I'm inclined to think most illegal votes came out of the woodwork because they hated bush- but a net side effect was a thrown gov. election.

If there is a revote- expect far less turnout. This will definately favor Rossi- as most moonbat Bush haters really couldn't care less about who the Governor is-

Then why did your fancy east coast lawyers and USC experts do such a poor job of showing this to the Judge?

What other possible explanation is there? They never say how else this distribution of excess ballots in Gregoire precincts and loss of ballots in Rossi precincts could occur.........

The net is worthless without the breakout. Netting to zero variance would mask some branch with embezzlement....

Even if the winner of a particular precinct is irrelevant (and I don't think it is, because I think more heavily Democratic precincts are more likely to have Democratic favored fraud, just as more heavily Republican precincts are more likely to have Republican favored fraud) you still end up with some precincts that have far more discrepancies than others. If nothing else, this should providing a starting point for closer investigation as to what happened in those precincts. Saying that the winner of the precinct is irrelevant doesn't in turn make the discrepancies irrelevant. They should still be investigated.

Based on Precincts:
95% Confidence Interval--[189 to 238]
99% Confidence Interval--[181 to 246]

Based on Voters:
95% Confidence Interval--[228 to 280]
99% Confidence Interval--[220 to 289]

I think the comparison by voters is the better comparison than the comparison by number of precincts.

Again, 121 is way, way, way outside any of these confidence intervals.

Bill H

"...there is no logical reason to relate the absentee accounting discrepancies with who won the precinct. Which candidate won the precinct is irrelevant..."

Nonsense. If the distribution of error is significantly different than would be expected by random chance, it is absolutely relevant to ask why.

I understand whay you would prefer that the question not be asked, but that's your problem. It has nothing to do with logic.

The only remedy available to the court without becoming an activist court, is to simply "install" Rossi Governor. That just isn't going to sit well with the people out here. Not this Democrat anyway.

It is my understanding a re-vote just isn't possible under the law.

I say, pass the gas inititive to cut off their funding and throw out every democrat or republican that is siding with king county at the next election in 2006. It's only a year away.

Does the presence of the apparently voterless ballots predominantly in precincts won by Gregoire make it appear that she got the benefit of those questionable ballots? That may be as far as it can be taken when dealing with secret ballots.

Without knowing where they came from or how the votes on them were marked, it's impossible to say with reasonable certainty how they affected the election's outcome.

Their presence makes it questionable whether we have a full, true and correct count of the legal votes in King County. But, that may not matter under our election contest laws.

Okay, I'll address that. I can't offer the correct explanation for the bias, only that the pattern cannot be explained by chance. Just because one can't propose a logical explanation does not mean that a logical explanation does not exist, as Westneat suggests.

But let me suggest one logical explanation -- some of the discrepancy is attributable to innocent or previously explained circumstances e.g. some of the ballotless voters are attributable to voters who did not enclose a ballot in their security envelope. Some of the uncredited ballots are attributable to federal write-in voters. These discrepancies should be more or less evenly distributed. Some of the negative discrepancies are attributable to the 96 recently discovered uncounted ballots. Those are known to have been disprortionately clustered in Rossi precincts.

If, let's say, an overzealous election worker chose to throw a few extra ballots on the pile, she might have chosen to stuff them in Democrat precincts or legislative districts, under the false assumption that they wouldn't be as noticeable. That's not very clever, but that could explain it.

By the way, I think I would be similarly curious about such a non-random pattern that appeared to favor Rossi. The real point is that King County failed to follow the law on ballot reconciliation and that's why we have this discomforting surplus of ballots and the weird patterns to begin with.

He is? He voted for Rossi, so by definition then, as a partisan commentator, he'd be making statements that were biased to Rossi.

A court that throws out an election and installs Rossi isn't an activist court? I'm quite sure a lot of people would think it is.

If you want to commit fraud for your candidate during the counting process, the simplest way to do it is to pick a precinct which your candidate is likely to have carried heavily, take a random handful of ballots which have already gone through the counting machine, and put them back in the pile to be counted. This is much less risky than hauling in a pile of specially-marked ballots from the outside and spreading them out randomly between precincts.

