Get out and vote

November 1, 2004 8:30 AM

Andrew Tanenbaum never ceases to amaze me. This delightful surprise was revealed today to visitors of one of his web sites -- which (of late) probably includes 96% of MeFi.
posted by RavinDave (50 comments total)

Both links render blank white screens to this Safari user at this moment, alas.
posted by digaman at 8:39 AM on November 1, 2004

Sorry, working fine now.
posted by digaman at 8:41 AM on November 1, 2004

Interesting. Nice to see it was a CS guy. I actually didn't know which way (politically) he was going to go.
posted by dig_duggler at 8:43 AM on November 1, 2004

he seems like a nice guy --and a big thanks to him from all of us. When you look at places like that RealClear Politics place (or whatever it's called), you see how slanted it all could actually be.
posted by amberglow at 8:46 AM on November 1, 2004

Wow, how interesting. Tanenbaum's CS books are classics. (Note the site has been Slashdotted, so it's pretty slow right now)
posted by gwint at 8:51 AM on November 1, 2004

amber ... ya know, I'm so tickled by this revelation and so impressed with the good Doc's efforts over these last few weeks that I'm gonna drop my (legitimate) beef with the DNC and toss Kerry a vote.
posted by RavinDave at 8:52 AM on November 1, 2004

Mirrors at:
electoral-vote2.com, electoral-vote3.com, electoral-vote4.com, electoral-vote5.com, electoral-vote6.com, electoral-vote7.com, electoral-vote8.com, and electoral-vote9.com
posted by zsazsa at 8:52 AM on November 1, 2004

This is pretty nifty, but I don't get why it is cooler than other electoral sites like http://004k.com etc.

[/apologies for not getting it]
posted by mecran01 at 8:55 AM on November 1, 2004

I guess I'm with mecran01, in that I don't know why this revelation is so staggering to y'all. No pejoratives in there. I'm just saying if somebody could explain the radness of this I would appreciate it.
posted by Hildago at 9:01 AM on November 1, 2004

he's a compsci hero, uber-geek guru kind of guy.
and it was a pretty good site - best i've found for tracking what's happening.
posted by andrew cooke at 9:02 AM on November 1, 2004

It's because Andy Tannenbaum has geek cred (via his books and the creation of MINIX*). You get it by actually doing something cool and techie, versus just posting about on Slashdot.






*It was his refusal to make significant changes to MINIX, which he uses to teach operating system design, that got Linus Torvalds to create Linux.
posted by tommasz at 9:02 AM on November 1, 2004

mecrran01 ... I guess it's a geek thing. Tanenbaum isn't just some nameless schlep to most of us who've been online for a long time. Many of us know him from the intimate pre-net days and even recall his famous exchange with Linus Torvalds. Great fun. Great memories.
posted by RavinDave at 9:05 AM on November 1, 2004

Did I say "pre-net"? Duh! I meant "pre-web".
posted by RavinDave at 9:06 AM on November 1, 2004

I have to admit I was downright giddy at the prospect of the votemaster coming out of the closet. Even though I didn't know much about Mr. Tanenbaum until today, he has been one of my heroes for the past few months.

mecran01, don't apologize, just know that some of us prefer graphic representations over columns of numbers.
posted by whatnot at 9:08 AM on November 1, 2004

amber ... ya know, I'm so tickled by this revelation and so impressed with the good Doc's efforts over these last few weeks that I'm gonna drop my (legitimate) beef with the DNC and toss Kerry a vote.

For real? w00t! : > (and us Dems never drop our beefs with the party, so you'll fit right in--we always are pushing them)
posted by amberglow at 9:22 AM on November 1, 2004

Ok, now I get it. The point was that it was a *secret* that uber-geek was running this site, and now he has revealed himself.

[/sound of brain turbo kicking in]
posted by mecran01 at 9:35 AM on November 1, 2004

I respectfully dissent. The practice of color-coding states on the sole basis of the most recent poll is extremely misleading. I prefer this guy, who looks at trends rather than single polls. He's also a Bush supporter, so I feel like I'm getting a worst-case scenario.
posted by PrinceValium at 9:54 AM on November 1, 2004

PrinceV, the electoral vote guy doesn't just use the last poll, he's got a scheme described on his site where he averages the past several days of polls to arrive at his numbers.
posted by mathowie at 10:08 AM on November 1, 2004

He's done a whole lot to publicize traditional and novel (i.e., "2004-specific") problems associated with polling, and in an accessible way. He's also been admirably transparent and responsive in terms of the processes he uses to arrive at his conclusions. Kudos to him.
posted by stonerose at 10:14 AM on November 1, 2004

Looking at the map, I feel like I'm playing The Political Machine, which is a bargain-bin, humor-laden presidential election politics simulator, which I wholeheartedly endorse playing today and tomorrow, just for fun. (Pitting Hilary against Condi was a hoot)

[/videogame stump]
posted by WolfDaddy at 10:25 AM on November 1, 2004

To answer the question of why I am in The Netherlands, my lovely wife is Dutch and long ago we decided that she couldn't do her kind of work in the U.S. whereas I could work anywhere.

