Monday, November 7, 2011

End of year predictions

The end of the year is approaching, so the annual flurry of predictions must be right around the corner.  Or maybe that smell is just a septic pumping truck, the contents are similar, except there are regulations covering the disposal of septic waste.

Here are my predictions:

People will predict stuff, and for the most part only their successes will be remembered.

Some people will predict the same things they have been predicting for years (or maybe even decades), and if they are eventually “right”, no one will ask about all the times they were wrong, and even of they did it would be shrugged off as “I was right, just off on timing”.

2012 will not be the year of Linux on the desktop.

And because I feel compelled to make one real prediction, Windows 8 as a desktop OS will be as disappointing as Windows 7 has been successful.

No matter what is predicted or what actually happens, randomness will not get the credit it deserves as people look both forward and backwards in time. Admitting that “life is a crap shoot” doesn’t get you the respect it should.

Dice, random or predictable?

I’ve listened to a couple of interesting books in the past several months, and a recent episode of the Freakonomics podcast does a great job of summarizing a lot of ideas into a one-hour show.  Short version: random stuff happens, and that makes prediction hard.  Really hard.  Also: so called “experts” are usually wrong- and the more adamant and certain an “expert” is, the more likely they are to be wrong.

The Freakonomics “Folly of Prediction” episode does a great job of distilling a lot of research into an easily digestible audio format.  (Note: If you aren’t familiar with Freakonomics, you should be- they make economics entertaining, challenging, and informative.  I’ve read both books and am a regular listener to the podcast.  Unrelated to this post, the recent episode on quitting was another great one).  Some of what they bring up in the predictions episode of  Freakonomics podcast is covered in much greater detail elsewhere, including a couple of books I listened to earlier this year.  The predictions podcast briefly discusses prediction markets, which seem much more promising than traditional pundit-centric pontification style prediction.

Note: I listened to both as audiobooks, Audible is not perfect, but for the commuter and frequent traveler they are great.  (I’ve also heard audiobooks are great for people who “exercise”, but people who do things like that clearly have too much to live for and are just punishing themselves for it).

The first book I listened to was The Drunkard’s Walk by Leonard Mlodinow.  Here’s an excerpt from Stephen Hawking's Amazon Review of The Drunkard's Walk:

In The Drunkard’s Walk Leonard Mlodinow provides readers with a wonderfully readable guide to how the mathematical laws of randomness affect our lives. With insight he shows how the hallmarks of chance are apparent in the course of events all around us.

The Drunkard’s Walk covers a variety of probability topics, from the significance of randomness to some history of the study of probability, and uses many illustrative anecdotes (including a look a the Monty Hall problem and others where “common sense” appears to let us down).

The second book was Future Babble by Dan Gardner.  From the author’s site:

Future Babble, a critical look at expert predictions and the psychology that explains why people believe them even though they consistently fail.

Future Babble is focused on prediction, but as random events and probabilities are challenges to prediction this book does have some content which overlaps with The Drunkard’s Walk.

Both books are overly negative at times, and thoroughly dismissive of many “experts”, but together they make a compelling case for a healthy dose of skepticism.  These works do highlight issues of bias and fallacies which lead us into making or accepting seemingly “logical” but wrong predictions, being aware of these biases and fallacies can help us identify and avoid them.

One of the recurring lessons of all of these works is that the more confident and adamant someone is about their predictions, the less likely they are to be correct, and the more likely they are to deny when they have been proven wrong.  A lot of this goes back to Philip Tetlock’s works including Expert Political Judgment, a skewering of political pundits’ ability to predict much of anything.  Tetlock often speaks of “hedgehogs and foxes”, a reference to the phrase:

Four-toed Hedgehog, Atelerix albiventris, 3 weeks old, in front of white background

The fox knows many things, but the hedgehog knows one big thing

Red fox (4 years)- Vulpes vulpes

from the ancient Greek poet Archilochus.  The hedgehogs are those with an ideology or single big idea, they hold onto the idea and rationalize around it.  Hedgehogs tend to use absolute words and are very confident in their predictions- hide from these people (television, especially cable news and talk radio are full of them).  Foxes, by comparison see much more variability in the world and are prone to use what we often derisively call “weasel words” such as “probably” or “likely”.  Foxes are also much more likely to admit they were wrong when history proves their predictions in error.

I am not saying that nothing can be predicted, and I’m not tossing stones at my risk and metrics friends- I am just suggesting that we pay attention to the realities of the world.  And the reality is that random events happen and have a large impact on our lives, and that some things which appear random are not.  And that means predictions are often hard, if not impossible.

I’ll leave you with a final quote, this one from the great philosopher Yogi Berra:

“It’s tough to make predictions, especially about the future.”