Slashdot Mirror
Early Voting Problems, Open Source Alternative
Techdirt makes note of some problems cropping up already for early voters in the presidential election. CNN covers some of the issues, including machines in a West Virginia county which recorded some votes incorrectly because of an alignment error. A lengthy discussion of the problems was also featured on NPR. Reader Rooked_One points out a related story at NPR about a voting program called PVOTE, written in Python and only 500 lines long. "Pvote is not a complete voting system. It is just the software program that interacts with the voter. Other necessary functions, such as voter registration, ballot preparation, and canvassing, are not part of Pvote. It is especially important that the voter interaction be correct because it is the only part of an election that must take place in private, whereas all other parts of an election can and should be subjected to public oversight and verification."
Even if you have a short programm you still cannot guarante that it works because there's still a system surrounding it. In fact you could even manipulate the CPU hardware to give you false results.
The _only_ practicable and moderately secure way to do an election is by pen and paper and manual counting. It's done all over the world and it works near flawlessly. Everybody, not just programmers, can watch the process and see what's happening. There's no "black magic" involved and it's completely transparent.
As soon as there is some form of technology involved, people will cease to understand it, therefore making the whole system intransparent and prone to manipulation.
Everything in Oregon is weird. We can't pump our own gas, we don't pay sales tax, and we do all of our voting by mail. It makes no sense, and it's ripe for corruption (though nobody has called the "C" word so far. At least not lately)
But it's kind of nice. No computers, no machines, just fill out your ballot and mail it in. I got my ballot in the mail yesterday. I've plenty of time to research the state and local ballot, so I can make an informed decision.
The Internet is generally stupid
That's good. It means that the trusted computing base is only a POSIX kernel, a C standard library and a Python runtime. Only about 100MB of source code to audit to ensure that this 500-line program runs correctly.
I am TheRaven on Soylent News
Why not simply make voting a public action? I'm voting for Obama. There. Done.
Because that opens it up to vote buying and voter intimidation.
If Bill Gates promised everyone $1000 to vote for him, he could buy 56,000,000 votes which would put him in the White House.
For intimidation, you don't have to intimidate everyone - just a small percentage in a few key states. Imagine if the CEO of WalMart told their employees, "If you work in Ohio, and don't vote the Right Way, you'll be fired." Even if it isn't an official, enforceable policy, it will still have a large percentage of employees worried for their jobs come their next performance review - and they will vote accordingly. With over 2 million employees, even 10% of WalMart employees changing their vote could affect the outcome of the election.
That's assuming that people with baseball bats don't just show up at your house and tell you How You Will Vote - Or Else.
Reading code is like reading the dictionary - you have to read half of it before you can go back and understand it.
Why can't the US do what we do in Canada? You don't have to make this complicated.
In Canada, we show up to our polling station with our voter card, show the card and receive a ballot. We take the ballot, which has the names of the candidates and their party in large font very clearly, and put an X in the big circle beside the candidate we're voting for.
Thats it! No fancy machines, no complicated forms, and no computers to go wrong or be hacked.
See this image:
http://www.elections.ca/yth/images/sample_ballot.gif
I'll Find You Peer, If It's The Last Thing I Do!!!!
Almost every place I've ever voted has had optical scans, generally of the connect the arrow type. They mail out sample ballots (marked sample and on different paper, no funny business there) several weeks ahead of time. You walk in with your voting card and/or proof of ID (the laws are getting stricter and require both in many places, now), the volunteers (almost always old people, setup so representatives of both major parties are present) mark your name off as voting and hand you an official ballot. This ballot should match the sample you got in the mail, or there'd be PR hell to pay.
You walk over to the booth, generally a portable table with a cardboard privacy screen fitted around the top. There's a pen of the proper type in the booth -- you can ask to have it replaced if necessary. For each race or proposition, it tells you how many you can mark -- for some state races you can select multiple candidates and the top X number are picked -- and you mark what you wish. For most candidate races (in most states) there's also a blank entry you can mark, and write-in your choice.
You don't have to vote for all races or propositions. If you screw up, you take the screwed up ballot back (using the privacy procedures below) and they give you another. (I've never actually screwed up, but this is what the prominently posted instructions say to do.)
When you are done, there's either a privacy cover (if the cards are printed on both sides) or instructions tell you to turn the marked sides in. You leave the booth and return to the sign-in area, where another volunteer takes the ballot and feeds it into an optical-scan machine right there -- you watch them do it and hear it beep and increment the ballot count. Again at this point, if it fails to read (tho I've never had that happen), you can get a new ballot and try again. The ballot itself is deposited in a lock-box for recounts, if necessary.
