VVPBs, what are our options?

[Bill Bored]

I posted this in another thread. It was sort of stream of consciousness, but I like it! So here's a new thread to discuss it more fully.

There are two basic aspects to the e-voting problem: Lack of transparency and lack of accountability. They are obviously intertwined.

The transparency problem can only be fixed by getting rid of the machines entirely and going to something like hand counted paper ballots. The accountability problem might be solved by security enhancements, random auditing, etc. using the same basic technology as we have now but with voter-verified paper records (audit trails).

I summarized the arguments for each and offer a third alternative. Please discuss!

1. We MUST HAVE transparency, even for non-technical voters, which I suspect not only includes the mentally challenged and the non-computer literate, but also the vast majority of election lawyers! This we can call the Land Shark approach, in honor of the WA attorney bringing a suit to bring it about.

2. We MUST make the system hack-proof beyond any reasonable doubt. There are ways to do this, but it wouldn't be transparent. Call this the Geek approach. None of the bills currently in Congress comes close enough to this standard in my opinion. And again, there would be no real transparency, except perhaps to the Geeks themselves.

3. Fuck the system (in essence by ignoring it) and just audit enough voter-verified paper ballots (or as many as we can afford to audit) to rule out any counting fraud with statistical certainty. This is kind of a semi-transparent approach. The statisticians would understand it, and so would many voters, at least in principle. Call this the Sledge Hammer approach. The issue is who would pay for it, regulate it, etc? Currently this power seems to be left up to BOEs including the EAC at the federal level. But we might argue they can't be trusted and so it should be put into law, or we may have to rely on the courts to make it happen.

Of course, without voter-verified paper records, none of these approaches will work, except possible the Geek approach, which again lacks transparency. I am interested in the Sledge Hammer though. If through this process, enough mistakes are uncovered, it might lead to other fixes, or it may be sufficient to deter fraud on its own.

What do y'all think of this little synopsis?

[In Truth We Trust]

1. Paper ballots NOW!!!! Hand counts Now!!!!!

That's what I think!

[BeFree]

2. Paper Ballots

Isn't a Paper Ballot just a record of a vote? A Paper record, right?

So, in the interest of that singular goal of ours, and in the interest of the BOE folks, let us find the fastest, easiest way to get Ballots recorded on Paper. One way of paper recording, that everyone is familiar with, is the cash register at the grocery store. Can we cast our ballots in a similar way? Millions of dollars are traded every day on such a system. Very transparent.

Your Sledge Hammer approach is something that we should have been ready for in 2004. We may have to gear up for that come 2006 if they don't give us what we want now.

[Andy_Stephenson]

3. Read my sig line.

If you want a paper ballot...ask for a paper ballot. No Reciepts...No Records...No trails.

Call it what it is...A Voter Verified Paper Ballot. Nothing more nothing less.

[BeFree]

4. Hey Andy, good to see ya...

Tell ya what, if you think I am talking 'Trail', show me where I am.

I'm talking about the fastest, easiest way to get a Paper Ballot counted in a way that is not a reinvention of the wheel.

You, Andy, convinced me that Paper Ballots, hand counted, ain't gonna fly too far. So, what I am thinking is still paper, but counted fast.

Fast is what everybody wants, eh? It's done everyday, day in and day out at the cash register, it's an option for us to look at, IMO.

[Bill Bored]

6. OK, how would it work exactly?

We don't buy votes like we do soap of course. So where does the paper come from, where does it go and how is it counted? Are you talking about OpScans, bar codes, or what? What kind of machine is it? How will it be tested or audited?

The problem is that $millions have already been spent on this stuff and they aren't about to throw it away. So it needs to be audited.

[BeFree]

8. Exactly?!? Not sure, but

Quote;"
The problem is that $millions have already been spent on this stuff and they aren't about to throw it away."

The idea is that the DRE's would have a printer attached.

