A Lonely Fan of Chamberlain; Probability and Expected Value

There must be very few of us, and maybe I’m alone. I’m referring to fans of Neville, Chamberlain, the British Prime Minister who in 1938 gave up a goodly chunk of Czechoslovakia to Nazi Germany rather than risk war. It was the apex of Britain’s policy of appeasement. Today there are few, if any, worse things you can call someone in the foreign policy game than an “appeaser” or a “Chamberlain.” Few if any diplomatic “disasters” so ridiculed and mocked than Chamberlain’s 1938 decision at Munich. The decision at Munich for appeasement, now eighty-four years old, still haunts every Western foreign policy decision, like a malevolent spirit summoned forth from hell to argue for war.

Of course, in hindsight, Chamberlain made the wrong decision, but that does not mean his decision was wrong.  This apparent paradox has to do with probability and expected value.

Many people have blind spots when it comes to probability. For example, the political website “538” and its founder Nate Silver gave Hillary Clinton about a 70% chance of winning the 2016 election. When Trump won, 538 and Silver were skewered for being “wrong.” That take, itself wrong and innumerate, illustrates well a fundamental misunderstanding of probability (possibly also a displaced rage.)  Let’s imagine I hand you a gun with ten chambers, three of which are loaded at random. I ask you to aim at your right knee and squeeze the trigger. Of course, you’d wisely refuse, even if the 2016 American electorate chose to fire away. That’s what a 70/30 probability is.

The issue many people have with probability is that they know what happened and have a difficult time purposely dis-remembering what they know. They know that Trump won, so they assume that the correct predictive percentage of the election was 100% Trump, not 70% Clinton. They forget the cardinal rule that all past events (including the 2016 election before November 8th, 2016) were once in the unpredictable future. And although the past carries with it the aura of inevitability, the outcome of decisions, whether they involve countries–––elections and foreign policy––– or whether they involve individuals–––marriage, having children, or choosing a career–––are generally highly contingent and therefore difficult to predict.

In following an appropriate process to assess probable outcomes, we do our best to gather data, listen to advice, and construct a decision tree of outcomes. The decision tree is particularly important if one branch of the tree ends abruptly. Examples of foreshortened branches include declaring war, having a child, or squeezing the trigger of my imaginary, randomly loaded ten-chamber gun. As my young son once told a friend upset with a baseball card trade, “No backsies.”

For certain events, very limited in scope, we can use mathematical probability to come up with an expected value for our decisions. This is easy to illustrate with a 50/50 coin toss. If you place an even-odds bet on heads or tails, your bet will have an expected value of zero. But let’s say instead I offer you $2 for guessing correctly and still only ask you for $1 if you guess wrong. Your expected value is  positive $0.50, calculated as (plus 50% of $2) minus (50% of $1.)

Now I make the same lopsided offer, but instead of $1, I ask you to bet a sum equal to the value of every last earthly possession you own. The expected value is still positive for you, but a loss would be devastating, so it’s an unacceptable risk. No one should make that bet. The math fails to capture the real world upside and downside, the real world risk and reward.

The point is that except for a small and generally irrelevant set of examples, math fails to express accurately the risks and rewards of decisions. The probabilities of outcomes are too murky, and the effects of the outcome are even murkier. Consider the past few years of living with Covid and all the risk and reward decisions we’ve had to make based on highly imperfect information and advice. ¹

Back around to Chamberlain’s appeasement decision in 1938 when all the horrors of the Second World War were in the unknowable future. In the knowable past were Hitler’s reestablishment of Germany’s armed forces and armed occupation of the German frontier with France, both violations of the Versailles Treaty. The only country Germany could be said to have violated was Austria, which became part of Germany in March 1938. But that happened without any armed conflict and with the approval of the Austrian people.

So, when faced with Hitler’s threats to go to war unless Germany was allowed to annex the ethnically German part of Czechoslovakia, Chamberlain’s decision tree was either to risk war or take a chance at peace. His foreign minister Lord Halifax, in agreement with Chamberlain, was against risking war and expressed it as a choice between “a certain war now, against the possibility of war, perhaps in more unfavorable circumstances later.” ²

Given the horrors of the First World War and subsequent developments in the lethality  of weapons, Chamberlain assumed that a Second World War would be at least equally as horrible, if not more so. He and everyone else also assumed that a “rematch” between Germany against the Britain and French allies would be a long and close fought war. On paper, it was Britain and France that had superiority of numbers in most aspects. No one expected France to be conquered in six weeks. As well, the Holocaust was generally unimaginable in October 1938.

I’d have taken the same gamble as Chamberlain did in 1938. A certain and terrible war would have appeared worse to me than a possibility of peace and, if peace could not be made, a terrible war anyway.

Right decision, wrong outcome.

 

1. For a probabilistic take on Covid, here is a link to a Twitter thread I found compelling in the author’s thought process. The last tweet of the thread is included.

Bob Wachter @Bob_Wachter

And let's be kind to each other as we all grapple with these tough & very personal choices. To me, it's worth erring on the side of safety. Others will find that the joy of "normal" is worth accepting a modest risk. Neither choice is wrong. (25/end)

2. Whether war would have been more or less “favorable” to Britain and France in 1938 or 1939 is a subject historians debate. Some cite the addition of Czech forces as a major factor favoring 1938, although the British and French assumed the Czechs would have been quickly overwhelmed by Germany given geography and relative forces. On the other hand, British aircraft production accelerated greatly into 1939 and 1940 and war starting a year later may have made the difference in the Battle of Britain. The answer is unknowable.