Luck?

[SDragon]

In another thread, I had mentioned the idea of luck as a game mechanic, which evolved into a discussion of luck in real life-- particularly, whether or not exists. I said I think it does, while Brainface and Snakefing say it doesn't. With permission, I'm copying Snakefing's and my portion of that discussion over to this thread.

[in response to one of Brainface's comments; if I get permission, I'll copy his portion over here as well, but I'm personally focusing
 There are people who seem to continually defy the odds, sometimes to very unbelievable levels. Just look in the Guinness Book of World Records for some examples. Unless you can think of some other explanation for Roy Sullivan.

<rant>
Is there some reason to think that there are more Roy Sullivan-types than statistics would predict?

People get struck by lightning and survive all the time. Seven times, that is certainly unusual. And surviving that many is certainly "lucky" from Roy's point of view. But for this to provide evidence that Roy has some special attribute, you'd have to compare Roy's experience to the number of people who've been struck once or twice and survived vs. died. Then take into account the fact that Roy's profession as a forest ranger puts him at higher risk, and the number of forest rangers out there over the years, etc. You might find that it is actually reasonably likely that there would be someone out there who had been hit many times and survived. And it was Roy's fortune, for good or ill, to be that person.

Even if you did find that such a thing was quite unlikely, you'd still be left with at least three possible explanations:

1. Roy is just a statistical fluke - out of the ordinary run of events but not for any particular reason. Things like that happen from time to time, in one area or another.

2. Roy has some kind of characteristic that attracts lightning but at the same time protects him from the effects. No idea what that might be or how it would work.

3. Roy has some kind of non-material characteristic (call it "luck") that causes odd events to occur. No real idea how it operates.

Even if option 1 seems statistically unlikely, it could be that it is still the best explanation (in absence of further information). Since we have no idea how options 2 or 3 might work, they seem inherently implausible and that counts against them just as the statistics count against the first option. The statistics only rule out option 1 if there is a more likely alternative explanation.
</rant>

We now return you to your regularly scheduled programming...

"[W]hen you have eliminated the impossible, whatever remains, however improbable, must be the truth[.]" --Sherlock Holmes


Okay, so, first off, I don't think Roy Sullivan-types are, as Snakefing puts it, more common then statistics would predict. I think this is just a matter of how wide or narrow your perspective is on the issue, and how specific you get for the outcome. Certainly, the odds a a human surviving seven lightning strikes are very high. They would roughly double if the person has to be specifically male, and they jump even higher when his name has to be "Roy".

All I have left to say at the moment is in direct response to your first possibility. What, exactly, is luck, if not a statistical fluke?

[Lmns Crn]

What, exactly, is luck, if not a statistical fluke?

Did we talk about the important distinction?[/b]

That is to say, the difference between saying that something is "lucky" as an event that's rare and fortunate (say, you won the lottery), and a person or thing that's "lucky" (say, a person who mysteriously has increased likelihood of winning the lottery).

[snakefing]

There's no doubt that Roy was "lucky" to have survived seven lightning strikes, or "unlucky" to have been hit that often. The question is: Does this say anything about Roy specifically, or was he just the victim (or beneficiary, depending how you look at it) of happenstance?

As LC says, you can look retrospectively at something and say that it was an unusual and fortunate ("lucky") event.

This kind of luck is purely retrospective - you can't predict it and you can't use luck in the past to predict luck in the future. This is what I would say is a "statistical fluke." Something that is unusual, can't be predicted in advance, and can't be used to predict the future.

But if you want to think of luck as a thing or a feature that is inherent to a specific thing - that's a different thing. It's hard to talk about it because the concept itself is quite slippery. Suppose we accept that Roy Sullivan was inherently "lucky" in some significant way. What does that mean? More likely to be hit by lightning in the future? More likely to survive? More likely to survive a car accident? More likely to win the lotto?

It's this latter type of luck that is usually captured by RPG luck mechanics. A guy who rolls a naturally 20 just when needed is lucky in the retrospective sense. A guy whose "Luck" characteristic allows him an extra chance to try and get that natural 20 is lucky in the second, inherent sense. He's more likely to have fortunate events than the guy whose Luck is lower.

----
Hang on St. Christopher on the passenger side - Tom Waits

[SDragon]

What, exactly, is luck, if not a statistical fluke?

I'm actually dismissing the plausibility, outside of fantasy settings, of luck being anything but a statistical fluke.

That is to say, the difference between saying that something is "lucky" as an event that's rare and fortunate (say, you won the lottery), and a person or thing that's "lucky" (say, a person who mysteriously has increased likelihood of winning the lottery).

The only way to demonstrate that difference, of course, if for the hypothetical "lucky" person to repeatedly have rare and fortunate cirumstances. Surviving a lightning strike is rare and fortunate, but it happens. By the seventh lightning strike, you've developed a trend of rare, fortunate events. By then, I think it would be fair to say the person is lucky. Of course, "surviving a lightning strike" can be replaced with any rare, fortunate events, such as winning the lottery.

[Polycarp]

The only way to demonstrate that difference, of course, if for the hypothetical "lucky" person to repeatedly have rare and fortunate cirumstances. Surviving a lightning strike is rare and fortunate, but it happens. By the seventh lightning strike, you've developed a trend of rare, fortunate events. By then, I think it would be fair to say the person is lucky. Of course, "surviving a lightning strike" can be replaced with any rare, fortunate events, such as winning the lottery.

