Wednesday, May 26, 2004

Audit trail recounts won't happen

We actually like the idea of touch screens and other voter-interfaces that provide private voting to the disabled, eliminate over votes, warn voters of under votes (and potentially offer an explicit "under vote" choice), and increase the rate of accurate interpretation of the votes.

We believe that the record produced by the system must be the ONLY record of vote, and that the voter must be able to VERIFY that the official record of votes records the votes that the voter intends, AFTER the votes are recorded and BEFORE the votes are cast. If the voter does not like what is recorded on the permanent record, they must be permitted to exchange the record for a new opportunity to vote.

There are additional requirements, but I shan't go into them here. Architectures such as that offered by the Vogue Election Systems' AUTOMARK illustrate the type of VOTE-RECORDING system I described.

I guess it is fair to say that the some bill may not intend to deceive the voter -- but none-the-less that is what they would do. In my opinion it is quite cynical to ceremoniously present the voter with a summary of what the voter intended, ask the voter to carefully check that the paper correctly records what the voter intends, and NOT USE that verified record to determine the election result. Instead the system will proceed to determine the election results from a different record that initially or subsequently contains votes other than those the voter intended. If there is any doubt that this can and has and will happen, and is unstoppable, we should confer further.

Colorado law regarding "recount" has been modified by HB04-1227. I have inserted the revisions into the statute to produce the "new" law and pasted it below. Paragraphs 1.10.5-110(1)(b), (3) and (4) were revised. As you can see in (3) the law now permits a recount of votes cast using either "electronic ballot images" or "permanent paper records".

NOTE-1: While the term "ballot image" is defined, the term "permanent paper record" is not defined. However it is required in order to pass certification according to a new section including:

1-5-615

(p) SAVES AND PRODUCES THE RECORDS NECESSARY TO AUDIT THE OPERATION OF THE ELECTRONIC OR ELECTROMECHANICAL VOTING SYSTEM, INCLUDING A PERMANENT PAPER RECORD WITH A MANUAL AUDIT CAPACITY.

(2) THE PERMANENT PAPER RECORD PRODUCED BY THE ELECTRONIC OR ELECTROMECHANICAL VOTING SYSTEM SHALL BE AVAILABLE AS AN OFFICIAL RECORD FOR ANY RECOUNT CONDUCTED FOR ANY ELECTION IN WHICH THE SYSTEM WAS USED.

NOTE-2: The Secretary of State shall decide which method is used. It can easily be proven that counting of paper ballots can be inaccurate, the SOS is on record as disliking the recount of the permanent paper records, and it is many times more expensive to count votes on permanent paper records than on electronic ballot images. Additionally, it is impossible to prove that the count of votes recorded on electronic records does not reflect the voter's intent. Therefore, it is highly unlikely that the votes on the permanent paper records will ever be counted. (Past experience has shown that the SOS will not release these records for independent assessment.)

NOTE-3: To achieve privacy, there is a requirement to randomize the ballot images. This makes it impossible to prove that any voter's votes are INCORRECTLY recorded.

There are numerous additional problems, and you are absolutely correct when you say, "VVPBs are insignificant in CO unless they are the first line of count." Your reason (the recount law) is not correct, but the new statutes are a sham -- another intention to deceive voters into the belief that their vote will be counted as they intend.

When you mention the form of the verified ballot, we believe that the voter should use the same ballot whether they hand mark or machine mark the votes on their ballot. Every ballot must contain the full ballot text so that when the voter verifies their vote, they know what they are verifying -- rather than a shorthand "AMENDMENT 28 YES".

The problem with a two ballot system is that the ballot that is counted controls. Fraud and errors that creep into this ballot decide the election. In almost all cases the "other ballot" will not be counted. Intuitively I dislike the two ballot approach because I fear that elections will be thrown to the courts to decide whether the candidate that won the electronic ballot vote or the candidate that won the paper ballot vote is the winner. This cannot help but cause bad feelings and DISTRUST of our election system. And, if one race comes out different, then all races become suspect, and even all elections that use (or have used) the vendor's equipment.

It's not worth the risk to go paperless. It's not worth the risk to go dual-ballot.