The paper would be on a roll. As each vote is cast, there would be printed a running total of cast votes.

The printer would print to that roll of paper the voters choices that a voter can visually verify as being correctly printed.

Any misprints would be marked void.

As each voter makes their choices, and is satisfied the ballot has been correctly printed, their ballot is marked as verified on the roll of paper.

Once accepted by the voter the DRE adds the voter's choices to the previous selections and prints the running total, unseen to the voter.

What you have, at the end of the day, is a paper ballot with a record of votes cast and the totals.

On that paper record, is printed every addition the computer made. Every addition, every computation.

After the last vote cast, there would be printed on the end of the roll, the totals, and after a quick visual check of the additions made by the computer as printed has been made by poll workers, the total as recorded could be verified.

[Bill Bored]

10. I don't get it.

The machine is still counting the votes and therefore would need to be audited. The VVPB prints the voters' choices which is fine, but even if there is a running total on the VVPB, no one knows if it's correct or not. You could for example, as a voter who happens to be paying attention, see that your choices have incremented the totals by 1 vote each, but you won't know if the total before that was right to begin with, and this assumes you are actually even checking this. You can't put the onus on the voter to see if the machines is working properly. The voter only needs to verify his/her individual ballot.

So am I missing something?

It seems that we either hand count all these VVPBs and do not even use the machines for counting (just printing!), or we audit enough VVPBs to rule out any errors that could change the outcome of any race.

[BeFree]

14. OK, let's try it this way

On that paper is every addition the computer makes.

The ballot would look like this: (simplified, I know)

Vote for one only.

Kerry
bush

Amendment #1
yes
no

After the first ballot was cast the running total would look like this:

Kerry 1
bush 0

Amendment #1
yes 1
no 0


After the second vote cast the running total would look like this:

Kerry 2
bush 0

Amendment #1
yes 1
no 1


After the third vote cast the running total would look like this:

Kerry 3
bush 0

Amendment #1
yes 1
no 2

After the fourth vote:

Kerry 4
bush 0

Amendment #1
yes 1
no 3

And so on. Each time a voter votes her choices, her ballot would be printed so verification would be made. If the computer screwed up, it would be quite evident by looking at the running total before the verified vote, and the running total after the verified vote.

In fact, every time a vote was printed, the running total would have to match the sum of one race. So, after 4 votes were cast, the sum of the Presidential race would add up to 4. If the voter missed a vote, it would be easily found, as no number on a verified ballot would follow a selection. Like this sample undervote:

Kerry 0
bush 0

The running total following would be no different from the running total before.

Hope this helps.

[Wilms]

15. If I got it...

This running total is printed on the VVPB??

Or on a separate roll of paper??

If this running total is printed on the VVPB, how do you prevent the voter from seeing that??

[BeFree]

16. I don't know, exactly, but

The new running total would be printed After the voter verifies her ballot.

It would be on the same roll of paper, running total + verified vote = new running total.The old and new running totals would be hidden from the voters view.

Like this:

Running total after 30 votes:
Kerry 29
bush 1

The 31st Verified ballot is cast:

Kerry 1
bush 0

New running total:
Kerry 30
bush 1

The 32 verified ballot is cast:
Kerry 1
bush 0

New running total:
Kerry 31
bush 1

[Wilms]

17. But how, then, do you prevent the voter from seeing that??

(I think seeing a running tally is a no-no.)

[BeFree]

18. Hi Wilms, howyadoin?

Did I answer it ok?

Really, the verified ballot part would be maybe six inches long. So a six inch window would be fixed for that ballot. If it was ten inches, it would be a ten inch window.

The running totals would be hidden from view by spacing those totals an inch or so away from the verified ballot part.