Remember the "gambler's fallacy" here.  The chance of being hit by lightning once is the exact same as being hit by lightning a second time; the weather does not "remember" who it hit last time.  It's like assuming you have a high probability of getting tails on a coin toss after 9 consecutive heads - the chance of every coin toss is 50/50 regardless of previous coin tosses.

If someone wins the lottery twice in a row, you would call them lucky, but the chance of winning the lottery the second time is no greater or lesser than winning it the first time (assuming the number of people buying tickets remains equal).  It is statistically just as likely for John to win the Lottery twice as for John to win the lottery once and Bill to win the lottery once, yet we don't call that "luck" because of observer bias.  Every outcome is statistically equal but the outcome of John winning twice seems especially fortunate to us, so we call it "lucky."  This is a result not of the statistical improbability of the lottery, but a result of the impressively good result of getting $200 million (or whatever).

[snakefing]

The only way to demonstrate that difference, of course, if for the hypothetical "lucky" person to repeatedly have rare and fortunate cirumstances. Surviving a lightning strike is rare and fortunate, but it happens. By the seventh lightning strike, you've developed a trend of rare, fortunate events. By then, I think it would be fair to say the person is lucky. Of course, "surviving a lightning strike" can be replaced with any rare, fortunate events, such as winning the lottery.

Slight, but important difference on my part. I think it would be fair to say the person was lucky. Whether that will carry into the future is really where the question lies.

[SDragon]

The only way to demonstrate that difference, of course, if for the hypothetical "lucky" person to repeatedly have rare and fortunate cirumstances. Surviving a lightning strike is rare and fortunate, but it happens. By the seventh lightning strike, you've developed a trend of rare, fortunate events. By then, I think it would be fair to say the person is lucky. Of course, "surviving a lightning strike" can be replaced with any rare, fortunate events, such as winning the lottery.

Remember the "gambler's fallacy" here.  The chance of being hit by lightning once is the exact same as being hit by lightning a second time; the weather does not "remember" who it hit last time.  It's like assuming you have a high probability of getting tails on a coin toss after 9 consecutive heads - the chance of every coin toss is 50/50 regardless of previous coin tosses.

If someone wins the lottery twice in a row, you would call them lucky, but the chance of winning the lottery the second time is no greater or lesser than winning it the first time (assuming the number of people buying tickets remains equal).  It is statistically just as likely for John to win the Lottery twice as for John to win the lottery once and Bill to win the lottery once, yet we don't call that "luck" because of observer bias.  Every outcome is statistically equal but the outcome of John winning twice seems especially fortunate to us, so we call it "lucky."  This is a result not of the statistical improbability of the lottery, but a result of the impressively good result of getting $200 million (or whatever).

That's only true when you view each incident as isolated. The odds of winning the second time are the same as the odds of winning the first time, sure, but the odds of winning both times are higher. In fact, I believe the odds of multiple wins grow exponentially. The odds of winning any given coinflip, for example, is 50%, no matter what the outcome was on any other given coinflip. The odds of winning two consecutive coinflips, however, are higher, and the odds of winning 42 consecutive coinflips are damned incredible.

[Matt Larkin (author)]

I'm actually dismissing the plausibility, outside of fantasy settings, of luck being anything but a statistical fluke.

Agreed. Statistical outliers will crop up without the need for a supernatural explanation like a person actually being "lucky." Whereas, we commonly use the expression of "lucky" just as means of talking about someone that has already beaten the odds (confusing it with the concept of someone mystically predisposed to beat the odds).

[Polycarp]

That's only true when you view each incident as isolated. The odds of winning the second time are the same as the odds of winning the first time, sure, but the odds of winning both times are higher.

The odds of John winning the lottery twice are higher, versus every other result.  However, that outcome is no more or less likely than any other result.

Think of it this way.  We would call rolling a 100 on a d100 a "lucky result" (assuming that higher is better for this roll).  Rolling two 100s in a row would be amazing.  Yet the chances of that happening are just as remote as rolling, say, a 67 followed by a 22, and when somebody rolls that we aren't amazed by their good fortune even though it is equally unlikely as rolling two 100s.

The point is that, from a statistical perspective, there's nothing "special" about John winning the lottery twice or rolling two 100s.  They are just as likely as any other possible outcome.  We only call them "lucky" because we attach special symbolic significance to those particular results and think of them as fortunate.

The concept of "luck" comes from a biased interpretation of statistics.  How else can we explain the phenomenon where one result is considered "lucky" while all the other equally likely results are not?

[limetom]

Gambler's falacy or it's inverse?[/url]  

[SDragon]

The concept of "luck" comes from a biased interpretation of statistics.  How else can we explain the phenomenon where one result is considered "lucky" while all the other equally likely results are not?

Because a quirk in the definition of the term "lucky" makes it only apply to good luck. If you call heads twice in a row, and win both times, your good luck is equal to the bad luck of the guy that called heads twice in a row, and had the coin land tails twice in a row. Of course, since the odds of one tail and one head are higher then both flips being the same (at least, presuming that it doesn't matter which flip has which outcome), that would be irrelevant when talking about "equally unlikely" scenarios.

By the way, if the 67 and 22 were somehow called, then I'd say that's lucky. Obviously, if no numbers were called at all, then I wouldn't say that the pair of 100s were lucky, nor would I say that the 67 and 22 were lucky. In a case where numbers weren't called in any way, then neither situations would be either fortunate, nor unfortunate, so "luck" would be completely irrelevant.