Thursday, September 30, 2004
Nonlinear feedback
So this is from the blog of a pretty friend of mine from Ohio:
"And, based on their handy-dandy interactive map, Ohio is no longer considered a swing state: it's definitely 'leaning towards' Bush. Dammit."
Even though I'm not allowed to vote in America, it's always bothered me what kind of an effect that unofficial polls by the mass media has on the voting population. For example, in the case of my concerned Ohioan friend (who I assume is a Democrat), the effect of hearing that her candidate is behind in the polls may be to cause her to become demoralized, so that she does not bother to vote on the rationale that her vote no longer makes a difference. On the other hand, the effect of the poll may be to make her more determined to vote than ever, hoping that her extra effort will swing the vote in the other direction. Either way, it seems clear that seeing the results of the poll would have an impact on her decision on whether or not to vote. What is not clear to me is whether the two effects I posited cancel each other out perfectly, when extrapolated over the entire voting population of Ohio. If they did not cancel, then what we would have is a kind of nonlinear feedback mechanism, where a poll of the voting population affects the voters' intentions, which in turn affects the next poll taken, etc.
And of course, it is not clear whether this voter-poll feedback would assist or impede the candidate with a marginal lead in a "swing state", but it is reasonable to say that it decreases the chances of a turnaround in a state where polls repeatedly claim that one candidate has a double digit lead over another. It isn't difficult to imagine then, the kinds of abuses of this effect the mass media could employ, in order to influence the final vote. The possibility for media abuse is all the more disturbing given the overwhelming support of the current president in the American mass media, ever since a certain day in September three years ago.
I don't claim to understand all the details, and I can't even show that such a feedback loop exists. However, I do know that concern over the effect of polls and media manipulation instituted a ban of polls in South Korea, for a set amount of time leading up to an election day. For example, media outlets in Korea could be forbidden from publishing any polls of the voting public starting a month before election day, and up until the vote takes place.
Something to think about for American voters.
So this is from the blog of a pretty friend of mine from Ohio:
"And, based on their handy-dandy interactive map, Ohio is no longer considered a swing state: it's definitely 'leaning towards' Bush. Dammit."
Even though I'm not allowed to vote in America, it's always bothered me what kind of an effect that unofficial polls by the mass media has on the voting population. For example, in the case of my concerned Ohioan friend (who I assume is a Democrat), the effect of hearing that her candidate is behind in the polls may be to cause her to become demoralized, so that she does not bother to vote on the rationale that her vote no longer makes a difference. On the other hand, the effect of the poll may be to make her more determined to vote than ever, hoping that her extra effort will swing the vote in the other direction. Either way, it seems clear that seeing the results of the poll would have an impact on her decision on whether or not to vote. What is not clear to me is whether the two effects I posited cancel each other out perfectly, when extrapolated over the entire voting population of Ohio. If they did not cancel, then what we would have is a kind of nonlinear feedback mechanism, where a poll of the voting population affects the voters' intentions, which in turn affects the next poll taken, etc.
And of course, it is not clear whether this voter-poll feedback would assist or impede the candidate with a marginal lead in a "swing state", but it is reasonable to say that it decreases the chances of a turnaround in a state where polls repeatedly claim that one candidate has a double digit lead over another. It isn't difficult to imagine then, the kinds of abuses of this effect the mass media could employ, in order to influence the final vote. The possibility for media abuse is all the more disturbing given the overwhelming support of the current president in the American mass media, ever since a certain day in September three years ago.
I don't claim to understand all the details, and I can't even show that such a feedback loop exists. However, I do know that concern over the effect of polls and media manipulation instituted a ban of polls in South Korea, for a set amount of time leading up to an election day. For example, media outlets in Korea could be forbidden from publishing any polls of the voting public starting a month before election day, and up until the vote takes place.
Something to think about for American voters.