Double-counting happens by accident (in which case one would expect a random distribution); there's no reason it couldn't happen on purpose (in which case one would expect more errors in the precincts which went heavily for one candidate).

Would the proper remedy be to: (a) have a revote, (b) accept all the ballots as recorded, with the effect that Gregoire won by absurd millions of votes, or (c) disregard all votes from King County?

The current situation isn't quite that extreme, but if those options would have been possibilities in the extreme case they should be regarded as possibilities under appropriate circumstances.

---
(1) RCW 29A.68.050: "After hearing the proofs and allegations of the parties, the court shall pronounce judgment in the premises, either confirming or annulling and setting aside such election, according to the law and right of the case.

"If in any such case it shall appear that another person than the one returned has the highest number of legal votes, said court shall declare such person duly elected."

Note that the court first annuls the election if "the law and right of the case" demand it, then if "it shall appear that another person than the one returned has the highest number of legal votes," the court awards the office to the plaintiff. Clearly the court is authorized to do the first without doing the second.

(2) Washington State Constitution, Art III, sec. 10.

Rossi could have won precincts where the box was stuffed- does this mean they were stuffed for Rossi? -no, it could just as well mean that the cheaters efforts weren't enough to offset legitimate voters and the precinct was sloppy in accounting + accepting provisionals. It is a stretch to say the sloppiness reflects how the majority in a precinct voted. However the existance of a pattern is too disturbing to ignore.

I agree with the analyst referenced earlier that the evidence distingusing bungling from willfull neglect from total fraud has not reached the clear and convincing fraud point- I'm guessing this is where Bridges is at- though the line between the three is pretty tough to see in KC.

In his comments though- "what voters should expect to see in an election contest" he is really setting the table for setting ANY election aside that can't reconcile the #voters, #ballots, #errors to well within the margin of victory.

This is as it should be- whether the winner is Republican or Democrat- the books need to balance to significantly within the margin of victory.

"...whether the winner is Republican or Democrat- the books need to balance to significantly within the margin of victory."

I'm sure that most of us posting here would prefer that Rossi was in the Governor's Mansion, but what this is really about is ensuring the fairness of the system, to be certain that no one, or group, can ever "game" the system to their advantage. It really isn't about who won, but how. We want to be confident that our votes do matter, otherwise what is the point?

When all is said and done, if we still have Gregoire in the Governor's Mansion, but we have a vastly improved election system, I'll be happy. Because at least then I'll know that we have a fair chance of getting someone else in there the next time.

Yes, it is somewhat comforting to mire ones self in statistical bliss when of course you cant "prove" fraud. But when the irregularities clearly prove that the margin of error exceded the margin of victory, the court should set aside the election and a re-vote should take place. The GOP has met the burden of proof in that regard.

If people find this hard to take, I refer them to an incident recorded in John Fund's book in which poll workers delivering completed ballots to the counting station took the long way to get there so that others had time in the back of the truck to stick nails in stacks of punchcard ballots. They poked out all the chads marked "Gore" so that if the person had voted for Gore, no change was made, but if the person had voted for Bush, the ballot was marked for both Bush and Gore and would be discarded as an overvote.

We should all consider ourselves morons I guess for thinking that there were legitimate reasons to want to see election reform be the central theme of the post election debate, but instead, as unequivocally pointed out by the observer, this court case was never about election reform, and election reform will not occur because of it, but in spite of it.

Meanwhile Andy states, "whether the winner is Republican or Democrat- the books need to balance to significantly within the margin of victory". While this has a happy shiny sheen to it as statements go, the reality is that it is not a practical one. If the vote difference is 100, but discrepencies equal 101, do you redo the election? What if the difference was 10 and the discrepency was 11? Or 129 and 130, for that matter. These would be infinitesimal margins of error in a state with 3 million voters, and the accuracy rate required to meet the goal that discrepencies were less than the vote difference would be next to impossible, and therefore not a practical standard.

No meaningful reform can occur until the dems lose their power base in the legislature. They have demonstrated no interest or committment to reforming the system from which they've benefited.

