Much fun has been made of Ron Sims' claim at his Wednesday press conference that King County Elections was "99.98% accurate" which Sims also believes is an accuracy rate that any bank would envy.
As noted, I corrected Sims during the conference to explain that given the numbers he used in his calculation, the correct accuracy rate is not 99.98%, but 99.8%. [I don't accept, by the way, that King County elections was even that accurate, only that 99.8% is the correct percentage given the raw numbers they're using in their claim of accuracy]. The prepared text of Sims' remarks posted on his website wa corrected by the end of Wednesday to read "99.8%". If for nothing else, Sims should be commended for making that correction.
Thursday's P-I had this headline: "Sims defends King County vote as 99.98% accurate". The headline must have been written by somebody other than the reporter, Neil Modie, who happened to be sitting next to me at the press conference and mentioned the correction:
election critic and conservative blogger Stefan Sharkansky pointed out that the executive stated the accuracy rate inaccurately -- that it actually was a slightly lower 99.8 percent.The word "slightly" is debatable. By way of comparison, SeaTac airport had 358,894 "aircraft operations" (take-offs and landings) in 2004, or about 1,000 a day. 99.8% accuracy would mean two crashes every day. 99.98% accuracy would mean a mere six crashes every month.
Some of Sims goofier defenders argue, for example, that he might have simply misspoken, after all, his prepared text on the web reads 99.8%. But, no, he didn't simply misspeak, his original prepared text clearly says 99.98%. And this miscalculation isn't an isolated incident. An earlier King County Elections press release, based on subsequently revised numbers, incorrectly calculates that an error of 1,217 out of 899.199 represents ... 99.99% accuracy (the correct number would have been 99.86%).
As to the argument that banks would envy King County's accuracy, I've been amused that talk show host Dave Ross and various lunatic fringe webloggers are coming up with statistics to "prove" that Sims was right about the inaccuracy of banks. Yes, I'm sure it's easy to find all kinds of meaningless statistics from the banking industry that can be spun to make Ron Sims look like an accounting genius. e.g. "Only 97% of bank wastebaskets in Fargo, ND are emptied once a week, so King County elections has 30 times fewer errors" or something.
But I think most normal people will admit that if a bank teller counts her till 3 times and is still off by $2 per $1,000, somebody will lose their job and/or go to jail. Unless they work for the aptly named "First Sims Bank"
Posted by Stefan Sharkansky at February 11, 2005 10:57 AM | Email ThisThe people don't want a 99.8% or 99.98% accuracy. We want 100% accuracy, and we want every one of the voters to be legal. And alive.
We shouldn't be in this mess in the first place. The elections should've been run in a way that if Gregoire won by just 1 vote, the state would overwhelmingly agree that she is the rightful governor.
Posted by: Jonathan Gardner on February 11, 2005 11:13 AMI, though, do not understand why the results should not be 100.00% accurate, the same as any viable bank - which Simms seems to be making as the measuring standard.
Posted by: Jonathan on February 11, 2005 11:31 AMKing County's result of 99.8%, by definition, is therefore fully acceptable and within the legal confines of a legally accurate election. Anything that is 99.75% or better meets the State's standard for expected accuracy.
Posted by: Nelson on February 11, 2005 11:37 AMThe point is that the margin of error is greater than CG's margin of victory. Therefore the election result is illegitimate. It's unconscionable that the Democratic stance is that CG may not have won, but we should just accept the results anyway because she won by "innocent errors" and not deliberate fraud.
Why kind of democracy is this!?
Posted by: Colin on February 11, 2005 11:38 AMIt's only linguistic contortionists like Mr. Goldstein that can take a quite obvious concept such as the general accuracy of banks, and spin a few random web examples in to a defense for the general innacuracy of the November election.
But this never passes the smell test where the rational are in attendence.
The average rational Washingtonion is not impressed with the Goldstein et. al. involutions and neither will be Judge Bridges.
Posted by: Jeff B. on February 11, 2005 11:39 AMI'd like to see where you come up with those numbers, (on second thought, no, keep your pants on, please).
It sounds like you're mixing together different concepts to come up with the results you want:
While "the public" may be happy with results that are less than 100%, if there are errors made IN COUNTING that are above the margin of error and that can be corrected by a recount, then a recount will fix them.
