|« Darfur, Pass 3||Do McCain's people check his statements? »|
[Update 11/22/2008]: What really surprised me is the stunning lack of interest in this idea after I've posted it, and I've heard nothing about it, any place I've submitted it has basically rejected the idea out of hand and has made no comment on it. With all the comments I hear in so many places complaining about difficulties in developing a scalable election system that is auditable and fraud-resistant, to hear nothing tells me that nobody is really interested in a solution, they just want to complain.[End Update]
I took a nap this afternoon, and woke up this evening with an epiphany. After I posted it here, I posted a summary on Slashdot and offered it up for publication. After having posted the summary there, I had some ideas to clarify it so I'm changing this item slightly to emphasize or enhance some points I made earlier, and to add points I thought of later. So this article may change over the next day or two. My usual practice is to indicate when I update an article, but since it may change several times and it's the idea that counts, not exactly when the item was put in, I'm going to suspend my usual practice (of marking when I make changes) for this article.
I won't indicate places where things change over the next day or two, as any changes should merely be to expand on prior points or to clarify existing points. I think I've covered basically everything in how it works, anything further is just amplification.
I thought of a really simple way to provide low-cost, trustworthy, secret elections, where you can verify what the voter intends, and get a fast total at the same time. You can even make it work for places where there is no electricity, neither the voter nor the election officials can read, and even if the election officials can't count. And by fast, I mean under 30 seconds for each question (or office) on the ballot after the completion of the election. Here's what you need to do this:
For each question - a public office, a ballot proposition - you need for each voting booth at each polling place:
- One transparent box with a lid for each choice plus one additional box if there's a write-in candidate.
- If you want them not to have to indicate if they didn't vote, you also have one box for "abstain" or "no vote on this question."
- Lock or seal for the box. Can be one of those one-way cable ties that you have to cut to remove
- For "Yes/No" questions like bond issues, you only need two boxes.
- One token such as a coin, a steel slug, or other similar object, for each voter, for each question. Doesn't have to be a coin, but makes it easier to count mechanically as there are commercially available coin counting machines, some that are hand-cranked.
- Cover for the boxes for each question so the contents of the boxes can't be seen by the voter.
- Cover for each voting booth so the voter can't be seen while they're making their choice.
For each polling place, you need:
- One scale for each question per voting booth, or possibly just one scale.
- Box of paper and pens for write ins.
Person reaches into a dispenser that releases a token. The "dispenser" could simply be them reaching out of the polling booth to pick up a token, to someone handing them a token once it's confirmed their hands are empty, or you could simply make sure there was only one token at that voting booth for that question. For each question there is one box for each choice, plus one box for "write in" where write-ins are possible. They then drop the token into one of the boxes that is their answer for the particular question.
For the 2008 Presidential election we have the question of "Whom do you want for President?" For the answers we have "McCain", "Obama", any other minor-party candidate boxes who have qualified, and "Write in."
For a bond issue, we have answers of "Yes" and "No."
You need one transparent box for each choice for each ballot question, for each voting booth. So if you have an election of 3 offices each with 4 candidates, and 2 ballot propositions or bond issues, we'd need 5 boxes for each office (one box for "write in"), and a "yes" or "no" box for each ballot proposition. So we need 19 boxes for each voting booth, and for each voter we need 5 tokens. Tokens might be different for any specific office or proposition to keep them from voting twice on any office or they only get one token at a time for each office or ballot proposition.
The voter selects which choice they want for each office or ballot question, and puts their token in the box for their choice, either a particular candidate or a yes/no proposition.
If we want to know if they voted, these boxes for that question all can be supported on a board on a scale, where the scale is outside the voting station ("voting booth"), so we know if the person has voted, but we do not know what they voted. Since all of them are on the scale, all we know is that the weight has changed, we don't know which one, except if the person declares write in, then they get a piece of paper to write in their vote. (You don't know what they wrote, only that they did choose write-in.) Also, use of a scale at each question prevents sneaking in votes since, at any time, the number of voters and the weight on the scale of each token times the number of people who have voted have to be the same or fraud is obvious. We always know exactly how many votes have been cast, but we do not know how they have been cast until the boxes are removed from the voting station.
As the voter has nothing in their hands when they reach in the box, they can't stuff the ballot box with fraudulent ballots (tokens). They can see the boxes they select, nobody else can, so their vote is secret. We can, if we wish, put all of the boxes on a scale, so that they have to release the token to exit the booth, and thus can't steal it. If they change their mind on voting that question they give the token back, or we have the "no vote" box for them to dump the token without telling us they didn't vote.
Or if it's just a booth where they can reach in, we just confirm that their hands are empty when they remove them, then they can't steal the token, and we don't need a scale at each question. Or use a metal detector if they try to steal the token you get an alarm.
In that case if we make sure they can't steal the tokens (like to have someone sneak them to another polling place and use them at another site to change the election there), or we don't care if they do steal them, we just need one scale for the entire polling place to use at the end of the election. As long as they can't get extra tokens into the booth, they can't vote more than once.
The tokens don't have to be coins, you could use ball bearings, bb shot, bricks, or anything else you can both count and weigh that is consistent in size and weight. Not paper, however; we want something that won't absorb water so that if the voter's hands are sweaty it won't affect the outcome.
The tokens don't even have to be the same. You could even go so far as to have, for example, an election with four candidates. Assign each one of them to 1c, 5c, 10c and 25c, and the voter picks one of those coins, then drops it in the box. Now all you have to do is sort the change, and whichever coin was used the most is the winner. All you have to do is make sure they can't get extra tokens into the booth, either of the same value or of different ones. This also includes making sure they can't get more than one of the different tokens too.