For all the good our hand-wringing does now, we might as well be Cubans in Miami dreaming about what we'd do when Castro is overthrown. Even when the citizens pass an initiative, the dems have no trouble voiding the law...see I601.

What we're left with is trying to boot 'em out, fighting an election system that they control, and expensive and time-consuming initiatives to restrict them.

And find good candidates to defeat them on issues.

Today's Jonny Cochran 'if the glove doesn't fit' award has to go to Foreman vs. Handy. But here is a cross I would like to see to the SOS.
1) What do you mean by "voter intent state"?
2) Which section of the RCW is that from?
...Oh it's not from an RCW?
3) where does this come from?
4) Is it OK for the SOS to make up laws not enacted by the legislature?
5) Assuming that we are a voter intent state, how would the election be different?
6) Would it have been wrong for the KC to submit a fraudulant reconcilliation report?

Yes, Yes, and YES. Either that or flip a coin or play a hand of p0ker for it. These are all "fair" solutions when compared to installing a Governor who is a lame duck before the term even starts. Because we can all see what that gets us.. A "Governor" who has so little hope of re-election that she will ram through massive tax increases in a state where the people have uniquely stated in no uncertain terms how they feel about government stewardship of their money.

It may seem fine and dandy to you now, but remember it when the situation is reversed, that you argued against those that wanted to correct the problem.

RR

I disagree that there is some maximum level of accuracy, below 100%, that we must accept in elections because 100% is simply too hard.

I will conceed that obtaining 100% is significantly more difficult, time consuming, and expensive than say, 99.8%. However I think when the margin of victory requires it, we ought to go the extra mile to obtain it.

For example, we have an automatic trigger for a machine recount, and that's whent he difference in votes between the top two candidates is less than 2000. If the difference is greater than 2000, we are willing to accept some small number of mistakes, because they aren't great enough to change the outcome. However when the difference is less than 2000, state law recognizes that a higher level of scrutiny is required.

I would take this a step further and require that the margin of error in any election be less than the margin of victory, otherwise the election is uncertifiable. That means the state needs to have contingency plans in place, like the automatic recount, for those rare occasions when the margin of victory is this small. We don't need 100% accurate elections all the time, because most of the time the margin of victory doesn't require it, and it would be a massive waste of time and resources to do it for every election. But for those times when we need it, we ought to be able to obtain it, given that we are prepared for the time and expense it will involve.

Also consider how long it took to determine the margin of error. We cannot have elections that take months to certify.

There are things that can be done to improve the future accuracy of counts, but any election of this magnitude simply cannot be as perfect as you would desire, so you have to account for that reality in the laws you make.

So we should probably have laws that require counties to take reasonable measures to ensure the integrity of our elections. Oh, wait.. we DO.
And when the laws aren't followed? Then we just throw up our hands and say "oh well, we tried?"

‘The absentee ballot reconciliation discrepancies may not be evidence indicating that fraud intended to benefit one candidate or the other occurred.’

Alone, it may not but, with the accompanying testimony explaining it and the knowledge that those testifying should have, it may well indicate fraud. And our laws, as incompetent as they are, do allow for circumstances less than fraud. Judge Bridges has already acknowledged that the legislature has not contemplated these current circumstances. In other words, we have no idea how he perceives this situation.

‘Does the presence of the apparently voterless ballots predominantly in precincts won by Gregoire make it appear that she got the benefit of those questionable ballots?’

Maybe not by themselves, but with the other circumstances involved, the probability may be different.

‘That may be as far as it can be taken when dealing with secret ballots. Without knowing where they came from or how the votes on them were marked, it's impossible to say with reasonable certainty how they affected the election's outcome.’

Again, Judge Bridges has already acknowledged that the legislature has not contemplated these current circumstances.

‘Their presence makes it questionable whether we have a full, true and correct count of the legal votes in King County. But, that may not matter under our election contest laws.’

And again, Judge Bridges has already acknowledged that the legislature has not contemplated these current circumstances. You have turned into a pessimist without sound reasoning. No one knows what he is thinking. But it is clear he sees problems.