The problem is that there were errors made which had nothing to do with counting. You can count a million times, and if you can't account for the fact that more people voted than signed the voter roles (IE ballotbox stuffing), recounting isn't going to fix that (and that's only one issue that's been brought up). Fraud? Incompentance? Apathy? It doesn't matter - when the error is larger than the margin of victory, then the outcome isn't known.
Posted by: thecomputerguy on February 11, 2005 11:49 AMLike counting provisional ballots without verifying them first.
Like marking ballots in way that the ORIGINAL marks can no longer be reviewed.
Like ending up with more ballots than people who voted.
Yeah, errors...............
Posted by: smoke on February 11, 2005 11:55 AMUnfortunately for him, some folks make their living crunching numbers and understand that a bullshit factor is not a precise measurement. And when some of these folks guffawed when they recognized the BS factor, he was miffed and scolded them for failing to conform to his parents' high standards for home training.
Had he followed those high standards himself, he'd have put Dean Logan to a merciless grilling about the miserable procedures followed by King County in its document handling and vote accounting. He still hasn't realized that even a 100.000% accurate count of votes that outnumber voters by 1800 equals one botched and indeterminate election.
Posted by: Insufficiently Sensitive on February 11, 2005 12:15 PMThey love how "Enron Math" allows them to manipulate public finances and how it also allows them to manipulate vote totals.
We all owe a great debt of thanks to Aurthor Anderson Accountants and Enron officials for their giving us a new lable on an old Democratic Party method of operation.
No wonder it is the Democrats who as so opposed to WASL math standards being enforced. If we educate our children, then such explanations and calculations 99.96%, 99.98%, 99.8 percent----oh, its just "good enough for government work" Let's move on beyond this election. (NOT!)
Posted by: Bob on February 11, 2005 12:19 PMThrough my various careers I have had unique opportunities to mix and mingle with all sorts of pols. Most, from all sides of the spectrum, have had intangibles of some sort, an aura if you will, or a certain comportment.
I vividly recall just how unimpressed I was with Ron. Very, ummm, disingenuous comes to mind.
And you know that radio commercial about increasing your vocabulary because we are judged by the words we speak? Well...I did come to a judgement: What in the @#$% were the voters of King County thinkin????
Posted by: SnoCo Voter on February 11, 2005 12:22 PMToo late you already asked. From the Secretary of State site (http://www.secstate.wa.gov/office/osos_news.aspx?i=jrVma9RNpBM06%2f2DSMEUBA%3d%3d):
"Hand recounts are required if the difference between the two candidates is fewer than 150 votes and less than 1/4 of one percent."
less than 1/4 of one percent means less than 0.25%. And this is statewide, not just for one county.
I do agree with Mark's statement that we should expect better than what we had, and strive to improve the error rate in the future, which is why I support election reform.
Oh, and Shark, if you're including me as one of the "lunatic fringe webloggers" who come up with statistics to prove things, well I guess you would know wouldn't you.
Posted by: Daniel K on February 11, 2005 12:38 PMPlease try reading next time, huh? There is nothing there about "what the public expects". It merely says that a recount is required if the margin is not more than this. The part about the public expecting 99.75% accuracy is something that somebody read into that link... unless you have something else?!?
In actual fact you have Shark to thank for making such a big deal about the 99.8%, although he was correct to point out the number was 99.8% and not 99.98%, and it is good that Sims' office also corrected the mistake in their document.
While I believe it is important to have quantifiable numbers, I agree with Shannon that the focus should be on determining what are the problems and what are possible solutions.
Some people at the Vancouver Secretary of State meeting correctly pointed out that they need to not only enact changes, but they need to ensure existing laws are better managed, such as how we ensure information about felons and dead people is captured and used to ensure the correct registration of voters and mail out of ballots.
Posted by: Daniel K on February 11, 2005 01:08 PMThe 99.75% is part of election law. Does that mean that's what the public expects? Of course not, because the public will always expect perfection, whether that's attainable or not.
What the law does respresent is the will of the people via their elected officials. 99.75% was a number that was selected when the law was created as a threshold. If that was not an appropriate number it should have been contested, perhaps even via an initiative.
