mancalablog

posts to days old
of topics
with text
post

Voting

Paul (5 Nov 2004 5:59): So...guesses whether Diebold's opaque touchscreen voting stole votes in Ohio?

I was thinking about how electronic votes could be made publicly verifiable without losing ballot secrecy, using MD5 sums possibly made with passwords... but I don't think it could work, because
1) if passwords are long enough to be un-crackable, voters can't remember their password
2) whatever MD5-encoder you use might steal your password, and unveil your ballot-identity
:(
zong (5 Nov 2004 10:48): Doubtful, but there are reports that the voting software runs on MS Access (on top of Windows) and is "beta-testing" quality. If there are mistakes in recorded votes, it may as much be "eaten" as it may be "stolen."

(1) That's not necessarily true...

(2) I assume you are talking about the encoder on the voter's end?

Anyway, it all depends on what kind of verification you want. Do you just want assurance that votes haven't been tampered with? Zero-knowledge proofs and such might work? That way anybody can conduct verification (* I don't know anything technical about zero-knowledge proofs). Or do you want to verify whether particular information the voter has matches the one on the machine? This would require the cooperation of and be restricted to the original voter. That sounds more like the traditional PKE scheme, for which you need keys, passwords, etc.
zong (5 Nov 2004 10:56): In any case, these machines are ridiculous. They run out of memory -- votes are lost. They lose power -- votes are lost. Granted, these were not all Diebold machine issues, but I mean... is it so hard to make one of these damned things?
Paul (6 Nov 2004 13:39): I should clarify my idea which depends on solutions to (1) and (2)...

At the time of voting, your name and address plus a password you choose, would be one-way encoded into some kind of checksum. Now you make your votes. Now a row goes in the public "votes" table showing your checksum and your votes, while a row goes in the public "voters" table showing your name and address (which is already a public record by law, btw). The public can see who voted, and that the two tables have equal numbers of rows. Only you can see (by re-encoding with your password) which "votes" row is yours, thereby knowing that your own vote is recorded correctly. You can even check it before you leave the polling place. And as soon as it's posted publicly, tampering would be easily caught.

The trouble is, if a hacker guesses or an encoders steals your password, then they know which vote is yours! :(
Grant (9 Nov 2004 4:30): I think that, whatever happens with regards to that, the very first problem that needs to be addressed is the quality of the programming on these machines. Specifically, it absolutely needs to be verifiably secure. To me, that means it needs to be open source.
Paul (13 Nov 2004 10:16): While I support open-source, I think a stronger approach is needed here, because I don't trust that the code released is actually the code on the machine. :( You just can't trust nobody no more...
(Speaking of which, we might all want to avoid Lexmark)
zong (13 Nov 2004 17:36): Well, let me just say that I have a Lexmark printer at home, and the driver really really sux.
I mean, really sux.
Did I say sux? Yeah, it sux.

And it was for Windows 98. They never released one for Windows XP, but I found one that worked with Windows 2000.

There is one button on the printer that spits out the rest of a page when there is a paper jam. Everything else needs to be controlled through the driver. So, if the driver got confused about the printer state, well, ... time to power cycle by unplugging and re-plugging the power cord. This would be necessary, say, when the printer runs out of paper.

Really sux.
m (13 Nov 2004 18:54): the machine is inherently untrustable, so we should stop worrying about it and only examine its output.

we want some mapping of people->ids where each person knows his own vote ID and nobody else knows it (so, here in california, I was given a piece of paper with a number on it, torn off of my ballot, on which there was a matching number - I don't actually know what the point of this was, since I assume that I won't have a chance to ever again see the ballot with my number on it)

an ID->vote mapping should be made public

each person should be able to look at this and see that their ID matches what they voted for - this is pretty much as paul suggested, only people would be given a password instead of making one up, and there's no longer a need to encode or otherwise sign votes

the only concern at this point is the creation of additional, fake ID-vote pairs
if there was a list of people who voted, with the same number of names as there were ID-vote pairs, then anyone could check to make sure that they were on the list if they had voted, and weren't on the list if they didn't vote. All names on this list could also be checked to correspond with real (live), registered voters at any later point in time.

Is revealing who voted and who didn't bad? If we just have a list of registered voters, then there will obviously be more names than votes for the people who didn't show up to the polls. How do they now keep shady polling places now from filling out ballots for people who are registered and don't show up?

also, I made leek soup for dinner tonight. I no longer think I really like leeks.
Paul (13 Nov 2004 19:16): You and I are thinking on the same track about this, then. Note that the list of who voted is *already* public information, given to anyone willing to pay a lot for it (usually both parties buy copies). So the plan is pretty good.

