Playing Russian Roulette with Ohio's 3% recount rule

[many a good man]

In the game of Russian Roulette, there are six chambers in the pistol and one bullet. You would have to be crazy to take those kind of odds. But what if the pistol had 32 or 33 chambers and still only one bullet. Would you try it then? Some of us would probably pause for a moment and ask ourselves if the reward is worth the risk...

What if the reward was wealth and power and control of the world's most powerful nation-state?

**********************


When stealing elections, there are two equally important objecives: 1) get more votes for your guy; and 2) don't get caught.

The first is child's play for a professional software programmer. The second one requires more thought. To get away with it you would have to exercise all means at your disposal to minimize the risk.

We're all familiar with Ohio's 3% recount rule. You hand count precinct by precinct until you have counted 3% of the county's total. If hand count = machine count = original official machine count then you could count the remaining 97% by machine.

The smallest county in Ohio, Holmes, contains only 19 precincts. That means any single precinct would have the required 3% of the total county vote. If a recount were ordered, you would only have to count the ballots from ONE precinct.

If you knew in advance exactly which precinct was going to chosen for the hand recount, you could hack the other 18 with ZERO risk of getting caught.

There are 14 other counties where counting only ONE precinct will get you the required 3%. For the record, the number of precincts per county to fulfill the 3% requirement are as follows:

15 counties only require hand counting 1 precinct
33 counties only require hand counting 2 precincts
14 counties only require hand counting 3 precincts
9 counties only require hand counting 4 precincts
4 counties only require hand counting 5 precincts
1 counties only require hand counting 6 precincts
1 counties only require hand counting 7 precincts
1 counties only require hand counting 8 precincts
2 counties only require hand counting 9 precincts
1 counties only require hand counting 10 precincts
1 counties only require hand counting 11 precincts
2 counties only require hand counting 15 precincts
1 counties only require hand counting 18 precincts
1 counties only require hand counting 25 precincts
1 counties only require hand counting 31 precincts
1 counties only require hand counting 44 precincts

The ONLY thing the fraudster has to do is make SURE that NONE of the precincts being hacked is going to be chosen for the recount.

There are lots of things you could do to predict which precincts will get chosen in the event of a recount. Several have been discussed on this board. An incomplete summary:

1) know in advance which one(s) will be picked by the BOE
2) you could TELL the BOE which ones to recount
3) you could predict using mathematical anaysis which ones are most likely to be chosen (which ones will yield the number closest to exactly 3%)
4) you could use your knowledge of human nature and the personalities of the BOE workers to predict which precinct(s) are likely to be chosen

From this you could develop a list of "safe" and "unsafe" precincts to hack. Then you could spread evenly the number of hacked votes you need among all the safe precincts. That way if a hacked precinct is chosen, the small difference between the hand count and machine count could be attributed to things like chads falling out and undervotes. Since you knew how many votes you hacked, you could even have a set of punch cards on hand with the correct votes if you were really really devious.

To gain an illegitimate advantage of 2%, you'll only need to hack 1% of the vote. Eliminate the "unsafe" precincts and you'll have to hack about 1.5% of the safe ones.

The smallest number of average voters per precinct in Ohio is 242; the largest is 688. Flipping 1.5% of the votes in these precincts would only require changing 4 votes in the smallest ones and 11 votes in the largest.


**********************
Some of us would probably pause for a moment and ask ourselves if the reward is worth the risk...

[althecat]

1. Great analysis.. have been thinking about this but not as clearly as you

obviously have.

thanks

al

[Straight Shooter]

2. Take it one more step and compare it to the precincts which were counted.

Didn't Blackheart choose the precincts to recount, against the express request of John Kerry for random precincts?

Blackheart has been waaaay too involved on waaaay too many levels from the get-go of Election 2004. The fraud in Ohio, as the saying goes, "has his fingerprints all over it."

Excellent analysis, T&P. These are the same kinds of questions which keep me up at night and angry during the day. I never compared it to Russian Roulette, though. Interesting analogy. Blackheart has his reward all wrapped up, in the form of Governor of Ohio. Jesus Christ could run against him and would lose. Know what I mean?

edit: oops

[KaliTracy]

4. it wasn't Blackwell that chose the precincts

as far as I know (the volunteers would know more) -- and it wasn't Kerry that ordered them "random" -- I think that's in the Ohio recount statute which is why so many people got so upset.

But I do believe Blackwell had too much involved here on some levels, (last minute rules and regulations) and not enough on others (Making sure there was Machine Equity, for example).

