Agree with you regarding the lack of source code for PC-Pine. Cygwin Pine comes with source (though it isn't under an OSI-approved license).
The default mailbox format for PC-Pine is c-client MBX. This is more standard than 99.9% of the other options available under win32.
PC-Pine does support Unix mbox (as well as maildir, mtx, tenex, and mbox with CRLF instead of LF). It isn't always easy or intuitive to choose one of these alternative formats. Indeed, when I was still using Windows, I believe I might have made the mailboxes under other programs & given them to Pine to read. But it won't change the format of your mailbbox for you, so it is very possile to use unix mbox, or something else that can be grokked by more programs.
In PC-Pine's defense, cclient mbx performance is quite good.
Oops--here's the link for Mahogany. Of all those I posted, it sounds like the most promising: it is being built ground-up as an IMAP client and it has Python bindings.
Well, in addition to mutt, elm, gnus, and cygwin ports of KMail, Sylpheed, etc, there are a few lesser-known native open source MUAs on win32. Some are useful, some aren't:
Open Source other than Mozilla, all I can think of would be Pine.
I use Pine. I love it as an IMAP client (and of the cross-platform email clients, it is, with Mulberry (which I also use), still one of the top two IMAP clients out there). But it isn't open source. PC-Pine (the native port to win32) is a completely closed source product. It is available gratis (which is more than I can say for Mulberry), but without source. Furthermore, the Pine license for the *nix code is restrictive enough that many consider it "not Free:" You aren't allowed to release binaries of your own.
This being said, there aren't many open source email clients available natively on win32. However, many do work with cygwin. The *nix version of Pine (which, as above, might not be "open enough"), mutt, kmail, gnus, sylpheed and others work fine.
In the MozillaZine Forum, many discussed putting the win32 and linux binaries on a single stick & having them share profiles. Might as well throw in the Mac binaries too & then you'd have something really useful!
I was just responding to your post using the same maturity you offered. I was just more succinct.
In that case, I apologize for the childishness & am sorry it trumped supporting evidence and references I presented.
No, it shouldn't be. It should only be loosely governed by federal law to the extent of protecting the rights of individuals in that state (such as the privacy provision that the King County judge ignored).
OK. Sorry for misunderstanding your reply of "the handling of them is defined by federal law" in response to my suggestion that there should be "accurate/transparent/uniform standards for provisional ballots." Perhaps you should have said you disagreed with me, rather than implying we already had uniform standards.
No, it shouldn't have been. There's no reason to think a hand recount is more accurate (and in certain cases, such as the voting machines in Snohomish County, would certainly be less accurate).
Fair enough. But you admit that you were wrong when you said "Right, because it just concluded" to my statement "There isn't a mandatory hand recount being done now." (I assume you merely skipped over the word "hand" before replying to me.) I'm sure you acknowledge that in especially close races a hand recount is mandatory, so do you think that law should be changed too?
They are working to improve it. It's hard to say if it has been improved.
I agree with this revision.
No doubt we should work to improve those, but that is what the recount is *for*: to find those errors and fix them, and as long as errors exist -- that is, forever -- recounts will be given greater confidence than initial counts.
My argument would be that the first count shouldn't be certified as complete until all votes come in if it is so close that those votes could decide the race. (As you have pointed out, this snag will hopefully be improved.) If there is a third complete count, it would have very few ballots that weren't included in the second & it would be very difficult to defend having misread the ballots the second time around. Perhaps there is a lot of pragmatism into making the first count as quick as possible, even if we lose precision. But it seems quite wasteful to have a third count that renders the second useless. (The first was for speed, the last for precision, but why the counts in the middle?!) It wastes time and money. Either we should accept the second "precise" and "all-inclusive" count as definitive or we should say it is just as good as the third count & shouldn't give the third count ultimate weight. I'd opt for the later.
If specific errors were found in the second count & we want to discount it, we should make it easier to make revisions rather than going to the time and expense of completely redoing the count. If no specific errors were found, either don't bother counting or don't pretend it will be better than the second count.
If I'm allowed an interjection, phrases can have more than one valid meaning. While I agree with pudge that it wasn't the best phrase to use (because it is so commonly used in sampling and because other phrases would have less semantic ambiguity), I do not believe that the phrase only has the narrow definition which he attempts to prescribe. Indeed, the fact that he found many (common) "incorrect uses" should have been a clue. But, hopefully, he will respect the authority of the OED where the explicit definition and the examples of usage are incongruent to his perception of proper usage:
margin of error, a permissible or tolerable degree of deviation from a correct or exact value or target; a small allowance made for the possibility of miscalculation, change in circumstances, etc. (also margin for error); also fig. and in extended use.
1867 Chambers's Jrnl. 16 Feb. 106/2 For silver coin, the remedy or margin of error is fixed at one pennyweight per pound Troy. 1897 K. S. RANJITSINHJI Jubilee Bk. Cricket 169 In playing back to a fast bowler, the thing to remember is, that there is very little time to make the stroke, the margin of error being exceedingly small. 1927 A. L. BOWLEY & J. C. STAMP National Income 1924 vi. 52 The margin of possible error in the 1924 estimates must lead to caution in these comparisons. 1937 H. LONGHURST Golf I. xxii. 198 The blaster gives no margin for error above the ball, but an almost infinite margin below it. 1949 H. MACLENNAN Cross-country 7 The climate of Canada allows for less margin of error in economic life than does the climate of the United States. 1986 J. HUXLEY Leaves of Tulip Tree (1987) iii. 69 Julian..did not allow for the margin of human errors, and demanded implicit acquiescence. 1991 Newsweek 28 Jan. 23/2 If recent advances have brought the margin of error down to 30 feet or less..then the age of surgical bombing is finally at hand. 1996 Sunday Tel. 13 Oct. (Sport section) 3/6 The United goalkeeper had much less margin for error in the second half, when he had to save headers..and smother a close-range shot from Collymore.
