• Marco Carbone

    Yes, he revealed his identity this morning. I’m very impressed. It’s great to have someone with reliable statistical know-how behind such an important site.

  • Greg

    Well, I was just impressed that with his ego, he waited this long to tell us :o).

  • Jardinero1

    The problem with the most recent polls, especially in the state of Florida where 35% of eligible voters have cast their votes, is that the remaining cohort of eligible or likely voters is qualitatively different than the cohort of eligible or likely voters just two weeks ago. These eleventh hour polls are useless for making any predictions for that reason.

    Exit polling will be just as worthless since the cohort who went to the polls early is going to be different than the cohort who go on election day.

  • Max Lybbert

    I enjoyed visiting the site, and found the ability to float the cursor over states to see the latest poll data useful. However, the decision try as hard as hell to put every state in a column, even if it were a statistical tie (such as Florida, which seems to change flags daily) made it less-useful than Real Clear Politics. I usually visited Electoral Vote Predictor first, but I always stopped by RCP to see which states were really ties. Otherwise states changed allegience too often to be useful.

  • Fred

    Regarding the barely Bush/Kerry states — I argued the same way, that it’s ok to color the states, but the overall totals should make it clear that they shouldn’t be counted. The big total of 298 v 231 is *really* misleading without that caveat. I lost, although Andy did point out that the subtotals are right there, and he sometimes put explanatory text about this as well.

    In all, though, I agree, kudos to Andy for a site well done.

  • Max Lybbert

    Sorry, Adam, it’s an issue of reality, but not like what you insinuate.

    I got tired of seeing things swing massively within days. One day, I’d visit and Bush would be “up” with nearly 300 votes, and the next day, Kerry would have 280, and the next Kerry would be at 260. It always bothered me because I knew people weren’t really changing their minds, The site was simply taking all polls at face value, instead of accounting for margin of error. If Bush is up by 1% in a poll with a margin of error of 4.2%, should the whole state really go in his column? When tomorrow’s poll shows Bush losing 1.5%, did he really lose that, or is it general statistical noise?

    Sure, this was explained in the text, but I wanted to look quickly and get a general feel for what the polls really showed. RCP seemed to do that better, but I liked to see a real map, so EVP was my second site. That, and I liked seeing the current poll results by floating my mouse over the states.

  • Jardinero1

    I don’t know why everyone is so impressed with this guy. His methodology incorporates only the most recent polls in each state. This is a fatal flaw no matter how well he massages the numbers.

    The only polls of likely or eligible voters which may be said to have any veracity are the polls conducted immediately before early voting started. With each succeding day of early voting more “likely voters” were peeled off and became voters, period. This leaves behind a diminished pool of likely voters and becomes a less and less representative cohort of all likely voters.

    In a place like florida where 35% of the eligible voters have voted it is highly likely that a substantial plurality if not a majority of the original(three weeks ago) likely voters are gone.

    I think Tannenbaum’s map will prove to be massively wrong, in the swing states tomorrow. If you want to predict how the swing states will do, look at the polls from three weeks ago for a better indicator.

  • http://mossback.org Richard Bennett

    Dr. Tanenbaum’s election predictions were somewhat less than brilliant.

  • Jardinero1

    They sucked.

    While I am sure Tannenbaum is very numerate he should re-read the chapter on replicates in a statistics 101 textbook. That was his critical error as was the error of every polling organization on the planet for this election. The dynamics are explained in my missive above.

  • J.B. Nicholson-Owens

    Yes–how “amazing” is the site if its prediction are wrong? Tannenbaum’s site appears to be quite overrated.

    Perhaps the technocrats need to reassess how serious the mandate for President Bush is in this country. Four million votes can’t be ignored or blamed on Nader (and in 2000, Nader wasn’t to blame for Gore’s failure to take office). Democrats did it to themselves nationwide, losing a number of important races because they ignored their base and pursued a Republican-made agenda.

    Perhaps these losses will incentivize the Democrats to become a progressive party and rally people with their message, not their corporate campaign funding. And perhaps pigs will fly.

  • ben wolfson

    You say “however much he knew…” as if the prominence of Linux despite his disagreements with Linus Torvalds somehow shows he doesn’t know very much about OS architecture, while, in fact, he literally wrote the (OK, a, but a very highly regarded) book on it, which Linus used while writing Linux. The fact that it did happen to be the case that a monolithic kernel worked for Linux, that doesn’t unseat the theoretical arguments for microkernels; it’s contingent.

