An idea for automated vote counting using optical scan.

[Pobeka]

(Don't split hairs with me here, this is still a concept.)
------------------------
Part 1, ballot counting:
------------------------
Build a machine, and all it can do is count the voter marks by rows A, B, C, D, ... That is literally all it does. (and it does that by a looking at a physical x-y location on the card, sort of like a punch card setup but optically performed.) You *can not* change the program in this machine.

Before the election, candidates are randomly assigned to row A,B or C (and grouped together by race). Ballots are printed according to the outcome of the random assignment.
(Random assignment is actually the weak link here, but assume it's possible).

It seems impossible to me, that you could rig the vote counting at the machine, because the machine has no knowledge of which candidate is in which row.

Only the human beings know that column "A" is the dem candidate, column "B" is the rep, and column "C" is not used, or other. So the election officials perform the logical mapping of which candidate got how many votes. It is simple.

------------------------
Part 2, final tabulation
------------------------
This is done with human beings, phones, and many pairs of eyeballs.

You then would need to phone in the results to the next "level up".
This is done, as well as the results are posted outside the polling location.

At the next "level up", the subtotals are totaled, and all of their individual results are posted, and phoned in. It would be the responsibility of all the "lower levels" to insure that their individual results as posted by the next level up was correct.

You now have a system where "many pairs of eyes" can check and cross-check at every level in the system.

[tocqueville]

1. why is it so difficult to put a paper into an envelope

and then have the results on paper counted by legally responsible citizens under active supervision of local representatives of all parties involved in the election ?

it works in Europe (and in many other countries in the world) and the results are published within the following hour after the polling stations close. Fraud is very rare.

[yowzayowzayowza]

2. European & Cdn elections have..

a minor fraction of the number of races the average US federal election. They are not comparable. The shear number of federal, state, county, municipal and initiative votes our democratic system has amassed is beyond the scope of manual counts.

[tocqueville]

4. you don't have to run them the same day

don't you think we have "state, county, municipal and initiative votes" here too ?

[oscar111]

5. EU must elect city officials like we do: so they must have hand counts of

that race too.

Perhaps they just elect them at other times.

We could as well. Worth it to get hand counts. Hand counts are the holy grail for ending GOP election thefts.

[oscar111]

6. Toqueville wins: Hand Counts are do-able

so lets toss machines, all of them.

PBHC is the thing to do.

"papaer ballots hand counted"

[yowzayowzayowza]

7. "The Holy Grail" indeed...

the mythical pristine hand-counted incorruptable election.

"Those who cannot learn from history are doomed to repeat it." - George Santayana

Whatever, dream away.

[tocqueville]

8. what's so pristine with hand counting ?

Of course fraud happen in Europe too (not because of the hand-counting but because of local manipulation with the voters lists).

But what I know of it is always exposed and I don't know why some still try. It happens mostly in remote rural areas, and it's always local stories of a mayor wanting to be reelected.

I don't know about any MAJOR frauds in the BIG elections in countries like UK, France, Germany, Spain, Italy in the last 40 years...

[yowzayowzayowza]

11. Sure, I look forward to ...

having to go vote umteen times and then hang around to perform a function that a machine (with proper auditing) is perfectly capable of doing.

The answer to our election problems is in the 21st century not the 13th:

http://en.wikipedia.org/wiki/Quality_control

Certainly current undocumented voting(DRE, lever) and unverifiable processes(phantom tabulation) are an abomination, but letz not throw out the baby with the bath water. If "doing it by hand" were so superior, the industrial revolution would have been superfluous

[tocqueville]

15. come on, don't distort facts

In a period of 5 years I vote maybe 5 times, some times twice a year. Always on Sundays. Takes five minutes, hardly no lines at all. At 8 o clock, the same evening the first exit polls arrive, one hour after the "after debate" starts.

there are 30 million voters in France in a big election

13 millions in Canada

http://www.global-conspiracies.com/canadians_hand_count_their_13mil_ballots.htm

I don't have the feeling of living in the 13th century

BTW elections didn't occur at that time

[yowzayowzayowza]

18. Don't avoid the point:

1. Therez no reason to hand count when an appropriately audited machine can do just as good a job. Believe it or not, the industrial revolution was a good thing.

2. Breaking US elections down to the number of races on a typical Cdn ballot would require voting 3 to 4 times as often.

I don't have the feeling of living in the 13th century

I notice your regimen did not include participating in the hand-count brigade either.