Many states have a no-reason-necessary early voting allowed policy. You can either request an absentee ballot and either mail it in or take it to an authorized polling place up to poll closing time on the date of the main vote, or go in and early-vote at the county recorder's office. A few elections ago I did just that, requesting and getting an absentee ballot in the mail, which I filled out, sealed in the provided envelope, and dropped by my normal polling place on the day of the vote. They had a lock-box for them. It was much more convenient than voting as normal, but I missed the voting ritual and it felt kind of weird watching the results come in that nite having not actually voted that day.
States differ in how they check the ballots, but Arizona (where I am now) at least, requires an audit of several (IIRC two) percent of the precincts, randomly chosen (the "random" process of choosing them is encoded in the law, with at minimum witnesses from both parties, both to the choice and to the verifications, the audited precincts are not known previous to the vote so can't be avoided that way). These audits hand verify the count of the optical scan machines.
This system seems pretty reliable to me. I still can't understand why the entire nation doesn't just go opti-scan, as the machines can be used to count and get quick results as necessary, while the paper trail is there for anyone wishing to verify things.
The biggest problem I've heard about, doing it this way, is the lock-boxes disappearing a couple of times. Each one holds a few hundred votes. Of course, there's an accountability trail, but as they say, **** happens. Unfortunately, that's a problem for pretty much any after-the-fact verifiable system. But those cases are few and far between, it seems, and I've not heard of any of them actually affecting an outcome. There have been a few cases of other oddities as well, but nothing even close to the unverified touch-screen issues that seem to come up every year.
BTW, I've worked with touch screens, and whoever came up with the idea of having the untrained public v
Duncan
"Every nonfree program has a lord, a master,
and if you use the program, he is your master."
R Stallman
Voting isn't a (*(&^ing nail, stop trying to throw your coding hammer at it! This has gotten to be an example of obsessive compulsive disorder with these schemes. This is crazy. Open source or not, unless there is an independent deep forensics investigation of every single computerized voting kiosk at the end of the vote period, including disassembling the chips on the machine and all that stuff, it can *not* be verified in a timely, cheap and thorough manner. Oh, a "paper trail"? Why yes, let's look at that "new idea" to "insure" and "verify" the computerized vote! A plain empty box CAN be verified at the start of the voting day by many people looking inside and going "yep, empty!" And a paper trail is exactly what you get start to finish with plain paper ballots, no stupid computer and expense needed. Yes, examples in the past of ballot box stuffing, still way easier to keep tabs on it then running everything through obfuscated layers of chips and code. Paper ballots and empty boxes are WAY MORE the lesser of (in)security evils when it comes to voting, let alone being loads cheaper when it comes to co$t$. Empty box per precinct=ten bucks max, what do these computerized schemes cost, and how much has been wasted on them so far and how much "irregularities" do we get to read about and enjoy before this sinks in as just a bad idea overall?
First off Germany has a relatively simple ballot, its only complex because it's evaluated in a complicated way. The US to begin with is large with diverse kinds of governing bodies, and has far far far far more elected offices. So comparing it to German elections is silly.
Second Pvote is not 500 lines. it's 500 lines sitting on top of hundreds of thousands of lines of interpreter code, device drivers, and the tens of millions of lines of Linux.
Third writing a voting system, while non trivial, is not the hard part.
The hard part is twofold
1) creating a viable bussiness model for it's distribution, component agregation, certification, and service it.
2) designing a voting PROCESS so that you don't have to trust the third parties that build or maintain these or the people that operate them. Things have to be transparently secure.
Now the OVC system OPen voting Consortium has had a python based system for years. it's open source too. But more importantly it is designed so we dont' have to trust the programmers (it produces an intermediate paper ballot and physically separates the vote selection hardware from the vote counting hardware ---just as optical scan does.) And it has a well thought out and viable bussiness model that will allow for it's practical distribution and maintainence.
That is what the world needs. so if you want to help. Donate to OVC. They are struggling right now not because they can't write code, but because they have to win acceptance at the state level before any company is going to start marketing the system.
OVC has a very clever bussiness model in which the software is free and open, but companies support it's development through fees paid to certify their OEM component based systems as compliant with the OVC standards.
Some drink at the fountain of knowledge. Others just gargle.