I suppose my only real worry is that the voter->ID mapping needs to be verifiably one-to-one. That is, no fair giving every Green voter the same ID then saying there was only one Green vote. I'm not sure how/if this can be done while keeping ballot secrecy...
m (13 Nov 2004 21:39): assign the number before the vote is specified
then doubling ids leads to vote-check conflicts
Paul (14 Nov 2004 7:35): On a slightly different approach, Cobb and Badnarik are raising money (have $112k; need $150k by tomorrow) to demand a recount in Ohio. I doubt it will change the outcome, but I hope it might draw some attention to problems like Diebold electronic voting...
Paul (30 Nov 2004 19:23): So we know IRV has major problems, and Condorcet is best for a single-winner election. But what about multi-winner elections?
 Single Transferable Vote (STV) is good in ensuring each faction/party gets its proportional representation, e.g. with n winners, 1/n of the voters can elect one winner. But STV seems subject to the same flaw as IRV: In a 2-winner election between Goofy, Bush, Kerry, Nader, if Goofy gets 50% while the others get half what I hypothesized earlier, again Nader spoils Kerry... (Maybe this becomes less important as n gets up above 4 or so)
 But STV does one thing we need: ignore voters once they're fairly represented. If 51% rank Goofy,Pluto, while 49% rank Nader,Bush, the winners should be Goofy,Nader not Goofy,Pluto. How can this be done besides STV? (NB If 76% rank Goofy,Pluto then Goofy,Pluto should win, and do in STV.)
Paul (2 Jan 2005 13:42):
Paul (6 Jul 2005 22:27): From the Johnson County (Iowa) voter's guide:

"Can a homeless person register and vote?
"Yes. ... If you are on the street, you may use a street corner or location to which you frequently return. You must specify some address or location."
m (6 Jul 2005 23:43): whoah! awesome.
Grant (7 Jul 2005 20:06): It's almost enough to become homeless for, but not quite.
Paul (17 Mar 2006 10:56): IRV sounds to have done a good job in Burlington, Vermont, of convincing people they don't have to take the lesser evil of two big parties. Interesting...
Paul (8 Jul 2006 3:46): Check it out: A "better" voting system called Range Voting. I'm suspicious of their "no-opinion" votes (which they admit "coddle" tiny third parties), but that doesn't seem to break the system. Also of course it's no good for multi-winner elections (like city councils), and strategic voting could get complex. Anyway, check it out.
Grant (9 Jul 2006 23:57): I don't know. I think you'd get a whole lot of people voting 99 for the candidate they like and 0 for some candidate they don't.

And as long as people are going to game it and fill the ballot with extremes, why not just go with Instant Run-Off? It seems pretty good to me.
Grant (10 Jul 2006 0:02): Okay, that was a knee-jerk reaction. I am now starting to see some possible benefits of range voting over IRV. I guess the only bad thing is that it's kinda complicated.
Grant (10 Jul 2006 0:03): Avoid looking like a dummy like me: check it
Paul (10 Jul 2006 17:03): If it's all 0's and 99's it's equivalent to approval voting...I think the advantage of range over over approval voting is just psychological: people can express their views more (for example, someone told me approval voting is bad because you can't discriminate between two candidates you like (also some of us would get some secret satisfaction in giving 98's not 99's to least-worst candidates like Kerry))
Also, range voting is very easy to understand, allowing transparency.
Grant (13 Jul 2006 0:21): Range voting is basically the same no matter what range you allow.

But I get the feeling that somehow 0-99 would seem kind of complicated to a lot of people. Strangely, I think 0-100 might somehow 'seem' easier. But what about something like 0-10? Do you think many people would be resistant to range voting because it seems too complicated?

Or, seriously, do people have any real qualms with it? Or is it more of a battle to get people to care?

Mildly off-topic: I remember having a couple classes where we talked about the benefits of the electoral college system & two party system and stuff. But, uh, I think most of it was kinda bunk. I also remember Ms. Miller (among others) saying that we'd never realistically have another election where a president won the electoral college vote yet lost the popular vote. I remember thinking that was bunk, too.
Popular Vote (13 Jul 2006 0:22): I'm changing my name.
Paul (9 Sep 2006 3:33): electoral-vote.com is back!, except their server is totally overloaded. (electionprojection.com has similar info, I suppose)
m (9 Sep 2006 7:09): I thought Cantwell's race was expected to be a bit of a close one, but electoral has it as pretty solidly hers -- are they right?
m (9 Sep 2006 7:12): Also projection thinks that Lieberman is going to take Connecticut (I can no longer check electoral vote to see what they say about Conn. I see what you mean about their server) and when I try to click on the state to see what they've got to say they ask me for a password and something about premium content. buh.
Paul (10 Sep 2006 4:46): A look through `host electoral-vote.com` leads me to its Coral cache here. I think any blue-prez-'04 states (like WA=Cantwell) are surefire blue-senate-'06 states...unfortunately we need way more than that.
m (11 Sep 2006 6:20): bzzht
m (11 Sep 2006 6:26): So I guess Cantwell's the likely winner, but they're both fumbling around a bit. That 11 million dollar spending thing -- buh. Lobbyists. Fucking congress.
G (7 Nov 2006 22:39): YESSS.
Paul (8 Nov 2006 16:27): Yes! Thank goodness!
Paul (9 Nov 2006 18:01): from tongodeon:
Paul (10 Nov 2006 15:17): doh.
Paul (11 Nov 2006 6:54): Uh-oh. I just thought of a sinister reason for Allen's quick concession in VA: If he exercised his right to a recount, it would draw attention to the unaccountable-and-probably-rigged computerized voting machines.
Paul (1 Feb 2007 18:03): Here's To Optical-Scan Voting! and down with charlatan touchscreens!
Paul (13 May 2007 5:28): What would be awesome is if General Batiste could be the Republican nominee in 2008. Of course I'd far prefer a Democrat, but we desperately need the Republicans to have integrity like him.
R (20 Sep 2010 17:26): WA-peeps! Tim Eyeman has a new initiative. So, right there, you should basically be aware that there's something important to vote against come November. You know how CA is totes' effed? In large part because raising taxes requires a 2/3rds supermajority? This would do that for WA! I mean, when's the last time the legislature was way too cavalier about raising taxes? It's already tends to be a pretty risky political choice. If you have any faith in an elected government, you've got to allow the elected to try to govern. Vote "No" on 1053.
R (27 Sep 2010 16:37): Russ Feingold is awesome. And he could really use your support.
Paul (19 Aug 2011 19:04): My vote and comment on wikimedia image filter referendum: 0,0,?,0,0,?

