears a pretty safe bet to say that a consistently large majority of Catholics identified as Democrats through the 1960s. Richard Nixon's reelection campaign in 1972 was the first time a majority of Catholics voted for a Republican presidential nominee (although the universally-loved Ike might have come close in 1956).

From that point on, Catholics became a genuine swing vote. According to exit polls, a majority of Catholic voters voted for Reagan in 1980 and 1984, and for George W. Bush in 2004. And even when a majority of Catholics vote for the Democratic presidential candidate, it is almost always by a much thinner margin than was the case before the 1960s. Only once in the most recent eleven elections (Clinton in 1996) have Catholic voters given the Democratic presidential candidate a greater than ten-point margin in their vote.

So did Catholics become more conservative or assimilate to white protestant norms over the last century? There might have been some of that. But the evidence suggests that more Catholics began looking at the GOP when they did mostly because they were turned off by the socially liberal policies of the Democratic Party that came out of the 1960s. Catholics still preferred the kind of economic paternalism that characterized the New Deal and the Great Society, but they couldn't bring themselves to vote for the homo-loving, abortion-supporting, drug-providing George McGovern.

This kind of socially conservative/ economically liberal attitude had been evident in many Catholics dating all the way back to Dorothy Day, a Catholic who flirted with radical and socialist ideas, and loved the redistributionist policies of her progressive allies in the 1920s, but also hated their bohemian, sexually-open lifestyle. Until the 1960s, however, that would not have been a major source of political tension for U.S. Catholics. Both political parties were traditionalist on most social and moral issues. So until the 1960s, American Catholics who voted Democrat could have their redistributionist cake and eat their traditionalist pie, too. After the social maelstrom of the 1960s, it was not as easy for them. But had Catholics changed? Or had the times and the Democratic Party changed?

I'm sure you can now see where I'm going with this. Most Republicans today, when surveying the odds against them with minority voters, exaggerate the level of their personal agency in attracting votes. They talk grandly of winning more Hispanic votes, as if it's just a matter of Republicans putting up a few more Spanish-language ads or changing a couple of sideline policies and then they're back in the ballgame.

But what if most voters, minority or not, tend to vote the way they do because they feel genuinely comfortable voting that way? I got into this argument with Professor Mead over at his website Via Meadia last year. He loves to go on and on in the manner of an evangelist warning of Judgement Day about the decay and imminent demise of the Blue State model. He's also a big fan of immigration and routinely got smacked around by his commentators for it until he decided to remove the comment section of his blog.

Last year Mead brought up a poll that showed Jewish support for Obama was still strong in the run-up to the election, and then predictably segued into a liberal paean about tolerance that rather oddly focused on how anti-Semites misconceived of a worldwide Jewish conspiracy.

As you can see in the comment section, I thought Mead's focus misplaced and tried to redirect the discussion back to something I thought both he and I could agree was more important:

Mead briefly responded: "@Pincher Martin: why difficult? There seem to be a good many historical and cultural reasons for this, none of which reflect particularly poorly on Jews."

You can see in my response here why I thought Mead of all people should find it a difficult question to answer.

I doubt it. Blacks were shrinking as a percentage of the U.S. population during that period, and in any case Jim Crow laws made their vote in the south, where they were most numerous, insignificant.

Absolutely. But the significance of the analogy is in understanding that just because we don't know what effect Hispanics will have on 2016's election doesn't mean we can't understand what effect Hispanics will generally have on elections in the future (i.e., more votes for Democrats). I can't tell you when or how big the next earthquake to hit California will be, but I can tell you there will be one.

At some distant point in the future, the issues and social context might change so abruptly that it all becomes moot, much as the social landscape of the 1960s changed the nature of the Catholic vote. But conservatives can't wait for that happen. For one thing, we can't be certain it will in our lifetimes. For another, we can't be certain the change will favor conservatism. For a third, even now Catholics are more likely to vote Democratic than not, which suggests any change that comes in the Hispanic vote will be marginal, not categorical.

Hispanics are already costing Republicans elections in some states (like California). The question is whether such local phenomena will scale up to the national level, and when it will happen.

Hahahaha ! You keep better track of these special days than I do. I had no idea tomorrow was Tomb-Sweeping Day.

Read any good books lately?

No, I haven't read any book by Atkinson for quite a while - not since his book on the Persian Gulf War. The first Persian Gulf War.

I've been reading a collection of essays by J.B.S. Haldane.

Here's a videotaped reminiscence of the man by his student John Maynard Smith about a hilarious anecdote involving Haldane's run-down motor car, a fire, and the "method of Pantagruel".

It's one of the most hilarious stories I've heard in some time.

If you enjoyed the book on hyperpolyglots I reviewed last January, you might find this article interesting about a Jewish kid in New York who's learned 23 languages.

Here's a video of him speaking, or at least practicing, in twenty different languages.

Youtube video of Tim Doner

I can't speak to the quality of his language learning in most of the languages on display in the video, but his Mandarin Chinese is surprisingly good (the section where he speaks Mandarins begins at 8:30 in the clip). He has a good ear for the tones and a lot of confidence when speaking them.

I'm currently reading a book called Hard Line: The Republican Party and U.S. Foreign Policy since World War II. The book closely examines the foreign policy views of all the Republican presidents in that period, and two Republicans (Robert Taft and Barry Goldwater) who lost presidential elections but were more influential in the party than most losing presidential candidates.

While the book is almost exclusively about foreign policy, in the chapter on Richard Nixon and Henry Kissinger, the author does briefly address the liberal domestic policies of Nixon by repeating a quote from then-Nixon advisor, Daniel Patrick Moynihan, who borrowed a line from Disraeli when he said of the administration that it was a case of "Tory men and Whig measures."