Saturday, August 16, 2003
Polling 101: An AP story today is headlined: "Poll: Bustamante leads Schwarzenegger."
Uh, no. This is a classic case of the media maiking an assertion on very weak evidence. Let's assume that's it's a well done poll -- reasonable here. Let's also assume that the people who answered the poll really are likely to vote -- a more infy assumption. BUT.., the poll is showing Bustamante at 25 and Schwarzenegger at 22 with a FIVE point margin of error. Thus the true results are:
In other words, it's a statistical tie. Besides I suspect that for a lot of people's preferences right now are rather fluid, which is also why I put zero credibility behind initial polls that showed the actor with a big lead. Especially for Schwarzenegger support is potentially skin deep, which is a big problem for him. But even if I think Schwarzenegger's chances in this race have been overstated, this poll does not demonstrate that he is behind. Yet.
Thursday, August 14, 2003
It doesn't get much worse than this: Kenneth Arrow and others demonstrated a long time ago that there is no perfect type of election. Any method you choose compromises some set of valuable principles. Here and there I've heard pundits wax ecstatic about the very democratic election California is about to hold for governor. Look at all this choice? Look at how there were virtually no barriers to the ballot so anyone could run? This is a great day for democracy!
Yeah, right. Almost certainly a majority of voters, probably a huge majority of voters, won't get their first choice for governor. They may not get their second, third, fourth, or even fifth choice. Indeed, it might not happen, but it's certainly possible with this kind of voting scheme, that a majority of voters will get their last choice from the list of serious candidates.
Plurality means whomever gets the most votes wins. So in a race with 165 candidates it's possible, though highly implausible, that the winner does so with less than 1% of the vote. Far more likely is a winner with something like 20%.
All of this you knew already, but it is only the beginning.
Off-year elections like this usually draw incredibly small turnouts. But this is such a unique, high profile occurence that it's possible that the turnout will be quite high. Quite high means about 50%. So if the winner gets 20% of the vote then he or she wins with the expressed support of just 10% of the electorate. There's a mandate for you.
Only it gets even worse. Warning: Light math ahead.
Here's a really simple example. Let's say we have 5 voters (V1-V5) who choose among 4 candidates: Orange, Silver, Gray, and Purple. Each voter has his or her own ranking from most liked to least liked and, of course, votes for the most liked candidate. Each voter ranks the candidates as follows:
V1: Gray > Silver > Purple > Orange
V2: Silver > Gray > Purple > Orange
V3: Purple > Silver > Gray > Orange
V4: Orange > Silver > Purple > Gray
V5: Orange > Gray > Silver > Purple
Who wins? Orange (2-1-1-1) with a landslide 40% of the vote. But look at this a little closer. Orange wins because he is the first choice of two true believers. But he was the last choice of the other three voters (fully 60% of those voting). Let's push this even farther and say that not only do V1-V3 rank Orange last, but they consider Orange completely unacceptable as a governor. It doesn't matter. In this kind of election Orange wins.
At this point feel free to define Orange as your worst nightmare: Communist, Fascist, Green, Religious Right, Larry Flynt, Hillary Clinton, Pat Buchanan, David Duke, Carrot Top.
The example is hypothetical but not unrealistic. With the vote spread out over many candidates a candidate with a small but intensely loyal base can win even though that candidate is loathed by everyone else. Just about any other major election method would be better than this. Separate run-off. Instant Run-off. Approval voting. Borda Voting. All are flawed but all are better than this.
Wednesday, August 13, 2003
California Follies: Now that I'm back from that mythical region that lies between the East and West coasts I suppose I should say something about California: Uhm, well at least with the recall we don't have to hear about the Kobe Bryant case twenty-four hours a day.
Who benefits from this? The media. Schwarzenegger's presence makes this a superhot story at the slowest news time of the year. So the media is thrilled. They get to pretend they are covering "substantive" news which then gives them cover to talk about Bryant and Peterson all they want in between. Besides, elections are the one kind of important political topic that the media like to cover, because elections can be presented as sports events.
President Bush. This is nothing but good news for Bush. Not because he might get a Republican ally in Sacramento -- that's an overrated asset -- but because this story distracts from the only other "substantive" news story of the moment: That little business about why we went to war with Iraq coupled with the continued attacks against our troops.
Who else? Local television stations that pick up some extra advertising cash during a slow season. Campaign consultants. Whatever vendor prints the giant ballots. Comedians. Video rental outlets who suddenly find Pumping Iron a hot item.
In other words, lots of people. Just not the people who supposedly this is for: the taxpayers and voters of California. They get to pay for a hugely expensive (at least $66 million, probably a lot more), utterly unnecessary election that probably will yield them a toxic divided government until 2006. What a bargain?
I'll talk about this tomorrow, but to make it all worse the election rules that govern this election are terrible. Terrible. You'd be hard pressed to find a more undemocratic way to elect someone and still be able to call it an election.
Monday, August 11, 2003
DARPA for the traveling masses: On the 1st I was on a flight to Iowa (Their motto: Corn. Pigs. Presidential Candidates) and flipped open American's flight magazine. Imagine my surprise when I came across a puff-piece on DARPA. Oh, the timing. Probably because of the timing it made for interesting reading. It even included a nice bit of foreshadowing: "Pushing the envelope as DARPA does, though, can stir up occasional controversy." I'd say. Is it just me or did this futures market idea provoke more "outrage" than the far more potentially pernicious Total Information Awareness business?
Check out the piece on DARPA. They claim more innovations than Al Gore. (The internet, even.) My hope is that this most recent faux pas -- not a bad idea, just bad politics -- won't cause permanent damage. (Though I shed no tears for John Poindexter who betrayed his military oath years ago, and should have spent some time busting rocks in Leavenworth, followed by a tour as a cook's mate. But I digress.)