Exactly how would you "fix" the tally when punch cards are used?

[chorti]

Regarding the Hocking County scandal where the Triad tech came in and took apart the vote tabulating computer. I think we need to be thinking not just about the hardware but also the software. I'm not saying this happened, but let's assume for a minute, that Triad was trying to cover something up and needed to go around to all the counties where the "fix" had been made.

This machine was reportedly an old computer. It is likely that both the tabulating software and the database software are old - either DOS-based or Windows 3.11. So this machine has an old database program, FoxPro or dBase or something like that. And it has punch-card tabulating software, which probably spits out strings of data, with a header row announcing the id of the precinct, then a header row announcing a new ballot, and then all the rows of data, one for each contest that the voter is voting on. Then a ballot end row and another voter header row, et cetera, until you hit the end of precinct row and the next precinct header row. The precinct header row is important in Ohio punch card tabulators because the order of candidates is switched precinct to precinct. Plus some precincts may be voting on a certain local race down the ballot that other precincts are not voting on. But this is very basic software.

The smart software is the database program. There will be one database file with all the precinct ID's, the list of all races in that precinct, and information about the order of the candidates in each precinct. So the tabulator file creates a text file that is imported into the database program. The data in this file is then converted to meaningful vote tally information. This is precisely where any "fixing" of the data would occur. You could easily program this basebase program to flip every 100th Kerry vote to Bush, or whatever you wanted to do. You could program only certain precincts to be "fixed". This would be important knowing the 3% recount rule in Ohio. In any case you would want any "fix" to be very subtle so that any Dr. Phillips examining the precinct totals couldn't find any anomalies.

So now, you come to a recount. First, it would help if someone from the BOE would be in on it, in order to make sure that one of the "fixed" precincts was not chosen "randomly" for a recount. You should be home free unless someone actually looks at the database code. In this case you would not need to examine anything or change a hard drive or whatever.

But let's say the BOE was not in on it and the precinct chosen had been fixed. You would need to put in place a new database file that is not "fixed" so that it agrees with the hand count and you don't have to do a 100% hand recount. Downloading this file by floppy disk would be too suspicious. It would be much easier to say that the machine was burned out or the battery was dead or make something up - and give you an excuse to change the hard drive.

The database file should be able to recognize the precinct header information and determine the order of candidates for that precinct. This makes it unnecessary to call anyone with the precinct number in order to determine the order of candidates to make sure the computer is programmed right. The only reason you would need to call someone with the precinct ID would be to determine if that precinct is one of the ones that was "fixed". If so, you would need a software "patch" in order to unfix the tally for that precinct. That is, you would need a new database file that would correctly record and tally all the votes in that precinct so that it will correspond 100% to the hand recount. If its off a couple votes, even 5 or 6, from the tally recorded on election night, that can probably be explained away. And you are able to avoid the 100% hand recount that would definitely show the fraud.

[mattclearing]

1. You can't fix it.

That's the thing. They are making the machines output numbers regardless of the ballots being fed through the machine, if I understand correctly.

A hand count will reveal the fraud, because they can't go and change all the ballots.

It would take YEARS to make that many fake paper ballots that match the "official" totals.

[General Paranoia]

3. They may have been able to fix it after all

If you also have a card punch you can read the ballots in on a card reader and copy everything to fresh ballots on the card punch. Of course the copy changes a "few" things as the data goes by. You could produce a clean deck that exactly matches the results you want.

I am assuming that you can get fresh ballots through a card punch. Someone else may have more information on that.

The way you detect this has been done is not by the totals in a hand count but look at the holes (a magnifying glass would help). If you have nice clean holes with no fibers sticking out from the corners then it may have gone through a card punch.

Holes can also be punched by hand in under votes. I suspect they might not bother with inserting the ballots in the voting machine one at a time but instead might try to gang punch a group at a time. If this were the case you might be able to tell by noticing that the holes have rounded edges where a tool was used or a wad of chad went by. The voting machine keeps the stylus in the middle of the chad and doing a bunch at one time might leave an artifact visible to the eye. In a lab it would probably be obvious.

[chorti]

4. new cards aren't necessary

Let's say that the GOP believed that Ohio was so close and that with everything else they were doing - voter suppression, lack of voting machines, touch screens miscallibrated, it would only take a little bit of something to push them over the top. This could be as subtle as one - only one - vote flipping from Kerry to Bush in each precinct. This would give Bush an extra 0.1% or 0.2% statewide. Maybe that's all they thought they would need. If the hand and machine recounts are only off from the voting day tally by one vote, it would probably be acceptable to the BOE - explained as machine error. And they avoid the 100% hand recount. The remaining (97%)machine recount, using the software fix still in place, would equal the original Nov. 2 count, so no suspicion would be aroused there either.

