a game of reverse logic with 91 precincts

[freedomfries]

This is a game of reverse logic for tinfoilhatters, using the Clermont County BOE screens 08/02/05, 09:30:23PM, and 08/02/05, 10:49:17PM to play. At 9:30PM, with 100 precincts counted and 88% of the district votes tallied, candidate S and candidate H are at a tie in the district. How can the vote count of the last 91 precincts be programmed to put candidate S over the recount margin? It's easy to do and you have about an hour to do the calculations. This game works best in humid conditions.
1. First subtract the 9:30 precinct reports from the 10:49 precinct totals. Leave out all the early voting and the write in votes.
(191)S 16162 (100) S 6711 (91) S 9451
H 11689 H 5349 H 6340
T 27851 T12060 T 15791
Precinct Average 120 Voters 173 Voters
2. Then calculate the percentage difference at 9:30 and 10:49PM:
(100) S 55.64% (91) S 59.85%
H 44.35% H 40.14%
Margin of difference:11.29% 19.71%
Shift in margin of difference for last 91 precincts: +8.42%
3. Try adding +4.21% for S and subtracting -4.21% for H. How many votes could S gain? Do a quick head count with 10000 voters and +5% for S, -5% for H. If instead of 55-45, your percentage spread is 60-40, you gain 500 votes for S and lose 500 for H. A difference of ca. 500 votes per candidate won't be enough.
4. Try adding voters to each of the remaining 91 precincts, just enough to carry S over the recount cliff.
Remember that the SOE screen shots show an average of 120 voters for the first 100 precincts and an average of 173 for the last 91.
It looks like you are going to have to add a lot of voters per precinct to make it work. Try out adding 53 voters per precinct. 173 minus 53 would make about 120 voters to start with.
5. Now to calculate the number of vote gained when you combine the 4.21% margin increase and the 53 added voters per precinct, multiply 91 by 120 voters per precinct and the early percentages
S 91 X 120 X 0.5564 = 6075
H 91 X 120 X 0.4435 = 4843
Now subtract these numbers from the 10:49 PM 91 precinct totals:
S 9451 - 6075 = +3376
H 6340 - 4843 = -1497
6. Finally verify how much difference the combined % shift and vote patting would make to the final district totals, by subtracting the gained votes from the S totals, and adding the lost votes to the H totals:
S 59132 - 3376 = 55756
H 55155 + 1497 = 56652
7. Now enter.
Game end

[Old and In the Way]

1. We getting better at spying voter fraud, I think.

I have a question.

Regarding this observation:

"an average of 120 voters for the first 100 precincts and an average of 173 for the last 91"

Has anyone done a correlation to actual population of the 1st 100 precincts vs. the last 91? Is there 30% more population, on avaerage in those precincts vs. the 1st 100?

[Fly by night]

2. Great question. I hope TIA and others can address it.

Either population size of the precinct, or number of registered voters.
Both would be useful to know.

[ramblin_dave]

3. If votes were padded, wouldn't there be

a mismatch between the number of votes cast and the tally of voters from the poll books? If so, how could this be done? If the poll books are "adjusted" to match the padded count, then there are people on these books who show as having voted but who would know they didn't vote.

[autorank]

4. Poll books...great question.

These should be open for public inspection. That's how they catch politicians running for office who've never voted. The nightmare is that the doctor those with "real" names. What do you do, call up and say, hey did you vote ... they'd hang up immediately. And how could you verify the answer. Sounds like they were off their game this time and were trying to pull it out at the very end, like the '70's Raiders, but unlike the Raiders, they may have run afoul of the rules (the Raiders never hacked the score board!).

[adolfo]

5. Open

Ballots and Poll books are open for public inspection as required by Ohio law. That means anyone can count the ballots and audit the books. Any volunteers?

[mod mom]

17. Definitely count me in. I just got back in town and assumed this would be

organized but it appears not. I have been told that Hackett refuses to recount. the DLC must have gotten to him. :(

[roseBudd]

7. True but the signature needs to match, so adding sigs next to names

of people who did not vote is risky.

The one day I did Warren County BoE with Andy he spotted signatures that did not match.

[wli]

11. harder to spot bogus removals than fake additions

[autorank]

14. Important point. Additions would seem the order of the day.

In the case of Hackett-Schmidt, eliminating Hackett voters to ensure a win would not help with the overall 2nd District totals, which were vital when the great "humidity" crisis emerged. How would they do phony additions, if they even bothered?

[autorank]

6. GREAT WORK!!! ?'S

1) Do yo have a link for the individual precinct data by precinct for Clermont with time reported and counted on time (first 100) versus counted later (last 91).

2) This reverse logic clearly works and it would be a great video game -- "FIX IT! Screw with entire countries and regions by Fixing It--a strategy game for electoral manipulation. General audience."*

Is this how it might have gone: OK we'll win, don't have to worry much. We got rid of a bunch of Hamilton precincts (those Cincy suburbanites) and we just "kill" in this precinct. Oops, these totals aren't looking good, holy shit, look how he's doing in rural areas. Stop the presses, yeah, say it's humidity, screw 'em. Let them prove otherwise. OK, now what do we do? <insert reverse logic algorithm 1.0>. That's fine but the damn thing has Schmidt winning by just 1/100th of a point above the mandatory recount thresh hold. A 0.026 margin isn't what's called for. Go back and give it some "substance." Fine, that looks a lot better.
GAME END. Smoke a cigar, indoors.

Who knows, but this is interesting stuff, that's for sure. It's a circumscribed case, not like the whole country or the whole state of Ohio in 2004.

Thanks for doing this!


*Actually, this would be a great massive multiplayer online role playing game. You could adopt an identity, get cash and "escorts" as weapons and do it all on some "borrowed" faiclity.

[freedomfries]

8. you're welcome autorank

a ver circumscribed case, as you say.

[Kip Humphrey]

9. Circumscribed but highly mechanical!

[TruthIsAll]

10. A spreadsheet makes it clearer

Time	N	S	H	TV	TV/N	SP	HP	Margin
10:49	191	16162	11689	27851	146	58.03%	41.97%	16.06%
9:30	100	6711	5349	12060	121	55.65%	44.35%	11.29%
Diff	91	9451	6340	15791	174	59.85%	40.15%	19.70%
								
Chg	-9	2740	991	3731	53	4.20%	-4.20%	8.41%
PctChg	-9.00%	40.83%	18.53%	30.94%	43.89%	7.55%	-9.48%	-11.29%
								
Last91		9451	6340	15791				
Incr		6107	4868	1239				
Diff		3344	1472	14552				

GT		59132	55155	3977				
Chg		55788	56627	-839				

[freedomfries]

12. questions for TruthIsAll

I like your spreadsheet format,
but I don't understand how you figure out the Pct Change line.
Is there a way you could turn the most salient parts into graphs?

[TruthIsAll]

13. Response

PctChg = Chg / 9:30pm Value
40.83% = 27.40 / 6711

What are the salient parts you would like to graph?

[freedomfries]

15. sequence

This, if DUable, should go on separate threads, since we have moved from logical into mathematical thinking.
1) probability of -9% producing 40.83%, ect. ???
2) graphing the humidity factor: TV, TVN, SV, HV, SP & HP at 9:30, 10:49 and diff in bar graphs, then GT, and GT - Chg

[paineinthearse]

16. Voter-verified paper ballots

NOW!