I love it when people resort to this childishness, although they seem otherwise quite intelligent (if not particularly educated about particular subjects (this is not a slam--I don't know about everything & not even about everything I try to talk about either)).
In summary, though, we mostly (perhaps begrudgingly) agree that:
There should be a study conducted of how much votes swing with recounts. This may also help us find (and therefore compare) the errors in elections.
There is no guarantee we will have a voter-verifiable audit trail. Most think it is a good idea, so we will hopefully have one.
The handling of provisional ballots isn't completely dictated by federal law, but perhaps it should be for the sake of uniformity
A hand recount wasn't mandatory in this case, but perhaps it should have been.
Voting isn't perfect now, but is being improved.
There are also some points that I think I'm right about that I'd probably never convince you of, unless you read up on error analysis:
Error analysis can and should be used in vote counting, in the same way it is used in the physical sciences for any measurement
The law should acknowlege that the precision of the vote isn't perfect. If there is a procedure for ties (such as sending it to the legislative branch), perhaps there should be a procedure to handle cases where the lack of precision could undermine accuracy.
All counts should be done at a high enough level of quality that there'd be no reason to implicitly trust a later recount over an earlier count. The value of a recount would then include adding to the statistical confidence of a decision.
And, in the interest of fairness, a few concessions:
I shouldn't have used the phrase "margin of error." Though I would not be the first to use it to describe the lack of precision in a measurement, it is most commonly used by pollsters in sampling. People, especially those who have read a lot of newspapers & not so many scholarly science papers, would easily misconstrue me.
I shouldn't have said that the ability to govern shouldn't be changed based on how confident we are that the candidate won the majority of votes cast. This is quite radical & reflects my cynicism of government more than common sense. There are already mechanisms for ties & few (if any) change the powers of the office. These can be leveraged instead.
I should clarify the back-of-the-envelope calculations I make & what research I've done before making some claims. I should also be upfront about the things I am unsure of. Finding no record of statisticians being consulted or of a budget expenditure for a study into the precision of an election doesn't mean that such things didn't actually occur. It means I can't find evidence of them occurring & that I should suggest they should occur if they haven't already.
I shouln't resort to childish namecalling. I don't like it being used against me or seeing it be used against others. It doesn't convince me, and it doesn't seem to convince others if I use it against them, even if (especially if?) they have used it against me.
Objection: assumption of facts not in evidence. What you say is only true if it is possible to do it significantly better than it is done, and yet you have offered no evidence to support this assumption.
Do you have any evidence that it isn't possile to do it significantly better? We can both play the lack-of-evidence game. It is a great game to play as far as I'm concerned because the answer to it would be: we both admit it isn't perfect. Where's the objection to sponsoring significant studies to determine how not perfect it is and how to improve it?
Those are coming; they did not get them in time for this election.
There is proposed legislation to do this federally. There isn't any passed legislation to assure it will be in place at federal, state, county, and city levels by next year. While the notable bills that would give us voter-verifiable audit trails don't seem to be poisoned & should be passed, they aren't exactly being fast-tracked either.
So what? We had less percentage of error in this election than in 2000, when we DID have an audit trail.
And exactly how much error was there in 2000 and how much in 2004? You're making this up. (And after accusing me of doing something similar?!?!
Provisional ballots are new for this election, and the handling of them is defined by federal law
Some states had provisional ballots prior to the election. Congress made them required in all states this year. It is the states who determine which provisional ballots will be counted. Some throw out the provisional ballots cast in the wrong polling place. Some don't. Some require a form to be completely and accurately filled out. Some don't. There aren't even state-wide standards, as a 2002 court case in Colorado highlighted: three counties had three different standards in a statewide election.
Bull. It is possible to accout for that the statistical errors There are none, moron.
You'd agree that there are some if you knew what they are. Please provide a definition and say why counting votes is so perfect that there are none. If we can't find the systematic cause of all errors, explain why. If we can, please explain why we can't make a perfect count after we resolve all systematic errors.
If you don't want to go to the trouble of checking out either of the books I mentioned from the library, feel free to usetheweb.
Yes, you can say that. If you're an idiot. It's nonsensical.
Way to go calling everyone publishing scientific papers and even many engineers idiots! You can (and often should) apply error analysis to any measurement that you can make repetitively that has a different value each time you measure it. It isn't nonsensical at all. If the standard deviation is sufficiently low, you might have wasted time or money. But we can gather from many recounts that the deviation is higher than small spreads & so could be worthwhile.
Such thing can only make sense when you are doing sampling, which you are not doing.
Neither are scientists doing sampling when they measure the speed of light.
then you went back to talking about it anyway, because you're stupid. Statistical error ONLY applies to sampling.
You're stupid. Read the references provided & explain to me how making a fixed measurement (as done in examples in many of the references) is sampling. It isn't. Statistical error applies to any repeatable measurement that don't lead to identical results.
There is no such thing as experimental error in these matters. There's no experiment.
This is what is done, as much as possible. So you agree with the status quo, nice
This isn't done as much as possible. The fact that you think it is & don't even defend that thesis shows the idiocy of your quote-unquote arguments. You are an ostrich who has stuck his head into the sand of the way things are currently done & (worse) the PR of those currently doing it. Doing this doesn't make the real world go away.
If it were done as much as possible, the first count would be trusted as much as the second count! We'd already have voter-verifiable audit trails, accurate/transparent/uniform standards for provisional ballots/role purged, etc. I still vote using puch cards, which have long been known to be less than perfect (the criteria for a human counting a vote is very different from what the punch card machines can actually do). We've had alternatives, but haven't gone to any. Touch screen machines are still miscalibrated and parts of the process are still open to fraud by people of either political party.
