Playing It Smart: Can you learn more from the future than from the past?

The philosopher, George Santayana, famously said that those who cannot learn from the past are doomed to repeat it. More mundanely put, experience is the best teacher. Is it true?

In "Fooled by Randomness" (Random House), Nassim Nicholas Taleb writes that history only seems deterministic -- effect following from cause -- when viewed backwards in time. Instead, he alleges, events unfold under random influences. And, randomness implies that differing histories could have transpired with varying degrees of probability. Further, nothing inherently distinguishes what did turn out from what might have, except in retrospect.

For instance, is there really a lesson in Napoleon's ill-fated Russian campaign? Different weather might have lead to the success he expected. What may seem destined when traced backwards was hardly preordained at the outset or as circumstances evolved.

Another problem with learning from the past arises when the record is that of others. People who badly want some result tend to believe dire consequences that befell the next guy can't and won't happen to them. Like solid citizens who chase casino losses all the way up through the limits on their credit cards.

Then there are folks who've heard of gamblers snatching ecstasy from the jaws of agony by pressing their bets after every loss, so figure the strategy's good. It may work. It may not. Triumph or tragedy may be the end twig on another of many randomly extending alternate branch histories that could have grown.

Here's a simple illustration. What would be learned from the history of a person who accepted and won a $10 million challenge to play Russian Roulette with a six-shooter? If it was your own experience, would winning suggest you try it again for another quick $10 million? The same could be asked for losing. But if it was your own experience, you wouldn't be doing the asking.

If learning from the past is suspect, what about learning from the future? Preposterous? Maybe not.

Pretend you build a Russian Roulette simulator by putting six small squares of paper into a hat. Five are green, one red. Close your eyes and draw a square. Green you're rich, red you're dead.

Most simulators are fancier and address more complex situations. But the principles are the same. Gambling simulations are usually used to find the probabilities in games for which mathematical equations are too difficult to write or solve. The approach for Russian Roulette might be to draw from the hat, jot down the color, and replace the paper. Repeat this six million times then tally the outcomes. You'll likely find about five million greens and a million reds, telling you that the odds are 5-to-1 or the probabilities are five out of six you'll win. You could similarly get the probability of playing up to some given number of times without sending a slug into your skull, or the average number of draws before the big bang. The equations for Russian Roulette are so straightforward that you could get these answers without a simulator. But this would be the preferred approach, say, to analyze the chance of a straight taking a pot in seven-card stud.

You can also use simulation to create diverse future paths of events involving random factors. For Russian Roulette, these would show possible outcomes that collectively reveal far more than the anecdotal evidence of a few players dying or winning lots of dough, or of a statistic giving the average number of successes in a row. For example, you could map the fates of a million players with revolvers having different properties.

Simulation lets you make inferences from possible futures. But it doesn't let you predict which will happen. That's the point. Randomness create trees of alternatives, which can be explored for guidance in making decisions conceivably better than studying selected or even agglomerated results from the past.

The poet, Sumner A Ingmark, noted this about the notion:

What may look apparent from the viewpoint of history,
May seem quite aberrant when the future's still mystery.