And he was definitely involved in the merging of the precincts.

on edit -- The precincts were supposed to be drawn out of a hat -- or other randomized means. That only happened in a very few cases. In at least one case, the ballots were "ready" for recount when the volunteers got there, and a whole bunch were for *, then a whole bunch were for Kerry and so on, as if they had already been presorted....

[mulethree]

5. I've heard that as well

That Blackwell chose, but nothing to back the statement up.

Also that 86 of 88 were chosen non-randomly.

If you look at the recount reports you find that very few were
chosen completely at random in front of observers.

Quite a few were chosen less randomly, but in a fair manner - e.g. have one picked by a repub, one by a dem, one by a green etc.

Quite a few more were purportedly random. The elections officials claimed that they had chosen them at random. But they didn't do it in front of observers and not immediately preceding the counting - so several people had advance knowledge for some period of time.

Quite a few more where a dem and a repub elections official each chose one and they considered it fair. Of course If I choose safe precinct X then I know you'll limit yourself to certain precincts that would result in exceeding 3% by the minimum necessary.

Then there is Cuyahoga and probably others who eliminated whole classes of precincts (those under 550 votes), apparently a tradition which would open up all small precincts as 'safe to tamper with'.

The letter of the law says choose at random. The spirit is - a random sampling as an audit. So clearly a random or at least fair sample, but the spirit requires open-ness, so the party requesting the recount can be assured of fairness. It does seem that at least 75 counties failed to be both fair and open, and theres a great stench to it because it should - theoretically - cost nothing extra to do it fair and open.

[Qutzupalotl]

3. Good stuff.

:kick:

[megalith]

6. 3%, Statute or Policy

I had this pointed out to me by an attentive reporter today. And I always learn more by being wrong :)

The actual Ohio statutes, as listed by "Andersons Ohio Statutes" on the web, DO NOT refer to a 3% selection procedure. In fact they are very spotty as to any recount procedures at all.

The documents we have worked from, which are all sourced to the SoS web site, are from the SoS directive # 2004-58; and as such are an administrative directive rather than statutory law. The BoEsare bound to follow directives by Ohio statute, however. I have strong reason to beleive that the directive for recounts predate SoS Blackwell and have been Ohio SoP for many years. I will try to confirm this tommorrow.

So we sould refer to SoS Blackwells directive for a random 3% precinct selction ; rather than the statute requiring the selection. That said...



Muletree makes some good points about variation in the selection procedures as practised by differing counties. In my region what generally happened was the BoEs selected twice as many precincts as requires, that would provide combinations close to 3%, were geographically distributed across the county and/or were balanced between the major canidates. The BoE would then "vote" the final selection of precincts to be hand counted from this first pass selection.

Every BoE member I spoke to first claimed this was random, but then admitted that it really was not. Thay said they were simply trying to find representative precincts that when combined would come very close to 3%, and I do believe that was their motive. I did have two exceptions where the BoEs flat refused to discuss their preselection at all. Quite nasty about it actually.

The preselection always ocurred previous to the witnesses observing the recount, with the exception of Athens County, which was one of only two that actually randomly selected their precincts.

The fact that precincts were to be preselected was apparently such common knowledge that technicians assumed it to be the case when setting up their machines for the recount, which is the real import of Ms Eatons affidavit at Hocking County concerning the "cheat sheet" and Michaels questions to her. The point remains that a person with firsthand knowledge of BoE practise could make a very good bet as to which precincts are VERY unlikely to be preselected, and with good odds, the door to hacking the election is wide open.

Best


Orren Whiddon
SE OHIO Green Co-ord

[Carolab]

9. The Ohio reporter should be more "attentive"

These were posted on DU when the recounts began. They were posted as Ohio's own recount procedures. Anderson's site would not contain them because they are not "statutory".