Many machines can be fixed, but we don't fix them. Voting machines are contracted out & once we've bought them, we're basically stuck with them. Why can't we handle voting like we handle acquiring equipment for energy, science, aerospace, or defense? Though we contract a lot of work out, it is funded federally & scientists and engineers are hired by the government so that the contractor doesn't have free-reign.
That's not possible. You can only uncover actual errors, not infer (extrapolate, guess) errors
Bull. It is possible to accout for that the statistical errors and uncaught systematic errors, by adding an error to the number. When I measure something with a ruler, I don't need to know why I get different measurements in order to estimate my precision of those measurements. I can say that the length is the mean value +/- some error. It is the same with all measurements, including vote counting.
(this goes back to your lack of understanding of the difference between statistical sampling and actual counting).
No--but it is great that you keep going back to statistical sampling which you brought up & which I have repetitively said isn't what I am considering. You've failed to refute my claim that the differences in counts can be explained by experimental (statistical and systematic) error. You've failed to say why we can't or shouldn't use statistical tools to acknowledge and account for error. I'll give you a few: many Americans (such as yourself) don't really understand that measurements don't have perfect precision & have an even harder time knowing what "56+/-2" actually means. It would be slightly more expensive to do it "the right way" and act as if the accuracy and precision of our elections was as important to us as anything funded by the NSF, DOE, ORNL, etc.
And at some point, the more recounts you do, the more error you introduce, so "many" is unreasonable.
This is the stupidest thing you've mentioned to date. The only way to introduce significantly more error would be to damage or lose ballots (in which case we shouldn't even trust a small number of recounts). If you're really worried about this, look at samples of recounts. Compare the first three to the last three to three randomly selected. Once again, you scream your ignorance of error analysis.
First, this is largely unreasonable: you could only use what you learn in one election for the next election, and elections these days are changing so much (new types of voting, differences in voter demographics, addition of provisional ballots, etc.) we don't know if the difference between counts in this election would hold up for the next.
Well, then you should prove it. Recount ballots from the last several elections. If the evidence supports claims that you pull out of thin air, then we wouldn't be able to do it. But we
A recount is part of the process to make sure you count it as accurately as possible.
By throwing out the initial count in certified totals, we aren't being as accurate as possible.
If you make multiple measurements & they differ, you can reconcile those differences using statistics. No, you don't. You reconcile them by counting again.
I said can, but of course we don't. Duh. I know what we do now. That doesn't mean what we do now is right. How is the law of counting again and using the latest tally better in theory or practice than my proposition that we shouldn't just throw out the old count? Just because that is the way it is now?
We still have to care about that statistical error. There is none.
Machines and human eyes make fundamentally unpredictable and random errors in counting. If you could predict them please tell me how. Anything which isn't a systematic error is statistical error. Statistical errors that go uncaught are handled the way I have said statistical errors are handled--they're ignored.
How people "try" to vote is irrelevant, if they don't do so according to the law.
The law does recognize this, which is why hand counts are valid. The intent of the voter could be clear but our flawed machines might not be able to discern that. Those votes should be counted. If you followed all laws and directions when you cast your ballot, it should be counted. Unfortunately, you can "lose your vote" through no fault of your own.
Get it through your thick skull: there is no margin of error. That concept has nothing to do with actual counting of votes.
You're arguing semantics. The number of votes that changed between counts is greater than the spread.
But you did say the law assumes there is no error, which is clearly false: if it did, it would have no provision for recounts. Duh.
Read it in context. I was clearly describing statistical error. The same can be said to uncaught systematic errors. Recounts don't address errors which might be introduced in the recount that weren't present in the initial count. They don't address errors that can't be caught or are random between multiple recounts. They don't look at how many votes change between recounts.
The sorts of things that should have been done right the first time around. The sort of things that would mkae a third count different from either of the first two. Actually, no, some of those things should NOT have been done the first time around, because ballots came in after the count had been completed (Reed wants to change the law so this is less likely to happen in the future).
When I say should, I am not saying the law currently says this is the way it should be done. I am arguing from my idealism. But this semantic argument is old:
ME->Hey--the voting system isn't perfect we should change it.
YOU->That's not how we count votes. This is the way we count votes.
ME->But that isn't how we should do things.
YOU->It is right because it is the law because it is the will of the people because it is right.
ME->We could apply these tools to make things better
YOU->The law doesn't recognize those tools. We should do what the law says
ME->The elected leader should really be chosen by the true count of votes. Since every count has error, we should not only do what we can to minimize the error in a single count, but to add more certainty to the measurement.
YOU->You don't understand the law. You hate democracy. Duh. You hallucinate. I have no arguments that aren't phrased as inflammatory attacks. Duh. I like to prescribe feelings and thoughts to you and call you a liar when you sink to my level. Duh. The law says blah-blah-blah. That is the way it should be. Oh. Yes. There are
This is where the problem lies. What you're doing doesn't make any sense. You're using a ruler to count a discrete value: the number of pens and pencils.
No--this is an independent example. I am counting the length of the ruler by counting ticks on a ruler.
I put forth two different measurements requiring you to count to highlight differences.
Counting votes is like counting pens and pencils.
Possibly counting several millions of pens or pencils of different sizes and shapes and colors. Probably with a few of those multi-chamer contraptions which are both a pen and a pencil and a few highlighers which are neither. While some idiot pockets the pen you let him borrow and naother inadvertently leaves his pencil.
There are differences between the counts because errors can occur.
Is it or is it not possible for errors to change the count? How does one decide when an error has been made or when an error-free count has been made?
You've stated that you believe we can accurately count three votes (2 pens and a pencil). What about 30? 300? 300,000? At what point can we no longer get an accurate count? What if we broke the votes into smaller groups? Can we count 100 groups of 300 without errors?