=======================================================================

Colorado Statute as revised by HB 1227

1-10.5-110. Procedures for recount on direct recording electronic voting equipment.

(1) Any recount procedure for votes cast on direct record electronic voting machines adopted by a designated election official shall:

(a) Be submitted in writing to the secretary of state; and

(b) Provide for a recount of votes cast on such voting machine as prescribed by this section.

(2) A recount of votes cast on direct record electronic voting machines shall include early voting, election day voting, and election day emergency voting. The designated election official shall produce a recount report that shall include, but shall not be limited to, the following:

(a) A summary of the recount results;

(b) Political subdivision recount results; and

(c) Precinct recount results.

(3) A RECOUNT of votes cast on direct record electronic voting machines shall be conducted using electronic ballot images OR PERMANENT PAPER RECORDS. THE SECRETARY OF STATE SHALL DECIDE WHICH METHOD OF RECOUNT IS USED IN EACH CASE, BASED ON THE SECRETARY'S DETERMINATION OF WHICH METHOD WILL ENSURE THE MOST ACCURATE COUNT, SUBJECT TO JUDICIAL REVIEW FOR ABUSE OF DISCRETION. As used in this section, "ballot image" means a record of each vote cast by a voter that is stored on a removable memory device contained in such voting machine. All ballot images shall be randomized in such voting machine to assure voter anonymity.

(4) All recounts of votes cast on direct record electronic voting machines shall retabulate vote totals from individual ballot images OR PERMANENT PAPER RECORDS. Tabulating incremented or summary vote totals from individual direct record electronic voting machines shall not constitute a recount for purposes of compliance with this section.

(5) Before any recount of votes cast on direct record electronic voting machines, the designated election official shall store or record all previous preelection, official election, or postelection tabulations of votes cast on such voting machine on an external device such as a diskette, tape, or compact disc.

(6) No tabulation system software utilized by a designated election official to perform a recount of votes cast on a direct record electronic voting machine shall delete any previously tabulated results from such voting machine, including preelection, official election, or postelection tabulations of votes cast, in order to perform the recount.

(7) If the retabulation of ballot images from a removable memory device is impossible due to destruction, loss, or any other reason, the audit trail from the direct record electronic voting machine from which such device was removed shall be copied to a new transfer device and the ballot images shall then be recounted by the tabulation system software.

(8) After all of the ballot images from all of the removable memory devices used in the election have been retabulated by the tabulation system software, the designated election official shall produce the recount reports described in subsection (2) of this section to determine if the recount results and official election results match.


==================================================================


Paper ballot necessary, but not sufficient

Hello Al - Before I comment on your comments, below, let me say that I consider voter-verified paper ballots on touch-screens essential but not sufficient. If a touch-screen is to be used at all, it MUST have a voter-verified paper ballot, but there's more to it than that.

Whether they SHOULD be used at all is a separate question. But they ARE being used and/or are about to be used in many places.

If a bill that calls for a VVPB gets passed at this point, it may mean that if counties DO BUY THEM... then they would buy them with VVPBs... or WAIT to buy them so as to avoid an expensive retrofit in the near future. Counties' having to delay the purchase of paperless DREs gives us a chance to get them to reconsider a better alternative, such as optical-scan or other paper ballots... especially if they have to "deal with paper anyway." Case in point: Ohio, where the legislature recently agreed that VVPBs would be required in the near future. Suddenly LOTS of counties that had been on the verge of getting new touchscreens have delayed their decision (at last count, 22 out of 31).

Quoting alkolwicz :

> The "audit trail" bills are intended to deceive voters into believing that
> they are verifying the votes that will decide an election. This is not what
> these bills do.

It may be that some bills are introduced with a deceptive intent. I don't believe that's the case for all such bills. BUT... that being said, the Colorado perspective naturally is affected by the CO law that says only the original counting method can be used. Thus VVPBs are insignificant in CO unless they are the first line of count.

Other states (such as CA, where I live) have quite different rules about recounts. Here ANYONE has the right to ask for a recount, and if the original was done by electronic means, the recount can be done electronically, manually, or both. (That's why here in CA we point out that DREs that LACK a VVPB don't comply with state law... because they don't allow for a meaningful MANUAL recount.)