  • http://devnull.typepad.com/nullity/ nullity

    I just wanted to concur with Mr. wolfson. Notwithstanding a certain famous disagreement with Linus, Andrew Tanenbaum is seminal. Although it’s a generalization and an analogy I think it’s fair to say that in relatively the same way serious programmers know the late W. Richard Stevens they also know Tanenbaum (although I concede that many more programmers are busting out Advanced Programming in the UNIX Environment and UNIX Network Programming at work than, say, Modern Operating Systems, I think that’s a matter of practicality rather than seminality).

    As an example of a serious programmer, the original implementation of Linux was inspired by and named after MINIX.

    I wanted to post a comment to this effect yesterday but 1) I wasn’t sure if I was the only one reading the title the same way Mr. wolfson did and 2) Prof. Lessig is coming to my school on the 11th and whenever I see that picture of him leaning against the column with his arms folded, I read in his steely gaze “I am a dude with some serious kung fu.”

  • Max Lybbert

    I won’t argue about how much Tanenbaum knws about operating system design (and especially distributed OS design). I’ve been studying what papers of his are available for a few years now.

    I will nit-pick though. Nullity wrote, “As an example of a serious programmer, the original implementation of Linux was inspired by and named after MINIX.” While Torvalds studied Tanenbaum’s book in school, the design of Linux was inspired far more by his professor’s views of OS design (i.e. monolithic kernel — especially considering the fact that “Linux” is only the kernel). Torvalds was able to use lots of information from the book, such as the file system design and POSIX compatability, but I’m not sure how much the book “inspired” him. And Torvalds didn’t name Linux.

    But, due to recent Tanenbaum postings, I don’t know how much he understands surveys. His understanding of statistics and math is admirable. But I’ve lost some respect for his non-technical side now that he’s added his voice to the crowd claiming Bush stole this election because the exit polls were wrong.

    For the record, exit polls end up dealing with incredibly small non-random samples. A pollster hands out surveys at a few polling stations, but (like any survey) gets only a handful back. The ones returned are self-selected, not random, and may not break down demographically the same way the voting public does. For instance, in this election, 58% of the exit polls came from women. Somehow I think more than 42% of the voters were men.

    The pollster then has to massage the sample to get it to fit with pre-conceived notions of who would come out to vote, and that is where things broke down this time. The under-30 crowd overwhelmingly supported Kerry, but fewer under-30 voters came out than expected.

  • J. Toran

    “Could it be that reality isn’t doing what you want?”

    “…is a brilliant example of the brilliance of amateur (as in the Olympics) news on the net.”

    I really dunno if people just need to be told, in order to see a few things, or not…

    That’s why I can be SUCH an irritant, because I presume the.. uhhhhhh.. “irony” of all this is Rather palpable by now, Saturday after..

    ..but mebbe STILL not clear to some. Yes, this is a BRILLIANT example, awright, of what is the NOT-VALUE of amatuers. My Dad was a Professor of Journalism, but you folks combined together can’t seem to find a pot to pee in with both hands, nor a fly-on-the-wall to help…;-) Iow, if you can’t do, teach?.. and if you got SO much free time on your hands by teaching this crap here, then mebbe tenured professorship has outlived it’s usefulness.

    Wouldn’t bother me if it did all that much, especially in YOUR case Dr. Lessig et al, you being a great supporter of “put all the coders outta work, by calling non-’free’ immoral” philosophy-cult-religion, yourself…)-;

    Btw, I doubt if you actually stated “put ALL the coders outta a job”, literally, but what you imply is what you say, also.

    Bad part is, like in comedy, timing: Really hurts when, not just coders, but paralegals and other such upper-middle jobs get farmed out over-seas for relative-peanuts.. These two, together with other things, make for a “Perfect Storm” economically, although i don’t even have a 4-year degree m’self, so add afaik…


    PS The “Remember Me” functionality didn’t work this time, another great job by (presumably) an semi-amatuer coder…
    And as I’ve posted a few bazillion times, the UI of most-all blogging/web “tools” is almost-100% Pure BullCrap…

    Wonder why?