I'm a scientist and there are no laws governing the counting of pens. I'd count a lot of times in order to estimate the error. I'd acknowledge that no individual count could be said to be dead-on accurate with infinite precision. With enough measurements, the estimate of error could be lowered enough to believe the result with some percentage of assurance. If such assurance was not met, a revote or some other mechanism could be used such that the impact of the choice is made less significant.
Being aware of error is important, but that doesn't mean we can't accurately count votes.
Errors change outcomes. There is no guarantee that with the current system the way people voted will actually dictate the outcome. It could be dictated by a flawed count.
You make stuff up about how the count can't be trusted, and then you say therefore the count can't be trusted. It's called begging the question.
If the vote can be trusted completely, there would never be a need for a recount. Mistakes are real & few deny that. I see no circular reasoning on my part & only you avoiding the legitimate and logical question of what if the outcomes change due to these errors.
Just because I don't believe in the numbers you invent out of thin air
I didn't make them up. They dont, of course, represent the actual error. They are a more reasonable guess of the order of magnitude for some sources of error. I'd be happy to entertain any reasonable estimates that you might offer instead.
This has *nothing to do with statistics.* Statistics -- in this context -- are for drawing conclusions about larger sets of data based on smaller samples.
I've said we aren't trying to extrapolate anything from random sampling. But statistics is a broad field and isn't useful for only random samplings. If you make multiple measurements & they differ, you can reconcile those differences using statistics.
When you make a measurement, there is a combination of nonrandom systematic errors (likely something like single-county screwups or outright fraud) and random statistical error. I am quite optimistic and believe (though do not present as fact) that there was no fraud. We still have to care about that statistical error.
If you don't want to call that "statistics," fine. Rename it as you please. If you don't think this matters, offer proof. Offer an explanation for why each and every vote migh be different between two counts and say why they intentionally didn't count correctly the first time through.
I assume that is a question, despite the lack of interrogatory mark. If so, your question is incomplete: was the law followed, or not?
It was rhetorical, but thanks for pointing out the bad grammar. I expect you can find a lot of that in my posts, just as I can find some in your posts.
If the law was followed, then the will of the people was followed because the law is the will of the people.
Now that is some good circular reasoning. It is the will of the people because it is law which is the will of the people? No. The will of the people as to who their leader should be is how the people actually voted. If the counting process fails to measure that, the will of the people as to who the leader should be can't be followed. Yes, the people elected leaders who made the law as to how votes are counted and how close races are handled. No, those laws dont fundamentally change how they tried to vote.
When the law proves inadequate, you fix it for next time.
Which is exactly what I'm making a case for. There is evidence that the margin of victory is less thanthe margin of error. You seem to throw up strawmen & assume I mis-state current laws.
I already told you. You said it assumes no error. You're clearly wrong.
How am I wrong? They count two times. They certify the later result. At no time do they even say "oh crap--this guy won by fewer votes which were 'miscounted' the first time around." Nor is there any mechanism to judge the merit of an individual count. Recounts can be as bad or worse than the initial count.
How does the recount law account for statistical errors? It does not and should not, as I already noted, since this is not about statistics, and there are no statistical errors.
Ah. So you are now willing to admit that the law doesn't account for statistical errors, although you used my claim that they didn't as evidence that I didn't understand or was misrepresenting the law as it now stands.
The spread is most likely not statistically significant.
Your attacks are pathetic. Your blind faith in vote counting is pathetic. Your ad hominem attacks are pathetic. You saying that people don't understand statistics because they don't agree with you is pathetic.
I have 2 pens and a pencil on my desk. Every time I count them, I get 2 pens and 1 pencil. It doesn't make sense to try to find any statistical error in this count.
NowI pick up a ruler and put it against the pencil. I record the number of ticks I count on the ruler (this count is, of course, speeded by the way the ruler marks off groups of mm ticks in cms, etc.). My count of ticks will not be the same if I remeasure after I stop typing this message. The count wouldn't be the same if I called my office-mate over to perform the meaurement. It is too hard to measure--Our eyes can't easily resolve ticks. Our hands can't hold the ruler perfectly parallel to the pencil, nor hold the 0 marker to the tip. Perhaps the eraser or the tip get worn down slightly between measurements. If accuracy is important, it makes sense to measure many times so that you can establish a mean and a standard deviation. A lack of precision in measuing can have substantial impact on our final result. Our precision would be much worse if we had fewer ticks further apart & had to extrapolate more. One individual measurement is less likely to be the "true" measure of the pencil than the mean is.
Counting votes should be like counting the pens and pencils, but is like measuring the pencil. It is hard. Each recount gets a different number. It makes perfect sense to apply error analysis to voting. Though we aren't intentionally making a random statistical sample of votes, we are aware of error.
If you disagree & think that counting votes is like counting pencils and pens, please tell me why there are differences between the counts!!!
So in your world, a close election means there is no democracy? That's nonsense. That is, it makes no sense. It's irrational.
No it isn't. I again pose the question: If you can't trust the counting of the votes, how can you say that it is the will of the people?
Since you can't wrap your head around simple statistics, let's simpify it. There are 11 voters. 6 want to vote for A and 5 for B. If 2 of those votes aren't counted, how can you say the will of the people is followed. Candiate A will not win if both votes that weren't counted happened to be for him. Candiate B would win 5 to 4. But that doesn't reflect the will of the people.
Fortunately, we don't have errors of 2 in 11. But the same kinds of problems exist when you scale up the number of voters. Not only do people not win by margins of 1 in 11 and not only are some votes not counted, but some are counted more than once. Some valid votes are lost or destroyed. Some votes that are counted shouldn't be--some are cast by people who aren't state residents, some people manage to vote more than once, it is a real mess.
We need to do as much as we can to make sure that that mess is small enough that it won't change the results of an election. We aren't there yet.
In my world, close elections remind us that we need to be as careful as possile to lower errors in the voting and counting processes. We can make laws that minimize the errors and can encourage candidates to both seek votes from as many people as possible and have an election which are as fair as possible. We have few such laws.