RECOUNT PROCEDURES
1. All Voting Systems
a) The recount must be conducted by teams having equal numbers of
Democrats and Republicans.
b) Total votes cast must be compared to the number of voters listed in the
pollbook, poll list, or signature pollbook records. In the presence of at
least two election officials of different political parties, the records must
be available for visual inspection by witnesses. The witnesses shall not
be permitted to handle the records.
c) Absentee ballot envelopes returned after the polls closed may be viewed
by the witnesses. The observer may not see the actual ballots, only the
ballot envelopes, which must still be sealed.
d) Disputed ballots may be settled as they arise by the board or by a
majority of the employees designated as teams, if so delegated by the
board.
e) Ballots must be handled only by members of the board, director, deputy
director or other designated employees of the board.
f) Votes cast for write-in candidates must be checked to determine if the
candidate is a qualified write-in candidate and for an overvoted ballot.
g) Witnesses may observe the inspection of the ballots.
2. Punch Cards
Test the Program
a) Prepare a new test deck of ballot cards that are punched and then
manually count them. This deck must not be the same test deck used for
the official count.
b) Process the test deck through the computer to verify that the computer
count matches the hand count.
c) If the hand count and the computer count do not match, but the hand
count is accurate, all ballots must be manually counted.
The Recount
a) Ballot cards must be inspected for hanging chad attached by one or two
corners, mutilations, and other invalidities. If a chad is attached by three
or four corners, a vote shall not be counted for that particular candidate,
question or issue.
b) Over-votes and blank ballots may be separated from the stack at this
time and placed at the top of the stack after the header cards.
c) Ballot page assemblies and rotation header cards must be checked for
each precinct for candidate positions to verify that each candidate,
question, or issue has been properly identified.
d) The board must randomly select whole precincts whose total equals at
least 3% of the total vote. These precincts’ ballots must be manually
counted.
e) Run the manually counted precincts through the computer.
f) If the computer count does not match the hand count, and after
rechecking the manual count, the results are still not equal, all ballots
must be hand counted. If the results of the computer count and the handcounted
ballots are equal, the remainder of the ballots may be processed
through the computer and results tabulated electronically.
g) At the conclusion of the recount, the program must be retested using the
pre-audited test deck.
3. Optical Scan
Test the Program
a) Prepare a test stack of ballots that are pre-marked and then manually
count them.
b) Process the test stack through the tabulator to verify that the tabulator
total matches the hand count.
The Recount
a) If the hand count and the computer count do not match, but the hand
count is accurate, all the ballots must be manually counted.
b) Ballots must be inspected for mutilations and other invalidities.
c) Ballots must be checked for proper candidate position and to verify each
candidate, question, or issue has been properly identified.
d) The board must randomly select whole precincts whose total equals at
least 3% of the total vote and must manually count those precincts’
ballots.
e) Run the manually counted precincts through the tabulator.
f) If the tabulator count does not match the hand count, and after
rechecking the manual count, the results are still not equal, all ballots
must be hand counted. If the results of the tabulator count and the handcounted
ballots are equal, the remainder of the ballots may be processed
through the tabulator.
g) At the conclusion of the recount, the program must be retested using
the pre-audited test stack of ballots.
6. Direct Record Electronics (DRE)
Test the Program
a) Prepare a test cartridge.
b) Process the cartridge through the computer.
c) Verify that the cartridge results match the pre-determined votes cast for
candidates, issues and questions.
d) If the totals do not match, check programming and rerun the test
program until the totals match.
The Recount
a) Check the public counters and protective counters to verify that the
numbers on those counters correspond to verify that the numbers on
those counters correspond with the pollbook, poll list, or signature
pollbook records.
b) Check the rotation on those machines to verify that they match the
proper candidates, questions, or issues.
c) Process cartridges through the tabulator. If the totals are different than
the totals of the official count, compare cartridge totals against paper
audit trail report.
d) At the conclusion of the recount, the program must be retested using the
pre-audited cartridge.
The procedures described in this outline are the basic requirements for conducting a recount. If the
board of elections feels the results warrant further investigation at any period during the conduct of
the recount, it may institute more rigorous recounting procedures, such as hand counting a larger
percentage of precincts, using programming options that total over-votes and/or under-votes, hand
counting all precincts, etc.

http://serform2.sos.state.oh.us/sos/elections/statewide/provisions_recounts.htm

[many a good man]

7. RFC from mathematicians

I want to hear your take on the probabilities involved here.

My contention is that tabulation software can be written that a) leaves no physical evidence of fraud; b) can differentiate between test runs and official counts; c) always returns the same results; and d) never needs modifying after the candidate rotation is entered. And it can run on any MS-DOS or Windows operating system released in the past 15 years.

The only way this hack can be exposed is if 1) the current ballot count is close to the expected turnout; and 2) the complete hand count is taking place in one of the precincts chosen in advance to be hacked.

Thus the crucial importance of picking the precincts to hand count. I am looking to you to validate whether the odds I estimated above are accurate. I suggest that the level of risk is acceptable based on mathematical probablility.

A full hand recount in a hacked precinct IS THE ONLY RISK that has to managed.

:kick:

Note: I don't have data for individual precincts. I created a spreadsheet that has all Ohio counties, vote totals, and number of precincts. Therefore I was forced to use averages for precincts. Let me know if you want a copy of this data...

[many a good man]

8. ...

:kick:

[The Judged]

10. Lawmakers for lawbreakers and deliberations for premeditations ...

In the name of the Father, the Son, and the Holy Spirit. Amen!

[berniew1]

11. Very few counties did a random 3%; see Green summaries

http://www.votecobb.org/recount/ohio_reports/counties/allen.php