Alaska's Hand Recount Provision

[borealowl]

HB 94, an omnibus election bill which includes one of the best random hand recount provisions of any state, goes into effect today!

One precinct in each district (about 40) randomly chosen and representing at least five percent of the voters in the district, will be hand counted.

We shepherded this through a Republican-dominated legislature and governor, with the elections division chief testifying that it wouldn't cost much. It was a very bi-partisan thing - Alaskans of all stripes have this thing about civil liberties, and voting rights.

Here's a url with some other states that have audit provisions -
http://www.verifiedvoting.org/article.php?id=5816. It will be updated soon with Alaska's info, I'm told.

As someone who has been following U.S. election chicanery with distress since November 2000, I took the approach of working on my own state rather than focussing on trying to affect national legislation (although I've made quite a few phone calls there, too) - but I'm hoping more people worried about American elections will make sure their own state has its act together - you can really make a difference when you can go knock on doors and bug elections officials and state legislators. In states with bad problems really there ought to be large gatherings, strikes, or whatever it takes to bring public attention to the fact that votes aren't being counted properly anymore. E-mail groups for activists in each state can be really helpful. It was hard for me to believe for example with all the attention on Palm Beach that the voters didn't just rise up in anger when they brought in the paperless touchscreen computers a few years ago - surely some of those disenfranchized voters understood the Trojan horse character of these "improved" machines.

In Alaska, luckily, it's more a question of trying to forsee problems than fix them. We're not really broke. We recently had a successful recount for Senate which didn't uncover any problems. The elections officials were open, the precincts for hand recounts were chosen randomly, out of a hat, and the entire state was recounted by machine using newly programmed cards. It just wasn't Ohio.

Because of the foresight of our previous lieutenant governor, Fran Ulmer, Alaska has rules against campaigning by elections officials. Also, for years the division has prided itself on its independence. Diebold provides the machines, but no Diebold technicians are involved in elections or are even in the state. An Alaska employee programs the cards for each election, and the division does its own counting. Also there are unusually stringent crosschecks perfomed on the cards beforehand by bipartisan committees statewide and again regionally. (Some of you working on improving your own states' elections might want to call the Alaska Division of Elections for details on the checking procedure, etc. I don't know whether any other states maintain an equal level of independence from the vendor - ???)

I don't want to sound as if I think our system is perfect - I don't like a partisan secretive company being involved at any level, and I'm annoyed about the purchase of DRE's for handicapped people to supplement the optical scan machines - better solutions are available - but we do have a paper ballot requirement and until the new terminals can be outfitted with printers, they've being mothballed.

Here's more info about Alaska's new law:


http://www.legis.state.ak.us/basis/get_bill.asp?session=24&bill=hb94&submit=Display+Bill+Root


CONFERENCE CS FOR HOUSE BILL NO. 94

IN THE LEGISLATURE OF THE STATE OF ALASKA

TWENTY-FOURTH LEGISLATURE - FIRST SPECIAL SESSION

BY THE CONFERENCE COMMITTEE

Offered: 5/15/05

Sponsor(s): HOUSE RULES COMMITTEE BY REQUEST OF THE GOVERNOR

A BILL

FOR AN ACT ENTITLED

"An Act relating to qualifications of voters, requirements and procedures regarding independent candidates for President and Vice-President of the United States, voter registration, voter residence, precinct boundary and polling place designation and
modification, political parties, voters unaffiliated with a political party, early voting, absentee voting, ballot design, ballot counting, voting by mail, voting machines, vote tally systems, qualifications for elected office, initiative, referendum, recall, and definitions in the Alaska Election Code; and relating to incorporation elections."

BE IT ENACTED BY THE LEGISLATURE OF THE STATE OF ALASKA:


Sec. 15.15.430. Scope of the review of ballot counting.
(a) The review of ballot counting by the director shall include only
(1) a review of the precinct registers, tallies, and ballots cast;
(2) a review of absentee and questioned ballots as prescribed by law; and
(3) unless the ballot for the election district contains nothing but
uncontested offices, a hand count of ballots from one randomly selected precinct in each election district that accounts for at least five percent of the ballots cast in that district
(b) If, following the ballot review set out in (a) of this section, the director finds there is a discrepancy of more than one percent between the results of the hand count under (a)(3) of this section and the count certified by the election board, the director shall conduct a hand count of the ballots from that district.

[Wilms]

1. Quite cool.

Are the statistician happy with it?

[OnTheOtherHand]

2. umm, maybe not

(dunno if I count as a statistician, but hey, that sheepskin oughtta be good for something....)