The error of any count is at least 219 votes. You're making that up.
No. They counted the same votes twice & got different numbers. This was the difference in the two spreads. That's error. If there was no error, the counts would be exactly the same.
I realize the law behind recounts. I just say that the law inaccurately assumes no error. Since the latter statement is false, it disproves the former.
What about the law behind recounts as it stands have I gotten wrong? Saying that the law should be some way is different than saying the law is that way. I think I've been clear with which I'm talking about. How does the recount law account for statistical errors? It is mute on error. It says if there is a spread of an arbitrary percentage or number of votes that there will be an automatic recount and that that the results from that recount will be legally certified. We don't report election results the way that people report pulls. We don't say so-and-so won with 53+/-1 %.
You said something clearly false. You said the spread is "virtually non-existant[sic]." The fact is there is a spread. It does exist.
Fine. The spread is most likely not statistically significant. Election laws don't account for statistics--they don't care about them. Dino won. That doesn't mean that statistics don't exist or that you can't apply them to this case--it just means that the state doesn't choose to do this when they put someone in office. This is why who is put in office isn't always what eh will of the people would have otherwise had. The will isn't perfectly communicated through the counting of votes.
Show me a law that says the governor will be the one who the people want. There is none. It says that it will be the candidate who has the greatest number of votes in the final count. Given that counts aren't perfect, there is a difference here.
No. Whoever legally wins represents the will of the people, period. There is no illusion. There's only ingrates who hate democracy, like you.
I don't hate democracy. I love democracy so much that I want the will of the people to ring loudly and clearly over and above the din and noise of fraud and mistakes. Look--the spread changed by 219 votes between two counts. That is gr
If we can't count the votes with enough certainty, then decisions can be made by chance. That isn't democracy.
You miss the point of the recount. It's assumed that a machine recount will still be accurate, if the initial spread is greater than 150, unless it overturns the result, in which case they WILL have more recounts.
The error of any count is at least 219 votes. If the election is decided by only 42 votes, then that is below the error! I realize the law behind recounts. I just say that the law inaccurately assumes no error. It may be politically significant that Dino won in both counts, but it isn't statistically significant because his margin of victory was so slim.
If you want to be deceptive, that's not fine. There's nothing non-existent about it, virtually or otherwise.
I'm not deceptive. I have stated how the law will handle it. I just say that elections aren't accurate enough to handle cases where someone narrowly wins like this. How is that deceptive? What dream-world do you live in where all votes are cast perfectly and all recounts lead to the same numbers? We just can't count with a precision of 1 in 100,000.
You say that as though it is in any way meaningful. All that matters is who wins: that person is the governor, period, with all the rights and powers and obligations that entails.
No--that isn't all that matters. Accuracy in voting also matters. This wouldn't be a story if who won was all that mattered. We didn't see 49 other stories on governorships. Furthermore, given the closeness of the vote & the many put who chose not to vote, I don't think who would actually win really matters to as many people as you think.
As long as one candidate has more votes, no matter if it is 1 or 1 million, that candidate represents the democratic will of the people
Wrong. If you aren't sure of 1 in 100,000 votes, how the hell can you say it is significant?
That's a fact, definitionally.
I didn't say that anything else would happen. But this isn't some definition. It is a legal illusion.
Unless you dislike with democracy, of course, which you already said you do, so why I am bothering?
No--you said I don't. If you can't even tell me from you, we have bigger problems.
This was the machine recount. If the initial numbers had gone this way, there would be a mandatory hand recount. As it is, they've already conducted the mandatory recount & any additional recount will need to be petitioned for my the participants.
I agree it is well below the margin of error. But that doesn't matter. Scientists have a number, an error, and a unit. Everyone else (including politicians who write the laws) doesn't use "error" and often neglects "units."
Not a "margin of error," but yes, that is precisely what does and did happen. The difference was less than 2,000 votes and under.5%, so they had a mandatory recount. This is what happened!
I was actually thinking that such a number shouldn't be arbitrarily chosen & that it should have consequences for the Governor's term in office (other than the implicit hard time he'll have getting any partisan issues steam-rolled through). Furthermore, this is a particularly interesting case. The initial spread of 261 was greater than 150, which would have required a manual recount. Yet on the machine recount, the spread dropped to less than this threshold. If such happens, they should really require a manual recount anyway.
Well, in fact, it does. There's no getting around it. It's not a matter of opinion. It's a fact. Whoever wins the legal vote is the choice that reflects the will of the people, by definition.
If you want to be curt, that is fine. I was actually saying that this debacle of a virtually non-existant spread is indicative of how we really don't have a strong opinion for either party. They were calling us a swing state for the presidential race, but the Kerry-Edwards 7% edge is orders of magnitude larger than this. It is also larger than the FL presedential race of 2000.
Yes, the one with the most votes legally wins the election. But if you think that the margin of voter error is larger than the marger of victory then no, you can't say that the choice scientifically reflects the will of the people (or even of the voting people).
I'm such a nerd. I really think that elections should have margins of error and something should happen if candidates don't win by at least that much. Something tells me that 0.0015% quantifies as sufficiently small to be below it.
Oh well--I really do think that this reflects the will of WA voters (and I am one). We didn't have a strong preference. Approximately 50,000 (2% or 1000x the margin of victory for Dino Rossi) more people voted for president than voted for our governor. Our divided state house & senat also bears out how moderate we are.
I agree that this study is rather dubious, but I also think that winning the embedded market does win desktops with it. So how should Free OSs exploit this? You can make something as "cool" as a zaurus, but you will only attract the uber-nerd. What kind of embedded devices can be made with elegant interfaces like the ipod with great support on Free OSs? Should someone be pushing for an embedded handheld linux video player?
Slashdot Mirror
Agree with you regarding the lack of source code for PC-Pine. Cygwin Pine comes with source (though it isn't under an OSI-approved license).