[General Paranoia]

8. They were pretty ham-handed

Check this other thread:
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=171627&mesg_id=171627

I suspect it wasn't that close and even using all their outer defenses I think they were hard pressed by the time the votes got to central tabulation. That's why there are big and weird differences from exit polls.

I have read lots of comments about how sophisticated they were. Software wise they were amateurs. For God's sake they had to send a guy out to change hardware for the recount! It looks like they had a bunch of pious foot soldiers with a Taliban like loyalty but not too bright. They probably infiltrated into the voting system over 5-10 or more years. These people were so clueless that they printed out multiple poll tapes to figure out how to change the results!!! Nothing like hard copy that you don't personally shred to provide an audit trail. We won't even talk about remnants of print files on disk.

Everybody in the press keeps commenting about how there is no audit trail because of the computerized voting machines. Wrong wrong wrong. They are frantic to plug the holes in the dike which is looking more and more like a shower head. During the recount they'd holler out the totals so the machine operator will know! We just need to keep turning over rocks to see the tar footprints everywhere.:dunce:

[chorti]

2. another hard drive replaced in Union County

According to the Cobb recount team, another county had a hard drive malfunction just after the election. See: <http://www.votecobb.org/press/2004/dec/pr2004-12-16d.php>.

This from the Cobb observer in Union County:

"There were some problems with the automatic counter because the humidity had warped the cards and they would stick. They did run through smoothly when a new header card was put in. David Moots (a local BOE Board member) told us the day after the election that the hard drive on the machine, a Triad systems machine, failed and had to be replaced. They took the old harddrive with them when they replaced it but he sent a subpoena to them and got the old hard drive with a statement it had not been tampered with."

[Ellipsis]

10. means nothing

[FogerRox]

5. The exit polls matched the areas where punch cards were used

nearly perfectly. BUT in areas where Diebold Optical scanners were scanning paper ballots-(results were then fed into a PC--GEMS type tabulator)The exit polls were off from the optical scan areas.
Bush won where optical scanners were used--Kerry won where punch cards were used.

[eomer]

6. I didn't know that exit poll data at the county level was available

Can you tell me where to get that data?

[FogerRox]

11. this may not help--state data

http://www.exitpollz.org/

[chorti]

7. not true

The exit polls were off in both cases. In Ohio many of the Northwest Ohio counties use punch cards. Yet the final exit poll shows Kerry winning Northwest Ohio, 52 to 47. I'm sure this exit poll was off by several points in Kerry's favor, because it is so far from the 2000 result. But the actual results from NW Ohio have Bush winning 56 to 43. This would mean that the exit poll is off by 9 points, which, even with a relatively small sample, would be beyond the 1% margin of error. Seven of the 25 NW Ohio counties used optical scan, one touch-screen, and 17 used punch-cards. (BTW, the penultimate exit poll had a 50-49 Kerry lead here, making the actual result closer to the margin of error of this uncorrected exit poll.)

There are not enough touch screen counties in Ohio to compare to regional exit polls, so you would need to look at other states, and, yes, I agree that in many of these the results seem to be off by 2 to 5 percent. In Ohio of the seven touch-screen counties, I would say that two of them (Ross and Auglaize) seem to have Kerry lower than expected, although you would never notice a one or two percent callibration error. You could argue that Mahoning came in lower than expected too.

Touch Screen Counties
............Gore+Nader....Kerry......Diff
Franklin ..... 51% ...... 54% ...... +3
Lake ......... 48% ...... 48% ...... --
Knox ......... 36% ...... 36% ...... --
Mahoning ... 64% ...... 63% ...... -1
Pickaway ... 39% ...... 38% ...... -1
Ross ......... 46% ...... 44% ...... -2
Auglaize .... 30% ...... 26% ...... -4

STATE ....... 49% ...... 49% ...... --

[roseBudd]

13. your info is wrong

70% of all votes cast in OH were punch card. There were plenty of punch card counties Bush won.

Clermont, Butler, Warren just to name a few.

[Hamoth]

9. my thought: Some tabs defaulted undervts as Bush vts

Reading the recent standoff in ohio where a machine was counting undervotes for bush explained a lot to me. The difference is in the the tabulation. Like in NM where bush was hte default selection on many machines, a person who cast not a vote for pres would still vote for bush.

Could a similar default have been setup on tabulators where punchcard undervotes were SOMETIMES (about 1/3rd of the time) counted as votes for Bush?

[roseBudd]

12. PM me I have some interesting info along those lines n/t