> Instead, voters would be verifying a secondary document that: (1) does not
> necessarily reflect the voter's choices, (2) is printed in a form that is
> excessively difficult for voters to verify and costly for officials to use
> for independent counting and (3) is not the official ballot and will rarely
> be counted.

Voter-verification of a paper ballot, to me, means that if the voter's choices are NOT reflected on the paper they review, that they get to spoil the ballot and start over. It's not verification if the choices shown are not what you intended. If a bill doesn't allow the voter to do it over, then it is not a good bill.

The manner of printout for a VVPB from a touchscreen or DRE would depend on design, which surely varies from one vendor to another. Diebold Election Systems, in a bid, told Santa Clara County in CA that it already had the capability to print what they refer to as a "voter receipt." But they're talking about the on-board thermal roll used for the zero tapes and the tally tapes at the end of the day... the sample looked much like an ATM receipt. I would agree that both reviewing and hand-counting of such printouts would be unpleasant (but I'd take it, over no printout at all, if it means I can confirm my vote was recorded as intended).

On the other hand, I saw a demonstration of an AccuPoll printout, which came out on an 8.5 x 11 piece of paper. The print could have been larger, but that's a design change... it was simple to discern one's choices, though.

We've been told that some DREs are capable of printing an "image of the ballot cast" which may be a screen-shot of the summary screen on the DRE. If that is the case, it should be pretty easy to read. Certainly no harder than the summary screen on the DRE... But I haven't seen any of those personally.

Finally, which governs--paper or electronic: One of the things I noted about the Ensign bill is that the specific language about the paper document spells out that it GOVERNS in the case of any discrepancy.

Procedural rules about whether the paper should be the ONLY thing counted or the FIRST thing counted or should ALWAYS be counted are left out... presumably because states (like CO) won't want to be told how they should count; in other words, a "states rights thing" as we heard often in DC recently... and that may end up being a separate battle, which is why groups like CAMBER and CFVI are so crucially important.

Having a paper ballot to count so that the voter can verify that his/her vote was recorded as intended, whether that comes in the form of a VVPB or an optical scan ballot or a hand-mark, hand-count paper ballot, is the FIRST requirement.

By itself, it isn't sufficient.

But it is necessary.

Pam

Audit trail bills deceive voters - ignore problems

The “audit trail” bills are intended to deceive voters into believing that they are verifying the votes that will decide an election. This is not what these bills do.

Instead, voters would be verifying a secondary document that: (1) does not necessarily reflect the voter’s choices, (2) is printed in a form that is excessively difficult for voters to verify and costly for officials to use for independent counting and (3) is not the official ballot and will rarely be counted.

The fundamental requirement is this: each voter must be able to verify that their real votes that will be used to determine the election outcome are recorded as the voter intends and must be able to obtain a replacement ballot if they are not. (There are additional requirements such as: recorded votes must be permanent, every ballot must be secure and accounted for, and the interpretation of votes must be unambiguous, but these requirements are not documented here.)

We continue to ask for the following:

Trustworthy Elections Resolution

WHEREAS trustworthy elections are basic to democracy,

WHEREAS trustworthy elections require that each vote is anonymous, secure, verified by the voter and counted as intended by the voter,

WHEREAS paperless voting machines make it impossible for us to verify that our votes are correctly recorded,

WHEREAS paperless voting machines make it impossible to prove that each vote is correctly counted,

WHEREAS a receipt printer on a paperless voting machine would not solve these problems because the votes printed on the receipt can be different from the votes stored in the machine, and because votes on the receipts would rarely be counted,

WHEREAS accurate re-counting requires that the votes on the original paper ballots be examined and counted, and that the results from a previous count are not known to the people doing the re-counting,

WHEREAS failure to conduct trustworthy elections opens the door to undetectable errors and fraud and destroys voter confidence,

THEREFORE, we voters want to hand mark or machine mark our votes onto full-ballot-text paper ballots, to check our votes before we cast them, to know that the votes on our paper ballot are what get counted, and to have access to proof that every ballot is accounted for and that every vote is correctly understood and counted.