The default mailbox format for PC-Pine is c-client MBX. This is more standard than 99.9% of the other options available under win32.
PC-Pine does support Unix mbox (as well as maildir, mtx, tenex, and mbox with CRLF instead of LF). It isn't always easy or intuitive to choose one of these alternative formats. Indeed, when I was still using Windows, I believe I might have made the mailboxes under other programs & given them to Pine to read. But it won't change the format of your mailbbox for you, so it is very possile to use unix mbox, or something else that can be grokked by more programs.
In PC-Pine's defense, cclient mbx performance is quite good.
Oops--here's the link for Mahogany. Of all those I posted, it sounds like the most promising: it is being built ground-up as an IMAP client and it has Python bindings.
Well, in addition to mutt, elm, gnus, and cygwin ports of KMail, Sylpheed, etc, there are a few lesser-known native open source MUAs on win32. Some are useful, some aren't:
Phoenix Mail
Mahogany
JoeEmail
Python IMAP Email Client
This being said, there aren't many open source email clients available natively on win32. However, many do work with cygwin. The *nix version of Pine (which, as above, might not be "open enough"), mutt, kmail, gnus, sylpheed and others work fine.
In the MozillaZine Forum, many discussed putting the win32 and linux binaries on a single stick & having them share profiles. Might as well throw in the Mac binaries too & then you'd have something really useful!
If specific errors were found in the second count & we want to discount it, we should make it easier to make revisions rather than going to the time and expense of completely redoing the count. If no specific errors were found, either don't bother counting or don't pretend it will be better than the second count.
In summary, though, we mostly (perhaps begrudgingly) agree that:
- There should be a study conducted of how much votes swing with recounts. This may also help us find (and therefore compare) the errors in elections.
- There is no guarantee we will have a voter-verifiable audit trail. Most think it is a good idea, so we will hopefully have one.
- The handling of provisional ballots isn't completely dictated by federal law, but perhaps it should be for the sake of uniformity
- A hand recount wasn't mandatory in this case, but perhaps it should have been.
- Voting isn't perfect now, but is being improved.
There are also some points that I think I'm right about that I'd probably never convince you of, unless you read up on error analysis:- Error analysis can and should be used in vote counting, in the same way it is used in the physical sciences for any measurement
- The law should acknowlege that the precision of the vote isn't perfect. If there is a procedure for ties (such as sending it to the legislative branch), perhaps there should be a procedure to handle cases where the lack of precision could undermine accuracy.
- All counts should be done at a high enough level of quality that there'd be no reason to implicitly trust a later recount over an earlier count. The value of a recount would then include adding to the statistical confidence of a decision.
And, in the interest of fairness, a few concessions:Do you have any evidence that it isn't possile to do it significantly better? We can both play the lack-of-evidence game. It is a great game to play as far as I'm concerned because the answer to it would be: we both admit it isn't perfect. Where's the objection to sponsoring significant studies to determine how not perfect it is and how to improve it?
There is proposed legislation to do this federally. There isn't any passed legislation to assure it will be in place at federal, state, county, and city levels by next year. While the notable bills that would give us voter-verifiable audit trails don't seem to be poisoned & should be passed, they aren't exactly being fast-tracked either.
And exactly how much error was there in 2000 and how much in 2004? You're making this up. (And after accusing me of doing something similar?!?!
Some states had provisional ballots prior to the election. Congress made them required in all states this year. It is the states who determine which provisional ballots will be counted. Some throw out the provisional ballots cast in the wrong polling place. Some don't. Some require a form to be completely and accurately filled out. Some don't. There aren't even state-wide standards, as a 2002 court case in Colorado highlighted: three counties had three different standards in a statewide election.
You'd agree that there are some if you knew what they are. Please provide a definition and say why counting votes is so perfect that there are none. If we can't find the systematic cause of all errors, explain why. If we can, please explain why we can't make a perfect count after we resolve all systematic errors.
If you don't want to go to the trouble of checking out either of the books I mentioned from the library, feel free to use the web.
Way to go calling everyone publishing scientific papers and even many engineers idiots! You can (and often should) apply error analysis to any measurement that you can make repetitively that has a different value each time you measure it. It isn't nonsensical at all. If the standard deviation is sufficiently low, you might have wasted time or money. But we can gather from many recounts that the deviation is higher than small spreads & so could be worthwhile.
Neither are scientists doing sampling when they measure the speed of light.
You're stupid. Read the references provided & explain to me how making a fixed measurement (as done in examples in many of the references) is sampling. It isn't. Statistical error applies to any repeatable measurement that don't lead to identical results.
This isn't done as much as possible. The fact that you think it is & don't even defend that thesis shows the idiocy of your quote-unquote arguments. You are an ostrich who has stuck his head into the sand of the way things are currently done & (worse) the PR of those currently doing it. Doing this doesn't make the real world go away.
If it were done as much as possible, the first count would be trusted as much as the second count! We'd already have voter-verifiable audit trails, accurate/transparent/uniform standards for provisional ballots/role purged, etc. I still vote using puch cards, which have long been known to be less than perfect (the criteria for a human counting a vote is very different from what the punch card machines can actually do). We've had alternatives, but haven't gone to any. Touch screen machines are still miscalibrated and parts of the process are still open to fraud by people of either political party.
Many machines can be fixed, but we don't fix them. Voting machines are contracted out & once we've bought them, we're basically stuck with them. Why can't we handle voting like we handle acquiring equipment for energy, science, aerospace, or defense? Though we contract a lot of work out, it is funded federally & scientists and engineers are hired by the government so that the contractor doesn't have free-reign.
Bull. It is possible to accout for that the statistical errors and uncaught systematic errors, by adding an error to the number. When I measure something with a ruler, I don't need to know why I get different measurements in order to estimate my precision of those measurements. I can say that the length is the mean value +/- some error. It is the same with all measurements, including vote counting.