Hooker or dealer?
posted by Ufez Jones at 10:33 AM on November 1, 2004

Hooker or dealer?

Or maybe a bona fide liberal?
posted by stonerose at 10:38 AM on November 1, 2004

PrinceV, the electoral vote guy doesn't just use the last poll, he's got a scheme described on his site where he averages the past several days of polls to arrive at his numbers.

Does he now? I thought he decided to stick with lastest-poll-wins, which he was doing up until at least a week ago. He keeps changing his methodology, though, so you could be right.

Speaking of electoral voting sites, I check electoral-vote.com obsessively, but I'm also partial to Race2004.net, which I don't think anyone's mentioned. They average the three most recent polls, and are great especially for checking the results of polls over time -- no graphs, but the most detailed charts I've seen.
posted by rafter at 10:41 AM on November 1, 2004

Mathowie, I think he goes with the latest. Iowa is listed as "Weak Kerry," and the only support for that is a Zogby poll showing a 6 point lead. That's a complete outlier. Mason-Dixon has Bush with a 5 point lead, and Gallup shows Bush +2, and a third shows a tie, while polls prior to this weekend showed Bush ahead. So there's no way a trending analysis supports Iowa being light blue.

Same thing with Wisconsin, where Zogby shows Kerry +7 and Gallup shows Bush +8.
posted by PrinceValium at 10:57 AM on November 1, 2004

Tanenbaum: "Starting Oct 4 the methodology was changed. There were so many polls and they were so far apart that the most recent 3 polls per state were averaged."

Race2004 uses a similar methodology. Results are different; in a perfect world, they'd agree on what polls to average and work from the same dataset.

Daly Thoughts (ref'd above) doesn't seem to post their methodology.
posted by lodurr at 11:05 AM on November 1, 2004

I know that he dinged Gallup for oversampling Republican opinion; giving it more weight based on wildly optimistic GOP turnout (and lowballing projected Democratic turnout). I wonder if Tanenbaum RE-adusted their imput accordingly. Might account for the disparity, no?
posted by RavinDave at 11:25 AM on November 1, 2004

For those who can't get enough of this stuff, here is a compilation by Ed Fitzgerald of dozens of sites that are running electoral vote counts, with results pretty much all over the place.

For my money, the smartest approach is this one, by Prof. Sam Wang of Princeton, which tries to obtain a more precise result by doing a statistical "meta-analysis" of all the polling data. Assuming the undecideds do vote, Wang says Kerry wins, 323-215, pop vote 49.3-48.7. Full source data, methodology and explanations included.
posted by beagle at 11:40 AM on November 1, 2004

I can't believe someone managed to bring RavinDave back into the big tent!
posted by rcade at 11:52 AM on November 1, 2004

I like Professor Wang's map of the states adjusted for their percentage of electoral votes.
posted by y2karl at 12:00 PM on November 1, 2004

I've been watching Slate's tally, updated extensively with scads of methodology (although theirs is a little more holistic than most, all about trends and auras).
posted by grrarrgh00 at 1:00 PM on November 1, 2004

slate is calling a TIE in the EC? (no. no. no. GOD NO!)
posted by bluno at 1:05 PM on November 1, 2004

While we're talking about places on-line to track the voting, here's a collection of reference links for reporters covering the voting by MSNBC writer Jonathan Dube. (From Poynter On-Line, a web portal for journalists)
posted by mmahaffie at 1:08 PM on November 1, 2004

well, bluno--i've already heard about 1 faithless elector in W.VA who says he won't go Bush even if he wins, throwing his EV away on Cheney or someone, so that would break any tie there.
posted by amberglow at 1:09 PM on November 1, 2004

Actually, that wouldn't break the tie, because Kerry would still need 270 votes to win. All it would do would make Cheney eligible to be elected. Here's a great article about some of the weird things that could happen in the event of a tie: Tie Goes to the...
posted by llamateur at 1:51 PM on November 1, 2004

As llamateur said, Amberglow, in the case that nobody wins 270 electoral votes, the THREE candidates with most votes enter in a House of Representatives sweepstakes, where each group of state delegates vote as a block. To win the election in the House a candidate needs at least 26 votes. That wouldn't be good for Kerry.
posted by sic at 2:31 PM on November 1, 2004

ohhhh...what if i paid off that guy in WVa to give his vote to Kerry then? ; >
posted by amberglow at 2:46 PM on November 1, 2004

Same thing with Wisconsin, where Zogby shows Kerry +7 and Gallup shows Bush +8.

Have you paid any attention to Gallup's polling tendencies. Zogby is far more reliable (see: history) than the pro-republican gallup samples.
posted by The God Complex at 3:02 PM on November 1, 2004

John Zogby went on the Daily Show last week saying Kerry would win, so that throws his credibility into doubt for me.
posted by PrinceValium at 3:04 PM on November 1, 2004

John Zogby went on the Daily Show last week saying Kerry would win, so that throws his credibility into doubt for me.