I'll let Nelson respond for himself, but I believe that's what he meant when he stated what the public expects.
Posted by: Daniel K on February 11, 2005 01:16 PMSims is full of crap. He's covering up something. Nobody dances, sweats, and looks as guilty as he did in that press conference unless they have something to hide. It's just like the over-reaction of Clinton, when he said, "I DID NOT SLEEP WITH THAT WOMAN ...."
What surprises me is that Logan is willing to dive on a sword for these people. Gregoire definitely is not worth diving on a sword for.
But, at this point, he is too far into it to pull the sword out and hope not to bleed to death.
Posted by: DeadManVoting (aka Iguana) on February 11, 2005 01:31 PMYou'll have to forgive me for reading more logic into your post than you wrote.
To paraphrase what you said, "since the law provides for an automatic recount at .25% margin, errors of this magnitude are expected".
Here's the problems with your logic. First off, a recount only addresses PROBLEMS WITH THE ORIGNAL COUNTING PROCESS. Not other types of errors made by election officials. You can count a million times, and every counting method has a statistical amount of error built in, so the results will vary, within a range. What happened with the hand recount is that CG took the hand results and ran with them, claiming that the recount process "fixed" the errors in the machine counts. I don't even think we've even examined what the statistical error range for the hand recount is. A hand recount is likely to be as flawed or more flawed than a machine recount.
But if other types of errors are made (like stuffed ballot boxes, for example), the votes have to be removed in order for any counting process to show the totals. Incidentally if the other forms of "errors" (the term error is somewhat generous here) are larger than the error seen in the counting process, then it pretty much makes the counting process moot, don't you think?
Posted by: thecomputerguy on February 11, 2005 01:35 PMThere were 1800 more votes cast in King County than signatures.
Assuming that the rest of the process is 100% accurate, 99.8% is the best possible outcome - the real accuracy is probably even lower.
It would be much more accurate to say:
"KingCo elections was AT BEST 99.8% accurate with many issues still to be resolved that may lower that accuracy rate"....something like that.
SORRY!.
Posted by: Mr. Cynical on February 11, 2005 01:52 PMThat sounds reasonable on the face of it, but it isn't a feasible conclusion to make unless we expect and can achieve perfection.
If we assume that 0.025% is an acceptable error rate, statewide that would be 7500 votes (based loosely on a round figure of 3 mil total votes), so a vote difference of 7500 or less votes would have to be called into question.
Set it at 0.002% and the threshold would be 600.
Unless we don't allow for any mistakes we will always have the potential for margins of victory that are greater than the margins of acceptable error.
If this is the road we want to go down, we better be very careful what thresholds we choose because it could mean a lot of previously valid results would have had to be thrown out in past elections, and the effect on future elections could be very significant.
Posted by: Daniel K on February 11, 2005 01:55 PMI didn't get fired. Of course, I did get transferred to a job where I didn't handle any cash. Seems they were happy with my work overall, but those pesky problems in not balancing did bother them. Better to put me on a desk job where I couldn't screw up the cash.
BTW, when I went to my boss with the $500,000 shortage, she just laughed. We didn't carry anywhere near that much cash in the whole branch, so she knew it wasn't a big problem. Just a wrong key stroke that was caught that night by the proofers.
Posted by: north clark county on February 11, 2005 02:00 PMBut notwithstanding wrongful acts, the legislature set up procedures to do up to TWO additonal counts, one by machine and one by hand if the margin is still within that range. The presumption behind that is that recounting in two different formats is more likely to get you a final number that is closer to the truth.
It is based on logic and it was written into the law. Think about how you yourself count things, such as money or a deck of cards. Don't you first count them one way, say, face down, and then, if you're still not sure, do it face up, by suits (or denominations) or by groups of ten or any other different way.
This is the same thing. The final count is the final count. It's the only one that counts.
Posted by: Nelson on February 11, 2005 02:03 PMIts simple. The error rate needs to be lower than the margin of victory.
Otherwise its not known who won, is it?
Posted by: thecomputerguy on February 11, 2005 02:05 PMThen you have to also expect perfection, and I simply don't agree with you if you believe that is an achievable standard one can set with a law.