No--but it is great that you keep going back to statistical sampling which you brought up & which I have repetitively said isn't what I am considering. You've failed to refute my claim that the differences in counts can be explained by experimental (statistical and systematic) error. You've failed to say why we can't or shouldn't use statistical tools to acknowledge and account for error. I'll give you a few: many Americans (such as yourself) don't really understand that measurements don't have perfect precision & have an even harder time knowing what "56+/-2" actually means. It would be slightly more expensive to do it "the right way" and act as if the accuracy and precision of our elections was as important to us as anything funded by the NSF, DOE, ORNL, etc.
This is the stupidest thing you've mentioned to date. The only way to introduce significantly more error would be to damage or lose ballots (in which case we shouldn't even trust a small number of recounts). If you're really worried about this, look at samples of recounts. Compare the first three to the last three to three randomly selected. Once again, you scream your ignorance of error analysis.
Well, then you should prove it. Recount ballots from the last several elections. If the evidence supports claims that you pull out of thin air, then we wouldn't be able to do it. But we
By throwing out the initial count in certified totals, we aren't being as accurate as possible.
I said can, but of course we don't. Duh. I know what we do now. That doesn't mean what we do now is right. How is the law of counting again and using the latest tally better in theory or practice than my proposition that we shouldn't just throw out the old count? Just because that is the way it is now?
Machines and human eyes make fundamentally unpredictable and random errors in counting. If you could predict them please tell me how. Anything which isn't a systematic error is statistical error. Statistical errors that go uncaught are handled the way I have said statistical errors are handled--they're ignored.
The law does recognize this, which is why hand counts are valid. The intent of the voter could be clear but our flawed machines might not be able to discern that. Those votes should be counted. If you followed all laws and directions when you cast your ballot, it should be counted. Unfortunately, you can "lose your vote" through no fault of your own.
You're arguing semantics. The number of votes that changed between counts is greater than the spread.
Read it in context. I was clearly describing statistical error. The same can be said to uncaught systematic errors. Recounts don't address errors which might be introduced in the recount that weren't present in the initial count. They don't address errors that can't be caught or are random between multiple recounts. They don't look at how many votes change between recounts.
When I say should, I am not saying the law currently says this is the way it should be done. I am arguing from my idealism. But this semantic argument is old:
ME->Hey--the voting system isn't perfect we should change it.
YOU->That's not how we count votes. This is the way we count votes.
ME->But that isn't how we should do things.
YOU->It is right because it is the law because it is the will of the people because it is right.
ME->We could apply these tools to make things better
YOU->The law doesn't recognize those tools. We should do what the law says
ME->The elected leader should really be chosen by the true count of votes. Since every count has error, we should not only do what we can to minimize the error in a single count, but to add more certainty to the measurement.
YOU->You don't understand the law. You hate democracy. Duh. You hallucinate. I have no arguments that aren't phrased as inflammatory attacks. Duh. I like to prescribe feelings and thoughts to you and call you a liar when you sink to my level. Duh. The law says blah-blah-blah. That is the way it should be. Oh. Yes. There are
I put forth two different measurements requiring you to count to highlight differences.Possibly counting several millions of pens or pencils of different sizes and shapes and colors. Probably with a few of those multi-chamer contraptions which are both a pen and a pencil and a few highlighers which are neither. While some idiot pockets the pen you let him borrow and naother inadvertently leaves his pencil.Is it or is it not possible for errors to change the count? How does one decide when an error has been made or when an error-free count has been made?I'm a scientist and there are no laws governing the counting of pens. I'd count a lot of times in order to estimate the error. I'd acknowledge that no individual count could be said to be dead-on accurate with infinite precision. With enough measurements, the estimate of error could be lowered enough to believe the result with some percentage of assurance. If such assurance was not met, a revote or some other mechanism could be used such that the impact of the choice is made less significant.Errors change outcomes. There is no guarantee that with the current system the way people voted will actually dictate the outcome. It could be dictated by a flawed count.
If the vote can be trusted completely, there would never be a need for a recount. Mistakes are real & few deny that. I see no circular reasoning on my part & only you avoiding the legitimate and logical question of what if the outcomes change due to these errors.
I didn't make them up. They dont, of course, represent the actual error. They are a more reasonable guess of the order of magnitude for some sources of error. I'd be happy to entertain any reasonable estimates that you might offer instead.
I've said we aren't trying to extrapolate anything from random sampling. But statistics is a broad field and isn't useful for only random samplings. If you make multiple measurements & they differ, you can reconcile those differences using statistics.
When you make a measurement, there is a combination of nonrandom systematic errors (likely something like single-county screwups or outright fraud) and random statistical error. I am quite optimistic and believe (though do not present as fact) that there was no fraud. We still have to care about that statistical error.
If you don't want to call that "statistics," fine. Rename it as you please. If you don't think this matters, offer proof. Offer an explanation for why each and every vote migh be different between two counts and say why they intentionally didn't count correctly the first time through.
It was rhetorical, but thanks for pointing out the bad grammar. I expect you can find a lot of that in my posts, just as I can find some in your posts.
Now that is some good circular reasoning. It is the will of the people because it is law which is the will of the people? No. The will of the people as to who their leader should be is how the people actually voted. If the counting process fails to measure that, the will of the people as to who the leader should be can't be followed. Yes, the people elected leaders who made the law as to how votes are counted and how close races are handled. No, those laws dont fundamentally change how they tried to vote.
Which is exactly what I'm making a case for. There is evidence that the margin of victory is less thanthe margin of error. You seem to throw up strawmen & assume I mis-state current laws.
How am I wrong? They count two times. They certify the later result. At no time do they even say "oh crap--this guy won by fewer votes which were 'miscounted' the first time around." Nor is there any mechanism to judge the merit of an individual count. Recounts can be as bad or worse than the initial count.