You are attacking a fellow D (who'dathunk?) I read his statement that installing Rossi WOULD be activist. Which is why he followed with "I say, pass the gas inititive to cut off their funding and throw out every democrat or republican that is siding with king county at the next election in 2006. It's only a year away."

He specifically wants to avoid an activist judge. Personally I agree that we do not want activist judges on either side, but there would have to be some way of reversing a 'certified' election when it was wrongly certified (based on certifying numbers known to be wrong) If not, that gives the SoS far to much power, as he could certify anyone he wants to and totally disregard what the actual vote was.

Look--seriously--forget the fine print and statistical gumbo. Step back. Big picture. This whole WA voting process and event is a mess. The statewide electorate machine is critically ill. Stop focusing on the myriad of exotic mico-symptoms. Fix it. Take lumps. Fire people. Enact reforms. Voters will respect the tough decisions.

I side with the other blog commentator on the newspapers' (journalists') failures to be the 'investigative bloodhounds' of not too long ago. Why the limp posture? The public needs hard-hitting articles, not love-fest excuse diversity sessions.

Olive pits, my friend. Grow them or cast them into the yard-waste bag of journalistic mediocrity. Political parties aside, the public wants true house cleaning, not excuses. Want to sell more papers? Be a tick on a hound.

If races continue to be this close, then accuracy must be an absolute, not a luxury. Bridges needs to lay down the law on what it means to hold elections and to vote in America. This applies to Florida and Ohio too.

It's sort of a stupid question to ask the day after memorial day. How many Americans died so we could have this right? Saying it's too hard is probably the worst disrespect you could give to any veteran.

"It's sort of a stupid question to ask the day after memorial day. How many Americans died so we could have this right? Saying it's too hard is probably the worst disrespect you could give to any veteran."

Well said.

Mark

It is possible to accurately measure things in the millions with an accuracy in the ones. I've done it personally and been an independent audit of my data, as well as an extended (and repeated) data analysis indicating no biases in the raw data. (Similar to Stephan's 'quartile' approach, but on 8 different nominally non-participant variables.)

I, for one, would be perfectly content rerunning any statewide race closer than 500 votes - there just aren't that close. This was the closest _ever_.

It is tougher when there are more bodies involved, but the VERY minor change of having one 'fake' race pre-marked between the participants 'poll', 'provisional', and 'absentee' would wipe these 'stuffed into the box' errors out in one go.

The main reason it has taken _us_ months to figure things out is that WE don't (or didn't) have access to the raw data. Even with the discovery process, it isn't quite the same thing as being the people in charge of collecting and filling the data in the first place. Stephan's already got better analysis tools than appear to be used by KCElections.

Oh, and if you billed the county with the highest _per_capita_ error rate for any revote, you'd see the error rate plummet.

HOWEVER, the fact remains that if “accounting errors” were purely random, we would not have such a glaring bias towards extra votes in Gregoire precincts and extra credited voters in Rossi precincts.

Now, it is clearly a correct statement to say that the statistical results do not tell you what happened. They don’t tell you WHY there is this bias. The statistical results simply show clearly and incontrovertibly that there is a bias as described above.

After seeing this clear bias, you can either ignore it or you can come up with theories as to why it exists. This is what people here have been doing. Of course, most people here support the GOP, and are extremely suspicious of the Democratic machine that operates in King County. So, most of the theories that people come up with here are along the lines of fraud of some sort.

But, anyone can come up with theories. The problem is, nobody else is coming up with plausible theories as to why this clear statistical bias exists. And, I should point out, “sloppiness” is not an acceptable theory, because sloppiness is generally random, while the data shows that the extra ballots and extra voters credited is definitely not displaying a random patter.

This, to me, is what is so disappointing about what has occurred. There is nobody with oversight responsibility that is objectively looking at this data and thinking, “Hmmm … this is a very strange pattern in the data, I think we should investigate.” Instead, the SoS and his pal Handy started from the very beginning pulling the wagons in a circle and defending King County.

So, what are we left with? We are left with the loosing party looking at the data, seeing a highly improbable pattern in the data, and saying that it must be the result of something deliberate.