[tocqueville]

20. there is no reason to use a machine

1) when it takes a couple of hours to open some hundreds of envelopes under supervision of elected representatives and other local officials. The time winning of recording and transmitting the result of a machine is about the same unless there is a direct wire to CNN (and you don't want that, do you ?). The program can ALWAYS be altered. Bribing the local Democrat in your case plus a couple of hand counters and a notary public is far more difficult.

2) it's a matter of organisation and BTW going more often to the polls is good for democracy, not the contrary.

3) the counters here are designed by the parties. One reads, the other writes. They are often from different parties. If they lack people, any valid elector can be required.

[yowzayowzayowza]

21. Sorry, there is such a thing...

as too much democracy. I don't really need to show up to vote in a primary, primary-runoff and election every two years to select a qualified city dog-catcher, er rather MCRE - Municipal Canine-Retrieval Engineer. Unfortunately, the current quantity of US races (federal, state, county, municipal and initiatives compounded by primaries and runoffs) broken down to the size of Cdn ballots would require voting constantly; just RIDICULOUS. Voting less often on more races utilizing verifiable technology is a reasonable solution.

The program can ALWAYS be altered. ...and an appropriate audit will ALWAYS (1:1,000,000) catch the alteration, thus rendering the tactic irrelevant.

Machines don't steal elections, PEOPLE DO in the absence of sufficient controls on the machines.

On edit: gotta get the acronym rite!

[tocqueville]

23. the system is working in Europe and Canada

and the size of the US isn't a viable argument. Even if there are plenty of different elections in the US due to the federal structure, it wouldn't require a constant voting, unless seen in time. But what does it matter if there are elections in California and none in Utah ?

When it comes around your democratic control of machines by HUMANS would probably use as much manpower as a paper vote.

[yowzayowzayowza]

24. Difference is not size, but quantity of races.

n/t

[tocqueville]

25. OK how many elections do a state of the size of California

have in 4 years ? 25 ? at the same time ?

Besides in many European countries it's allowed to put several paper ballots in the same envelope, for different votes.

It's not very difficult to put 3 different coloured ballots, for let say a municipal, state and federal elections. The counting isn't much longer.

[yowzayowzayowza]

27. Post of DUer who actually..

administered US elections:

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x385783#386348

[tocqueville]

32. OK then use machines ONLY for the small local elections

because eventual fraud would anyway be too obvious.

I agree that paper count can be frauded too, as I posted before. But the risk is minimal, what statistics in Europe show. It demands too much complicity...

Then it can be discussed (I am not a specialist) if the electoral system in the States can be simplified or done another way. Nobody disputes that countries like the UK, France, Germany have a very high level of local democracy, even if they use different methods.

Besides there are other aspects than machines to be discussed too :

The voter suppression, specially of Blacks is very obvious and wouldn't be tolerated over here.

the suppression of participation for "felony" is used to an extent which isn't acceptable in Europe. Some countries let even felons vote, because they consider that it's a way of making them reenter society...

and most important : forbid donations of more than $100 to candidates.
Donations from corporations, etc.. are strictly forbidden and could even render jail and in any case uneligibility.

Most European countries have a system where the State gives the money the candidate need for their campaign. To be fair the amont is equivalent to the number of votes they had in the last elections. If they get more votes, they get more money, if they get less they have
to repay the State. And they get enough money.

All candidates must have the same TV time. Private advertisement and ads of private support organisations are forbidden.

Tough ? Morons are still elected in Europe, but maybe we have so far skipped the worse ones, at least since Hitler.


Thanks for the discussion I have to go to bed

cheers

[oscar111]

3. Toss ALL machines: complexity invites new tricks

The reply above me is right _ if we get results in an hour in other natons,

using PBHC

PAPER BALLOTS HAND COUNTED

then why introduce a machine?

machines introduce complexity, and in the crannies of that, tricks can be invented. Geniuses with more ability than me or you, dear OP {no insult intended, really none}, can cook up tricks where we see no possibility.

Witness magicians.
"no way he can levitate the lady!"

But it appears he can. So too with vote theft.

We must leave no room for crannies in which genuises can invent vote thefts.

[Pobeka]

17. The difficult part is in the counting.

Anyone who's hand-tallied large number of data-cards will confirm this. It seems like a simple problem, but really you don't know when an error occurred.

All I'm suggesting is a machine could do this task and make less errors.