[Wilms]

19. I figured that might be it, but wanted to check. n/t

[Bill Bored]

5. Andy, I'm glad you're feeling well enough to post.

I was hoping not to get too bogged down in the ballot vs. record definition in this thread, only because none of the bills actually use the phrase "voter-verified paper ballot" and if the "record" is going to be the official vote count, I would think this should suffice. There are those here who believe that the word "ballot" automatically implies hand counting and hence makes DREs illegal. I don't know if this is your interpretation, but it does confuse the discussion at times.

Even Ensign's VIVA bill only says "paper version of the voter's ballot" and "individual permanent paper record for each ballot" which clearly allows other ballot formats to be counted initially.

The issues I'm seeing in reading these bills is that they vary in their auditing requirements (if any), there security requirements for the machines (if any), who decides what do to in the event a discrepancy is found, etc.

If we can't verify the source code in real time, even if it's been made public, and we can't design a system with iron clad security, including the tabulators -- see my thread about Rush Holt's bill <http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x337682> --
then we need to think about how to audit enough voter-verified paper records to ensure that the count is at least accurate enough to correctly decide the winner of the election. I call this the Sledge Hammer approach because rather than trying to finesse the design, authenticate the code, etc. (the Geek approach) it basically says the heck with it! We don't trust this stuff and we never will, so we might as well do as many hand counts as necessary to be certain that the machine count is accurate enough to pick the winner.

Now, does there need to be additional legislation (beyond the VVPB requirement) to ensure that this can be done? Should there be a statistical algorithm used to go beyond the initial mandatory random 2 or 3 percent audit (if any)? Would a court have to decide to allow these audits as part of an election challenge, or can they be done by anyone willing to pay for them, right there on election night? Should we start setting up some funds to be used especially for this purpose (i.e., Help America Recount, a DNC GELAC Fund, etc)?

Keep in mind that the Kerry campaign could have done this in Ohio had they been less naive, more computer literate, or whatever. I just don't want there to be any more Ohios.

Therefore, I think we need a strategy that can work given a simple requirement for a VVPB (or a rose by any other name). This may be all we are able to get into law, unless Land Shark is successful.

The Geek approach is a bit too much to ask for various reasons, although I get the feeling that Hillary Clinton might like it based on her comments about the elections in India on Jan. 6. But if we can't thread the needle, why not use the Sledge Hammer?

[msongs]

7. pls read our paper ballot proposal for California ---->

paper ballot is the ballot of record for declaring outcomes/winners or any recounts, secured, counted in public at the precinct/voting location, and results posted in public before actual ballots are sent to a central tabulation center:

www.msongs.com/vvpb.htm

Msongs
www.msongs.com/liberaltshirts.htm
Riverside CA

PS - have started sending this around in CA to political groups, media, etc. got a real thank you from the guy at sfgate.com, the san francisco chronicle, and form replies from actual dems legislators who have NOT responded yet.

[Bill Bored]

9. Sounds like a plan.

So the DREs will NOT be used to count votes then?

They are trying to get this into law in GA, but so far, the legislature is saying to just provide a paper record. The counting is still done on the machines, except for audits and recounts. Be prepared with a fall back position in case you don't get what you want. There needs to be a way to get enough random audits to rule out fraud without having to jump through hoops, based solely on the distrust of the technology which has not been or can not be proven to count correctly.

Good luck!

[Chi]

11. Sorry if this is outlandish

Just an idea, probably not thought through enough.

Voter votes electronically, paper receipt is verified (behind window) by voter visually,
voter pushes 'accept' button, and receipt pops out.
(meanwhile the carbon duplicate of that receipt stays in machine and on roll)
At poll close, machine counts and reports to central tabulator.
With vote information for each voter, is also a voter ID number (randomly assigned by machine),
this information is also uploaded to a State voter verification web page, where anyone who voted
could verify his/her ID number matches their votes.

[Bill Bored]

12. Can't give receipts because if the voter can prove to a third party

how he/she voted, votes could be bought and sold. Sorry.

[Chi]

13. Sorry, I must of missed that memo.. n/t