Of course, leaving the investigating up to the GOP alone is not the way to go about it. Of course they would jump to conclusions, because they are trying to win, and nobody else in the state is coming up with alternative theories that are plausible.

If Westneat and others really were interested in finding out what happened, and really are as objective as they claim to be, they would have seen this highly unusual pattern in the data and began their own investigation from an objective point of view. It would have been a great service to the community to do so.

Danny - let's turn your question on its head. What is the logical reason NOT to relate the accounting discrepancies with who won the precinct?

It is merely a data set. You can take a data set and run all sorts of analysis on it in order to see if you find a pattern in the data. This occurs all of the time in scientific endeavors, marketing analysis, econometrics, etc ... This is the sort of process, for example, that is used to find potential therapeutic benefits of chemical compounds. In marketing, this is often called “data mining.” When mining the data, you are not necessarily relating on variable to another because you have a reason to; instead, you relate all data to all other data and look for patterns of behavior.

This is all that Shark has done. He has found a very strange pattern in the data. One would reasonably expect that, if these were random accounting errors, the results of this analysis would have shown a random pattern. There is not a random patter. Now, it would seem, it is up to someone to come up with theories as to why this strange pattern exists.

You are right that Republicans are the wrong people to do this. But, who is going to? I think that fact that nobody is doing is the major frustration here. In the absence of anyone objective looking at this and trying to figure out why this odd pattern exists, the Republicans are left to do it, along with their very obvious bias.

So, again, I ask – please give us a logical reason why NOT to relate this data?

We have come too far as a nation and paid too high a price to vote (and sell this "freedom idea" around the world) to let this slop stand as the pride of our electoral system or something "...any bank would envy..." (quote attribution to Mr. Sims, [a.k.a. "Sims who swims & votes with those who swim with the fishes])

Ok, now try ensuring perfect accuracy when thousands of different people are counting thousands of subsets of data in dozens of different ways. Hardly the same thing.

Saying that a certain error margin will always be acceptable, regardless of the victory margin, is an open invitation for officials to commit fraud up to the 'legal limit'. Further, requiring that a revote be performed when the error margin exceeds the margin of victory is not particularly impractical; while such situations will occasionally arise even with well-run elections, victory margins will seldom be below the error margins of decently-run elections; having a revote once or twice a century will only add 1-2% to the costs of an election.

Al: Oh, and if you billed the county with the highest _per_capita_ error rate for any revote, you'd see the error rate plummet.

I'd like to see the revote costs assessed to counties in proportion to the absolute number of errors therein. This would create incentives for people in any county to seek out errors in other counties, so as to dilute their own share of the costs, and incentives for counties to prosecute mischief.

Key observation: if 66% of a county's people support one party, officials who 'fudge' things in favor of that party without consequence may be supported in that effort by a majority of their constituents. Officials whose 'fudging' causes the election of their party's candidate to get thrown out, however, will be supported by 0% of their constituents. What better way to discourage mischief?

"Ok, now try ensuring perfect accuracy when thousands of different people are counting thousands of subsets of data in dozens of different ways. Hardly the same thing."

Ohhhh, it's sooo difficult counting ballots. Soooo hard.

I remember after the first Presidential debate - all the liberals were making fun of President Bush's statements about the difficulty of being President. Where are they now? It's more difficult to count ballots than to be President? President Bush has an easy job, but King County Elections workers are overworked??

Al is absolutely right - we could take the margin of error down to a manageable margin, so if we re-ran every election that was within that margin it would happen every few decades at best. Most likely we'd re-run an election every 50 years or so.

But since you now want perfection, will no incremental change suffice? Why, after this mess, are you shooting for perfection? Is it to keep us from changing at all?

You bring up a very good point. The system as it currently exists is overly complex, and sorely lacks uniformity. We didn't have one election on November 2nd, 2004 in the State of Washington, we had 39 elections, because each county did it differently, with different methods, rules, and standards. That's one of the first things that needs fixing. One standard for the entire state. It actually isn't really all that important which standard, as long as it is uniform, it would make finding discrepancies much easier to find.