A vote without a discussion is not a sound way to measure community opinion. It is disturbing that the instructions say "read the statements and decide on your position". A sound voting process should say "read the arguments for[link] and against[link] and the discussion[link]".

I managed to find such statements, and I found that this argument against the proposal is compelling:

If the images are categorized, ISPs/schools/etc would easily find technical ways to censor based on the categories, and they would be very likely to remove the end user's choice about what to view.
m (3 Dec 2016 16:32): "How did all the people know who to vote for? Did Multivac tell them?"
"They just used their own judgment, girl."
post

Instant Runoff Voting

Paul (16 Jun 2004 5:03): Whether you already support Instant Runoff Voting or have never heard of it, visit irvwa.org. They need 200,000 signatures of registered Washington State voters by December 2004.
The idea is: when you vote for, say, President, you can rank your choices: if your first choice wins, okay; if not, your vote counts toward your second-choice. Cambridge, MA has been using this system for 60 years to elect its 9-member council (the IRV rules for electing to multiple-seat bodies ensure that, say, if 33% of Cambridge wants Republican councillors then the council will include 3 Republicans).
Paul (14 Nov 2004 11:35): Having read a lot at electionmethods.org, I no longer strongly support Instant Runoff Voting. They prove not only that Condorcet (pairwise) voting is better, but also that IRV is hideously bad whenever a third candidate is a viable winner.
Paul (14 Nov 2004 12:08): Consider the following fantasy:
44% - 1st:Nader, 2nd:Kerry
22% - 1st:Kerry, 2nd:Bush
33% - 1st:Bush

IRV would say: Kerry has fewest 1st-choice votes, so loses; now Bush wins.

But note that the pairwise defeats here are cyclical. This is an inherent ambiguity of the voters' preferences:
Nader beats Kerry (44% vs 22%)
Bush beats Nader (55% vs 44%)
Kerry beats Bush (66% vs 33%)
Condorcet drops the smallest defeat, Nader's over Kerry, then Kerry wins.

And if the Bush voters ranked Bush, Nader, Kerry? Then
Nader beats Kerry (77% vs 22%)
Bush beats Nader (55% vs 44%)
Kerry beats Bush (66% vs 33%)
dropping the smallest defeat, Bush's over Nader, then Nader would win.
m (15 Nov 2004 0:26): wait - by smallest defeat, you mean defeat in which the victor got the lowest percentage?
m (15 Nov 2004 1:10): Ok. I thought about your description of condorcet for a while, and I'm convinced that in an election with three political factions, no one group can swing the election their way by changing how they rank the candidates (which I initially thought would be the case after looking at your example where bush voters didn't rank the other candidates). I guess two groups can collude and make things go their way indefinitely, but that's the same thing as reducing the number of groups.

When there are more than three, it's confusing.
m (16 Apr 2015 9:57): Bruce Schneier on the problems with this year's Hugo Award:

(1) I think the best choice would be to do nothing. It's not at all obvious that this is anything other than a temporary aberration, and that any fixes won't be subject to a different set of abuses and need to be fixed again. I think the worst situation would be a series of rule changes in a continuous effort to stave off different abuses. I don't think highly of a bureaucracy that tinkers with election rules until it gets the results it wants.

(2) If we choose to ignore (1), the second-best choice is to modify the electorate. The problem isn't the rules of the vote; the problem is that a voting bloc was able to recruit voters from outside the usual community. Trying to fix that problem by changing the voting rules is very difficult, and will have all sorts of unintended consequences.

(3) If we choose to ignore (1) and (2), this is the thread to discuss how to fix the voting rules.


and here's the follow-up where he summarizes possible fixes to the Hugo voting mechanism
m (16 Apr 2015 10:00): But I mostly just wanted to point out that George R.R. Martin's blog IS SECRETLY A LIVEJOURNAL (maybe not so secretly)
m (16 Apr 2015 10:47): BLAST FROM THE PAST

Current Location: Santa Fe
Current Mood: depressed depressed
m (17 Oct 2016 12:40): holy shit I think Maine is going instant runoff