I don't offer proof of this assumption. I'll bet the actual error rate of hand-counting is also not known in election systems like we have here in the US with as many as 20 ballot decisions to indicate.

[Bill Bored]

9. This is what I think some Op Scan proponents think is going on.

Unfortunately, it's much more complicated than that.

You have straight party tally rules, undervote protection, overvote protection, ballot reject settings, etc. All this can be programmed on the same server on which the ballots are laid out, such as GEMS. Then it's downloaded to a whole bunch of scanners and that's what makes it so dangerous!

I think you've got a good idea in principle though.

[Pobeka]

10. Would it be less error-prone than hand counting?

This is the real question in my mind.

Both methods have error. Anyone who deals in large numbers of tabulations of data by hand has a sense of how severe the errors can be, and especially how hard they can be to detect.

I suppose the way you'd check the errors is by random auditing of a random selection of ballots, and then have a <large number> of people independently hand count each ballot until they arrive at the "truth".

[Bill Bored]

12. Given how many under/overvotes there can be,

a precinct count optical scanner would reduce VOTER ERROR.

As far as other errors, if the machines are programmed correctly, they should work fine. Hand counts are always a possibility if there is doubt, but this needs to be in the law in my opinion, and so does random auditing of Op Scan ballots just for starters.

[Pobeka]

13. I was thinking in terms of testing the system.

Not in terms of actual election operation. I should've made that more clear, but dinner was calling :)

Of course, during an election, some level of independent auditing is a requirement.

Over dinner, that auditing thought morphed itself:
--
We like to whine about the lack of exit polls.

Why not, build directly into the auditing system, a requirement that says if the audit and the final tally differ by <reasonable %>, it legally requires a recount and a hand-count. But this would happen at the *precinct* level.

[yowzayowzayowza]

14. Looks like we're on the same page:

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x387455#387546

[Pobeka]

16. Yep.

Why invoke a recount of the whole race if the auditing system can detect something's wrong at the precinct level?

[Bill Bored]

19. Which audit?

Do you mean an exit poll as an audit? That's apples and oranges IMO.

If you mean, what should be done if the random audit finds a discrepancy, I agree this should be put into law instead of some BoE or SoS regulation that can be changed at the drop of a hat. But to do this you have to get it right or err on the side of caution.

See this thread about New York's new hand count law:

<http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=386269&mesg_id=386269>

[Pobeka]

22. No, I mean the audit serves 2 purposes.

As yowza**3 says, the audit results are posted, and they by definition are far better than any exit poll, because the audit would sample far more ballots than an exit poll would.

But, come to think of it, audit results wouldn't be available until the precinct had closed, so the comparison to an exit poll doesn't make sense.

This is what I get for thinking while I type ;)

[Carolab]

26. What's wrong with the Open Voting Consortium's system?

It's already underway and gaining momentum...

http://www.openvotingconsortium.org/modules.php?name=Content&pa=showpage&pid=12

[Pobeka]

28. Because it is programmable.

Especially because it is programmable on a commodity PC, running a commodity operating system.

Which means I can fool the voter with a rigged verification program that reports "you voted for candidate A", when in fact the bar code is printed for candidate B.

Then by any number of means, the verification software is replaced at the end of the day with the "correct" verification software.

There is always a weak link in programmable software systems. It is a question of "who's guarding the guards?", and then "who guards the guard who guards the guards?". There is always some level open to attack in a system has knowledge of the candidates.

This was the genisis of my idea, that the counting system must be completely unaware of the meaning of the things it is counting.

[Carolab]

29. So, you've studied it, then?

And decided against it?

[Pobeka]

30. No, I haven't looked at it closely. This is my guess from 50,000 feet.

I need to look at it more closely, but I'm averse to any system that can be reprogrammed for each election.

[Pobeka]

34. This is scary stuff.

I wouldn't touch this with a ten foot pole as a system for automated counting. For human counting it might be ok.

The problem is that people can't easily read a bar code, they have to trust what the talkin' computer tells them.

Then the system implementors have to trust that someone hasn't messed with the python interpretor, or the programs that verify things like python interpretors.

[yowzayowzayowza]

31. Despite Open Source,

the system would still need mandatory manual precinct-level audits to rule out tampering. ALL automated systems MUST validate their operation EVERY election.

[Pobeka]

33. I think open source is a red herring.

I have yet to find a computer scientist who can look at source code on one computer, and *guarantee* that the binary program running on another computer is compiled/linked from that same source code.