I draw an analogy to my days in the Army. When we had a barracks inspection, it was best if roommates divided the work such that one soldier arranged both sock drawers, and the other soldier arranged both field kits, etc. The idea being that there is less chance of a discrepancy because the same standard is being applied to both. I might roll my socks to be 3 inches wide, my roommate might do his to 4 inches. If either he or I do both sets, they are more likely to be identical.

If the same standard is applied statewide for elections, it would make finding and correcting mistakes much easier.

Besides, just because it's too difficult with the current system doesn't mean we just say "Oh well, it can't be done". If we want it bad enough, it can happen.

Accountability.

Let's face it. A "citizen" of this state should be able to follow simple written instructions, i.e, how to fill out your ballot, mail it (if needed) on time, sign name, etc.

If they are too stupid to figure it out, their vote should not count in the current election.

While that may seem cold, I think it is logical that someone that impaired really has no valid opinion on current events. Rare exceptions could be accommodated.

Given that horribly unfair requirement to be a legal, registered, voter, AND reasonably aware of your surroundings, I think the vote could be 100% accurate without much trouble at all. All it takes is a specific standard to be met and the effort to count accurately.

My kids could do it. There is no excuse for the state, regardless of county.

But that's just my opinion......

Could it be Nick was hoping to have a "specific target number" for KingCo to exceed if KingCo were the last County to report??

Just asking.

Just asking Danny.

Did you even think to ask Danny?
Why don't you ask Mattingly now Danny? WHO requested that blind support letter?

Just asking Danny.

Actually, I thought about it overnight, and the recounts would make double-counting a less likely solution. In the machine recount, you'd end up with substantial discrepancies in precincts between counts, and in the hand count, double-counting on any significant scale would probably be impossible.

Thakns for revisiting the statistical analysis on the extra-vote and mising-vote data. I'm not in a position to dedicate a whole lot of time to this, so I am kinda slow on the analysis side, but I did start running my own "analysis" of the numbers yesterday and did the same precint level comparion of extra-vote for Gregiore verses Rossi as well as for those that were splite evenly. Here is what I noted that has not been listed in your most recent tables:

Precint For: #Precints CumNet VotRat
Gregoire 1875 680 0.3627
Rossi 693 121 0.1746
Neither 48 7 0.1458

#Votes CumNet %ofTot
334412 680 0.20%
229794 121 0.05%

The VotRat line shows that the rate of cumulative votes increase per precint in Gregoire precints is more than twice as high as in Rossi precints or in precints that were split.

Additionally, the cumulative net increase in votes for Gregoire compared to total votes cast was 4 times that for Rossi.

I don't know how to calculate the odds of that level of discrepency on such a realtively substatial data set, but I would guess that it is a very small number.

In regards to the various other ways you DID calculate odds, I see that even in the best of worlds, there was 7 x 10ee-6 chance that this would occur. There simply is no way someone can't come to the conclusion that these irregularities were not the result of intentional vote biasing.

Thanks for helping me to reconcile the facts.

Happens ALL THE TIME in millions of different situaitons. Banks do it. Even Costco Warehouses do it. Someone is charged with a set of data (a till), they handle the in and out flow of the data (money), and they reconcile the balance at the end of the day. There is consequnces for consistent failure to do so. There are proper procedures to follow. They know what they started with, they know every change of data as it occurs, and they know what they end with. They handle hundreds of changes a shift, and there are scores of shifts per day, every day in thousands of warehouses across the country. I can bet that the error rate in one year for Costco cashiers is less than King County's one election.

If being accurate mattered, they would make it add up. Supervisors would double check every "counter's" tallies as they processed batches. There would be consistent and thorough end-of-the-day audits of all batches processed. It is really very basic stuff.

The problem is, it never mattered. There were no expecations of accuracy in the past, no one's job was dependent on accuracy, and no one checked to see if accuracy were assured.

That should have never happend. People had a responsibility that they did not do. Look what it has cost us. Time, money, good will, faith in they system.

Don't take the easy way out. Demand it be corrected and demand that in the future our public employees (who work for us) be accuntable for their responsibilities. Management is the cause of nearly every operational failure, and we are the management If you don't want to manage, then at least stop getting in the way of those who do.