Ah. So you are now willing to admit that the law doesn't account for statistical errors, although you used my claim that they didn't as evidence that I didn't understand or was misrepresenting the law as it now stands.
Your attacks are pathetic. Your blind faith in vote counting is pathetic. Your ad hominem attacks are pathetic. You saying that people don't understand statistics because they don't agree with you is pathetic.
I have 2 pens and a pencil on my desk. Every time I count them, I get 2 pens and 1 pencil. It doesn't make sense to try to find any statistical error in this count.
NowI pick up a ruler and put it against the pencil. I record the number of ticks I count on the ruler (this count is, of course, speeded by the way the ruler marks off groups of mm ticks in cms, etc.). My count of ticks will not be the same if I remeasure after I stop typing this message. The count wouldn't be the same if I called my office-mate over to perform the meaurement. It is too hard to measure--Our eyes can't easily resolve ticks. Our hands can't hold the ruler perfectly parallel to the pencil, nor hold the 0 marker to the tip. Perhaps the eraser or the tip get worn down slightly between measurements. If accuracy is important, it makes sense to measure many times so that you can establish a mean and a standard deviation. A lack of precision in measuing can have substantial impact on our final result. Our precision would be much worse if we had fewer ticks further apart & had to extrapolate more. One individual measurement is less likely to be the "true" measure of the pencil than the mean is.
Counting votes should be like counting the pens and pencils, but is like measuring the pencil. It is hard. Each recount gets a different number. It makes perfect sense to apply error analysis to voting. Though we aren't intentionally making a random statistical sample of votes, we are aware of error.
If you disagree & think that counting votes is like counting pencils and pens, please tell me why there are differences between the counts!!!
. Every time I count
No it isn't. I again pose the question: If you can't trust the counting of the votes, how can you say that it is the will of the people?
Since you can't wrap your head around simple statistics, let's simpify it. There are 11 voters. 6 want to vote for A and 5 for B. If 2 of those votes aren't counted, how can you say the will of the people is followed. Candiate A will not win if both votes that weren't counted happened to be for him. Candiate B would win 5 to 4. But that doesn't reflect the will of the people.
Fortunately, we don't have errors of 2 in 11. But the same kinds of problems exist when you scale up the number of voters. Not only do people not win by margins of 1 in 11 and not only are some votes not counted, but some are counted more than once. Some valid votes are lost or destroyed. Some votes that are counted shouldn't be--some are cast by people who aren't state residents, some people manage to vote more than once, it is a real mess.
We need to do as much as we can to make sure that that mess is small enough that it won't change the results of an election. We aren't there yet.
In my world, close elections remind us that we need to be as careful as possile to lower errors in the voting and counting processes. We can make laws that minimize the errors and can encourage candidates to both seek votes from as many people as possible and have an election which are as fair as possible. We have few such laws.
No. They counted the same votes twice & got different numbers. This was the difference in the two spreads. That's error. If there was no error, the counts would be exactly the same.
What about the law behind recounts as it stands have I gotten wrong? Saying that the law should be some way is different than saying the law is that way. I think I've been clear with which I'm talking about. How does the recount law account for statistical errors? It is mute on error. It says if there is a spread of an arbitrary percentage or number of votes that there will be an automatic recount and that that the results from that recount will be legally certified. We don't report election results the way that people report pulls. We don't say so-and-so won with 53+/-1 %.
Fine. The spread is most likely not statistically significant. Election laws don't account for statistics--they don't care about them. Dino won. That doesn't mean that statistics don't exist or that you can't apply them to this case--it just means that the state doesn't choose to do this when they put someone in office. This is why who is put in office isn't always what eh will of the people would have otherwise had. The will isn't perfectly communicated through the counting of votes.
Show me a law that says the governor will be the one who the people want. There is none. It says that it will be the candidate who has the greatest number of votes in the final count. Given that counts aren't perfect, there is a difference here.
I don't hate democracy. I love democracy so much that I want the will of the people to ring loudly and clearly over and above the din and noise of fraud and mistakes. Look--the spread changed by 219 votes between two counts. That is gr
Right--but it isn't a forgone conclusion like you initially implied. The first recount was a forgone conclusion. This one will need to be paid for.
While I agree that there is a scientific concept of margin of error, it isn't legally recognized in an election.
This was the machine recount. If the initial numbers had gone this way, there would be a mandatory hand recount. As it is, they've already conducted the mandatory recount & any additional recount will need to be petitioned for my the participants.
I agree it is well below the margin of error. But that doesn't matter. Scientists have a number, an error, and a unit. Everyone else (including politicians who write the laws) doesn't use "error" and often neglects "units."
Yes, the one with the most votes legally wins the election. But if you think that the margin of voter error is larger than the marger of victory then no, you can't say that the choice scientifically reflects the will of the people (or even of the voting people).
I'm such a nerd. I really think that elections should have margins of error and something should happen if candidates don't win by at least that much. Something tells me that 0.0015% quantifies as sufficiently small to be below it.
Oh well--I really do think that this reflects the will of WA voters (and I am one). We didn't have a strong preference. Approximately 50,000 (2% or 1000x the margin of victory for Dino Rossi) more people voted for president than voted for our governor. Our divided state house & senat also bears out how moderate we are.
I agree that this study is rather dubious, but I also think that winning the embedded market does win desktops with it. So how should Free OSs exploit this? You can make something as "cool" as a zaurus, but you will only attract the uber-nerd. What kind of embedded devices can be made with elegant interfaces like the ipod with great support on Free OSs? Should someone be pushing for an embedded handheld linux video player?
In case anyone is interested, his other two projects are:
Optimystic--CD creation for GNOME
DrivelLiveJournal client for GNOME
Seeing as how OS X is not linux and many major components aren't free in any sense of the word, I see a difference.