Posted by: Daniel K on February 11, 2005 02:09 PMPlease explain how an error rate lower than the margin of victory equals perfection. (that is what you said, right?)
Posted by: thecomputerguy on February 11, 2005 02:17 PMNo it isn't what I said, so you'll pardon me if I decline to explain something I did not say.
Posted by: Daniel K on February 11, 2005 02:21 PMIt did no such thing. It just set a quarter percent (between numbers of votes for two candidates) as an arbitrary trigger for an automatic recount.
Statistics is a huge red herring in counting votes. Properly managed, an election should be able to repeatably determine a winner by a margin of ONE VOTE. Counting of integers is a different process than a measurement of some quantity, and shouldn't be confused with measurements and their statistics.
Posted by: Insufficiently Sensitive on February 11, 2005 02:23 PMSo this wasn't your posting then?
thecomputerguy wrote, "Its simple. The error rate needs to be lower than the margin of victory."
Then you have to also expect perfection, and I simply don't agree with you if you believe that is an achievable standard one can set with a law.
Posted by: thecomputerguy on February 11, 2005 02:28 PMPerhaps, but first you have to determine which integers (or votes) should be counted, and that's obviously not proven to be so clearcut due to all the qualifying or disqualifying criteria that have to be used, and the processes behind those criteria than manage their application.
I think well meaning people like yourself or myself, once immersed into the election process would soon discover just how hard (sometimes unnecessarily, but usually due to complexities) a seemingly simple process of running an election can be.
The net effect of this total error was a change of 390 in net margin, +261 for Rossi to +129 for Gregoire, or .0014%.
The number Sims was speaking of did not relate to any of this. It was the net difference between ballots and voters credited. However, some counties had more ballots than voters and some had less. If Stefan's numbers are correct, then the total absolute difference is somewhere are 3,800. That means that 0.04% of the ballots are not associated with voters credited. This has nothing to do with the recount threshholds.
Posted by: north clark county on February 11, 2005 02:40 PM*sigh* Yes, that was my post.
You stated, The error rate needs to be lower than the margin of victory, to which I responded, then you have to also expect perfection.
I did not say an error rate lower than the margin of victory equals perfection.
But unless you have guaranteed perfection, you will not be able to eliminate scenarios when you would be forced to throw out an otherwise valid election simply because the error rate was greater than the vote difference.
Eliminating that possibility isn’t feasible, and because the error rate would be based an accepted standard, however arbitrary, one could not use it to throw out the result simply if the vote margin was smaller.
Precisely. And that's why I specified a PROPERLY MANAGED election, which would rigorously enforce the existing rules to ensure that all invalid ballots entering the system were rejected prior to counting.
Logan's "count every ballot" rabble rejected rigor as a principle in their haste to welcome ballots from any and all sources.
Posted by: Insufficiently Sensitive on February 11, 2005 02:52 PMIf Republicans were trying to win a court battle with nuance, they'd be laughed out. To the ever-so-smart trolls here, stop straining out a gnat and swallowing a camel. It makes you sound foolish, and I highly doubt that you are as foolish as you are sounding.
Posted by: dkpcowboy on February 11, 2005 03:06 PMHow many errors do you think are in those precindt poll book reconciliations. They were supposed to account for every ballot they were issued. Did they??
Let's say the errors are 1800+++++++whatever. To Be Determined!
Posted by: Mr. Cynical on February 11, 2005 03:17 PMConservatives simply take their marching orders from whatever yokel is currently shouting louder on talk radio or a blog and continue to espouse that position without any regard for logic, argument, or -- yes, I'll say the word -- nuance.
Liberals always look for grays and shadings and logic. Conservatives only know black and white and only in what they're told to say and how to say it. No individual thought is permitted.
The joke is that it's done openly and without embarrassment. Hence, all the conservatives who rush to be identified as "dittoheads" by the likes of Rush Limbaugh and his ilk. "Dittoheads" obviously means that whatever Rush says, you say "ditto!!!"
Posted by: Nelson on February 11, 2005 03:33 PMBesides pickin fly poop outta pepper, they also like to claim victory when they're just being obdurate.