So let's start simply by ensuring that counties track the number of ballots received and reconcile this number with the total number of ballots counted plus ballots rejected. Let's also force counties to make sure that every ballot cast (including poll-site ballots and provisionals) is matched to a voter signtaure. Counties should also take steps to try to ensure that each signature is matched to a ballot, but this will be more difficult. Ideally, every person issued a ballot at a poll site should have to account for it before leaving.

Apart from the latter, these things should be possible to do with near-100% accuracy. As Eyago points out, lots of businesses do the same thing every single day as a matter of routine.

it would be interesting to see what you get if you retry your analysis with the so called "undervotes" -- you know, the precincts in which there are more voters recorded as having voted than there are votes."

So why doesn't Westneat do his own friggin analysis? Why does he rely on the Shark to do it for him? Doesn't Westy understand statistics? If not, why is he even commenting on it? I could care less how he "feels" about the numbers. The great thing about Sharkanski is he provides NUMBERS to back up his assertions. Actully he has his numbers in line "before" making any assertions, unlike Westneat and every other big mouthed lib who comes on here or claims to be an "editorialist" for the Seattle MSM rags.

As a mathematician, I share your interest in statistical patterns. But you have failed to propose, let alone prove, a plausible theory for the pattern here.

- Stefan admitted that there would be no point to putting fraudulent absentee ballots in D precincts (though he says it "could" have happened anyway ... uh, so?). Frankly, I think you could make an equally good case for putting fraudulent D ballots into an R precinct to draw less attention in the totals.

- ScottM suggested deliberate double-counting, a clever theory, but then admitted that this would have been caught by the evil recounts.

Your numbers also provide evidence against fraud. While there were more net overvotes in D precincts, there were still significant numbers of overvotes in R precincts and undervotes in D precincts. If overvotes and undervotes are evidence of fraud, then how do you explain these?

Your spreadsheet showing the count by precinct was telling, too. We're talking a handful of overvotes and undervotes in each precinct, not hundreds. This is not the typical footprint of fraud. (It doesn't disprove fraud either, but when you're trying to void an election due to circumstantial evidence, you actually need some circumstantial evidence.)

You ask what other explanations are possible. If overvotes are caused by voter sloppiness and by witness protection folks, is it possible that people in D precincts are more likely to be sloppy and/or witnesses?

Revotes have practical drawbacks that traditionally are considered serious. We could deal with those drawbacks if we wanted to, but recognize that this hasn't historically been our society's choice.

You are right there is no reason from a standpoint of simply having an objective of increasing the votes for Gregoire. However, activists for Democrats were probably more likely to be working in "friendly territory." Just a thought.

Actually Larry, if you were paying attention you will see that on numerous occasions here and previously I have NOT been asking for perfection, and in fact declaring that such a standard is not practical.

And, BTW, yes, when you have thousands of people counting ballots you will get errors, not because counting ballots is hard, but counting them without making errors is hard.

Bostonian, motherhood and apple pie are the core elements of our life. We either protect those or we live in motherless hunger. It's that simple.

Bottom line is errors are always going to occur, so we need to determine how to minimize their impact and what procedures are acceptable for catching them and perhaps rectifying them if there is still time to do so before an election is certified.

Which is why the Founding Fathers invented the Electoral College. Or was that because they didn't think the decision of electing a President should be left to the uneducated populace directly? But I digress...

Certain types of errors are probably unavoidable. If someone decides that he'd rather vote in his mother's legislative district than his own, so he registers to vote from her house even though he doesn't live there, his vote in that district would be fraudulent (even if he only votes once) but there would be no good way to protect the system against that sort of thing without adding some extremely intrusive laws to monitor how often people sleep in which dwelling, etc.

In King County, however, there seems to be no interest in preventing fraud or mistakes. Even the most basic check in an election--ensuring that the number of voters equals the number of ballots--seens to be regarded with disdain.

A convenience store clerk should not be fired if a man with a gun is able to come in and rob the place. On the other hand, a convenience store clerk should be fired if some cash is found to be missing after he sets his cash drawer on the counter in front of a 'friend' and goes to the rest room.