Actually, this is substance. Why _one precinct_? That looks sort of like a "license to kill" in any and all precincts that are likely to be less than 5% of the total vote. Or are the election districts so small that I should relax about that?

I'm a bit quizzical about the 1%, too, but it depends on how it fits into the broader context. I suppose there is already a recount provision for races that are quite close? (Of course we do need a random recount provision even for races that don't appear to be close.)

[borealowl]

4. question

Well, the statisticians seemed happy but there weren't too many involved, basically the legislators introducing this amendment just hit on something that seemed to make sense. It derived from the procedure decided on during the recount, which people liked - instead of picking 40 random precincts out of a hat, pick one randomly from each district.

I guess it would be possible to rig the election in some precinct with less than five percent of the voters, but in order to make a difference in the overall election result, that precinct would stick out as out of whack with patterns in other similar precincts. I think the idea was mostly to exclude from the hat the small rural districts which don't even have machines, which still use paper ballots to vote.

We didn't really expect to get this provision through - I'm not sure why not, there's really no good argument against a good audit - but I guess the stonewalling in other states had us spooked. Anyway, mostly we were hoping it would remain in the bill through multiple committee reviews. But I don't know what I'd change if I could, and I think it will have the intended deterrent effect, taken together with the other safeguards in this state. It will also be reassuring and make us feel less of a need to request a total statewide recount in a tight important race like the last Senate race in Alaska - it's not easy to raise the required fee either, which the same bill raised from $10,000 to $15,000. That was the tacit compromise - good audit provision, but more daunting for average citizens to try to pull together the funds for a full recount. Still, the amount of the raise was reduced - the bill in its original form, as submitted by the governor, asked for $50,000!

If I had my druthers, I'd run elections like Canada, by hand, but once elections officials have these optical scan machines they love em, it's much less tedious than hand counting, and it would be hard for them to give up the machines. Like getting SAT testers to go back to hand scoring!

Also, the elections people say these machines are more accurate than hand counts, which, if there's no monkey business, is probably true.

Note: when these were bought in, I think, 1999, our elections people looked at touchscreen computers and rejected them for the simple reason that there was no way of checking the results. They chose the optical scanners instead.

That was the common sense response, before this whole debate got off the ground really (before the Diebold/ESS/US govt propaganda campaign to get elections offices to accept paperless voting took hold).

[OnTheOtherHand]

5. sounds pretty good; as long as I am kicking the tires...

I was 'talking' about this general topic with Bill Bored a while ago; we were discussing Kathy Dopp's audit calculator, which basically makes the calculation: "If there are X precincts of which Y have vote corruption, and you randomly audit Z precincts, what is the probability of auditing at least one of the corrupted precincts?" That calculation assumes that you can tell for sure whether a precinct has been corrupted.

With one-percent "wiggle room," things get a little more complicated. Frankly, it shouldn't matter except in a very close race. (And I kinda suspect that if a state allows no wiggle room at all in the random recount, it will mostly end up with crooked recounts. But I think Bill tends to disagree with me about that.)

Probably an 'ideal' system combines a random recount provision; a provision for a wider recount in a close race; and a provision to obtain a recount of (a relatively small number of) individual precincts where the results seem bizarre -- all of which entails good ballot security in order to succeed.

I'm speaking very generally, not about Alaska. I don't know all the provisions of Alaska recount law, and in any case, Alaska definitely has some unique circumstances! About 300,000 voters scattered from here to eternity.... Given that you know the whole picture, I bet you're right that this new provision is an effective deterrent. (And, like you, I am not a hand-counted paper ballot purist, although op-scan definitely needs careful auditing.)

Thanks for all the information. My cousin used to live in Fairbanks, and I've puttered around Alaska three times -- but only in the summer, which hardly counts. My brother and I hiked in Denali _State_ Park for several days -- excellent views of McKinley, no tour buses. But I digress. ;)

[Einsteinia]

3. Great info! I'm submitting it to the CA "Gold Star"

Audit protocol's request for submissions.

If anyone has their own improvements to an audit protocol, see the Gold Star icon on the left hand column at http://www.califelectprotect.net

THANKS

[Bill Bored]

6. One per district might be OK but why exclude the smaller ones?

Unless you mean at least one per district until you reach 5% of the voters. And why "voters" and not actual votes?

The idea is to find a corrupted precinct, which could the first of many, or just enough to reverse the outcome of a close election on its own. You might not do that if you systematically exclude all the precincts with <5% of the vote count and if someone wanted to hack the vote, they'd know that the smaller precincts would never be audited. This is the reverse of what Blackwell did in Ohio when he excluded larger precincts to get no more than a 3% total. Either way is not a random selection.

What do you think?