Then you don't know much about Zogby. Sorry. He's been fairly strong over the years in predicting outcomes of elections because he's a very strong poller. Unlike Gallup, which uses highly questionable cross-samplings (like 8% higher republicans despite recent history suggesting that democrats turn out in higher numbers and have increased registration more than republicans). Look at Zogby's numbers from the last election; he was off by I believe .1%

The fact that he made a prediction doesn't mean he's partisan. It means he recognizes voting trends and the historical significance of undecideds trending 80% towards challengers, as he's outlined very specifically when talking about it.

I don't get your knee-jerk dismissal of his statistics given his history as a poller. It seems like more of the insanely illogical run-to-the-centre in american political discourse.
posted by The God Complex at 3:13 PM on November 1, 2004

If he's right, he's right, but I think a strict policy of neutrality is essential to a trustworthy poll.
posted by PrinceValium at 3:28 PM on November 1, 2004

what Zogby says, tho, has no bearing on his methods--as TGC said, he's not oversampling one side or the other, as Gallup does.
posted by amberglow at 3:31 PM on November 1, 2004

Saying who is going to win based on polling data and your experience in past elections makes you partisan? How do you figure that?
posted by dig_duggler at 3:34 PM on November 1, 2004

If he's right, he's right, but I think a strict policy of neutrality is essential to a trustworthy poll.

How does making a prediction affect the neutrality of the polling? One would assume he's making his predictions based upon his polls, not polling to back up his predictions.
posted by The God Complex at 4:01 PM on November 1, 2004

Electoral-Vote.com also has a predicted final results page, which calculates undecideds as breaking 2:1 for the challenger, as they have since the 60s. It predicts 306 to 218 for Kerry.
posted by D at 4:03 PM on November 1, 2004

He (Zogby on The Daily Show) called it for Kerry, and then made it very clear that he'd be 'huddled in a foetal position' for several days if his polling turns out to be off-base. He takes his reputation very seriously, is the impression I got from him, and his Kerry victory prediction (with which I would agree, looking at it from afar as I am) was, I think, based solely on the numbers as he understood them. Not partisan. Not that I see anything wrong with that.
posted by stavrosthewonderchicken at 4:07 PM on November 1, 2004

I found myself really beginning to like this guy. He remained credible throughout the campaign and was very honest about his methodology.

I am glad I don't have to call this election. I do agree with Zogby that, while there is a pretty even split between those who like George Bush and those don't, those who don't like the president, really loathe him. And Loathing goes a long way in politics. I really can't imagine many people with Democratic leanings (meaning people like me who don't like the party, but vote for it regularly) not getting off their asses and getting to the polls.
posted by gesamtkunstwerk at 5:45 PM on November 1, 2004

Help me out here - if electoral-vote is averaging the last three polls then why is Hawaii listed for Bush?
posted by PrinceValium at 5:59 PM on November 1, 2004

Hey, that guy in W. Virginia is my uncle. Holy small world.
posted by tcaleb at 7:26 PM on November 1, 2004

I'm one of those old netheads who saw the name Andy Tanenbaum up there and thought, Wow -- what cool website could that possibly be? Talk about your pleasant surprises!

As far as I know Electoral-Vote.com still uses the last poll, but that's judged by the polls' median date, not the release date or end date. I believe an average makes a certain amount more sense, because the way it is the map or any of the charts become very noisy due to inevitable sampling errors -- but from what I know about stats simply averaging two polls together without working from their internals is a pretty iffy proposition. It's a defensible position.

I don't think there's anything wrong with E-V.com, race2004.net, or nowchannel.com (the new name of 2.004k.com). All are trustworthy and valid enough in my opinion. If Tanenbaum has a flaw it's in uncritically including polls such as Gallup which have taken a lot of flak over alleged sampling bias and weighting issues. We'll only know for sure tomorrow night, one supposes.

Zogby is clearly a Democratic pollster who does most of his political work with Democratic campaigns. That's just the way the business is -- there isn't such a thing as a pollster who does equal amounts of polling for both parties. I prefer Zogby being up front about his connections and possible biases, though. Gallup leans Republican, based on their polling the last two cycles at least, but are still mainly known for their corporate and product polling, and are probably running on fumes from a generation of being hired by major media outlets. This year might seal their reputation, though. Mason-Dixon is IMHO a more reliable right-wing biased poll, if you want one.

Ultimately you pays your money, and you takes your chances -- because polling is an inexact science. This year the arguments have been not only about weighting but about the very underlying assumptions behind weighting, such as how you define party identification or likely voter. That's a healthy debate for the industry to have, so it's good that we've had a challenging season and an election that makes polling not only more important than usual (most of the elections in my lifetime were decided months before the vote) but specifically probably will confound almost all polling. Meanwhile many people have learned a lot about the process of polling.

The Mystery Pollster points out, by the way, that whatever calumny Gallup deserves because of their weighting, they are just about the only pollster at the top rank to release full results to the Roper Center, which permits people like the Mystery Pollster to dissect and critique their methodology.
posted by dhartung at 7:46 PM on November 1, 2004

You guys have seen this animation, right?
posted by betaray at 10:15 PM on November 1, 2004