Posted by: thecomputerguy on February 11, 2005 03:40 PMAt the risk of being accused of making yet another argument, I just want to say that if there has been nitpicking going on it has been by those who have been suggesting I'm saying one thing, when I saying another, or requesting me to repeatedly explain what I've already stated.
The alternative would be to ignore the misstatements, or ignore the requests for further clarification, which I guess I would do if I had no self respect, or couldn't defend my opinions.
Posted by: Daniel K on February 11, 2005 03:46 PMYou're the pot calling the kettle black. It's quite ironic that you are projecting your traits upon Conservatives and Republicans and you don't know it - the definition of irony.
You say that we take marching orders from talk radio, blogs, etc. I've never listened to talk radio, I didn't like Rush 15 years ago, and I don't take marching orders from anyone. That sounds like the typical liberal talking points from Air America, Christine Gregoire, and a whole host of liberals. So you spout their talking points, and accuse us of doing the same. See the irony?
I have yet to meet a liberal in Seattle who analyzes anything to see if it is correct or logical. I have a litmus test to determine of someone is a liberal - if they agree with King Ron Sims CAO, or see nothing wrong with it, they are a liberal.
Liberals believe that the government knows best, and should determine how they live their lives and spend their money. It's one step on the road to fascism.
Conservatives believe that individuals can best determine what path to take in the road of life. Hence, we want tax cuts and social security reform.
But when it comes to grays and shadings of logic, maybe you are correct. My favorite liberal nuance goes something like this: 'Well, sure, Saddam Hussein was a maniacal killer who committed genocide, used chemical and biological weapons against his own people, continually shot at our planes in the no-fly zones, and committed the two worst man-made environmental disasters in history - but since we didn't find stockpiles of WMDs after we overthrew his government, we had no right to dispose of him as ruler of Iraq.'
Sometimes less nuance is better. At least - that's what 61 million people believed last November!
Posted by: Larry on February 11, 2005 03:57 PMThe problem is that we do not know what the "errors," "discrepency," or "remainder" refer to. The King County election report does not clarify this, except to give suggestions about the types of errors the county experienced. For sure, the county's error rate does not include the votes of felons, or the deceased, which would increase the raw number of errors. But, are these county "errors" - or felonies committed by the various individuals? Until proscecuted and convicted, the county has to treat those ballots as valid, unless you believe that people are presumed guilty.
The liklihood is that these errors are data entry problems, and the county employees are no more likely to commit these types of errors than anyone else. Should we demand better of King County - definately.
Do we have enough information to make any informed conclusions as to whether the county intentionally commetted these errors - definately not. We do not know the type or distribution of these errors. As far as I know, we do not know completely how the error rate in King County compares to other counties. We also have different conceptions of what the error rate actually represents. Until all these conditions are met, it is difficult for me to make any informed conclusion about how the error rates impacted the election.
We have a little bit of an idea. According to info at http://alsoalso.typepad. com/also_also/ the error rate in Spokane county was twice as large as in King County.
We do a disservice to fairness and addressing the problem statewide if we do not highlight this fact and let election officials in Spokane think they aren't under the microscope too.
If election workers are dilligent in their duties, such scenarios will happen rarely. And when they do happen, a revote will do more to establish the legitimacy of the winner than would debates about how many 'magical mystery ballots' were cast for which candidates.
I would expect that if revote costs were assessed to counties in proportion to the number of 'fishy votes' found therein, such a rule would do a lot to clean up elections. People might not mind election officials who let slide some fishiness that might benefits their party, but nobody's going to like election officials whose mistakes cause them to have to pay for a new election.
Posted by: supercat on February 11, 2005 11:05 PMI should hope Spokane is under the microscope! I don't care what county it is, problems are problems. If someone tries to deny the problem, then you have a problem.
Whatcom County, where I live, had a half dozen extra votes show up out of now where during the recounts. I was extremely disappointed that they included them even though the majority of them went to Rossi. I frankly didn't understand why they felt they should be counted without any good explanation for why they were not in the first place. King didn't have good explanations for the dozens of controversial decisions they ended up making either.
Posted by: Mark on February 11, 2005 11:07 PM