On the other hand, if they can get different tokens, this allows for things like approval voting where you can vote for all of the candidates you favor and not vote for the ones you don't. Or in cases like corporate elections by stockholders where they can have as many votes as there are candidates, then they can vote as many times as they have tokens, whether all for one candidate or split their vote.
If they vote write in, they get a piece of paper to write their selection and put it in a container; we count those later. Write-ins are usually not a large part of most elections. If they are, then it's the same as what you have now and it doesn't change anything.
End of election, you remove each box, you don't have to even count the tokens or even open the box, all you have to do is weigh the boxes! Each box has the same weight empty, each token has the same weight, and whichever box weighs the most is the winner. And since the weight is an exact multiple, you also do get the exact count without even opening the boxes! And if you use tokens, and you want to verify the results, you open the box and dump the box contents into a coin-counting machine calibrated for those tokens. And since the box is transparent, you can even see the tokens and still might be able to count them without opening if the number isn't large or the results clearly obvious.
Unused votes can't disappear because they can weigh the boxes containing unused tokens to be sure that they weigh as much as the number originally issued minus the number of ballot tokens given to voters.
Also you don't even have to know how to read to vote! They can put the picture of the candidate, their identifier (party symbol, public color, a candidate number) or anything else showing whom they are, next to, on or above the box, and just remove the sign (if it's on the box) after the election. For a ballot proposition, they can have like a big red X/ thumbs down/frowny face for no, and a green check/thumbs up/smiley face for yes.
Hell, you could have six year olds running for school office in first grade, the voters don't have to be able to read, and the election officials can be the kids themselves, all they'd need to hold the election is to be able to set up boxes and weigh them!
And the officials counting the results don't have to know how to read either! When the boxes are weighed, whoever's box weighs the most wins. If you have to count the ballots (because the needle on the scale doesn't quite make it clear), you don't even have to know how to count! You break the seals, you empty each box onto one of two (or as many as there are candidates) tables (being sure they can't spill off). At each table, each person holds up one ballot token for each candidate, and drops it back in that candidate's box, and they repeat this until one of them runs out of tokens. If you have multiple candidates, one of them runs out first, so they are dropped, and so on until whichever one of them still has tokens left on the table and no one else does, that one with unmatched tokens left has won the election. This presumes winner takes the election, if you have minimums then you still have to count. By putting the tokens back in the box, we can do recounts any time we want if necessary.
- You get the total immediately because you don't even have to count the tokens, you can weigh the boxes. (This is how they handle the huge amounts of cash they generate at casinos in Nevada and Atlantic City; they don't count the money, they weigh it.)
- If you have a dispute, you open the boxes and either count the tokens manually, or you run them through a token counting machine.
- If we use coins for this purpose, we can use commercial-off-the-shelf coin-counting equipment. Which can be as simple as a mechanically cranked coin counter with a meter on it. I once saw a laundromat using one of these over twenty years ago.
- Extremely cheap to do, as none of the equipment needed is expensive to start with, none of it is going to wear out (other than perhaps the scale becomes inaccurate, or a plastic box cracks or if you're using a lock, it becomes rusted) and if it does it's cheap to replace. If the scale breaks, you use another one, or you don't even need a scale, you can still count the tokens.
- Voters don't even have to be able to read to be able to vote, they only have to be able to pick up (or be handed) a token to put into a box for the candidate (or answer for a ballot proposition) of their choice, which can use symbols on or near the box.
- Election officials can hold the election even if they can't read and can't count, they can still report who won and send the sealed ballot boxes to the control center for actual counting.
- Requires no electrically powered components.
- Works in any size election, large or small.
- Scales easy; does not get more out of hand or hugely expensive with larger elections or more polling places, all you need are to buy more cheap boxes and cheap tokens.
- Allows for very simple voting booths. No 'butterfly ballot' errors.
- Instead of counting each ballot (which takes a long time), you're simply counting all the ballots in each box simultaneously, so even if you have to do the counts centrally, all you need is the time it takes to weigh each of the boxes, not the time it takes to handle all the tokens. With one hundred butcher's scales, you can easily have two hundred people handling 500 boxes a minute (one box each 20 seconds). (You have two people on each box so both can watch each other.) If you had to count the results for two ballot questions with two candidates each, at 5,000 polling stations, you can do it in five hours or less. And the total should be exactly right.
- Works only for simple elections with small number of answers for each question, say, 10 or less; for an election with a lot of candidates this might become unwieldy. California Gray Davis recall where there were more than 100 candidates might not work. For those questions you could go back to a paper ballot, and probably would have to anyway. Or you could simply have one booth for the entire polling station for that particular question if there were a lot of choices.
- Doesn't work as-is for multiple-choice questions, like school board elections where you have eight candidates and 'vote only three' unless you figure a way that they can only insert one token into any specific box. But you can do that by fixing it so that dropping a token in the box seals its lid so no further tokens can be inserted. You could use something like a coin push slide on a vending machine. (But then you have mechanical parts that can break down. Better to use something like a U-, S-, or J-shaped curved plate that when you drop a token, it pushes the top of the plate and around, causing it to block the opening.)
- Not sure if it will work in elections where the winner is something other than highest vote getter wins; if one must get at least a certain number of votes or percentage of the election, it might not work. On the other hand, I think this is only a problem if you can't count.
- Voters choosing write-in can't do so secretly. This is where you use an additional scale or trigger on that box, and if you want, if they choose "write in" then they get a piece of paper and the other candidate boxes lock so they can't change their mind, similar to the point I make above for handling a multiple-choice question. (They could always write in one of the original choices if they do change their mind.)
- Voter can't change their mind once they drop their token. You could put a cover over all the choices, if voter is satisfied, they push the lever forward to dump the tokens, pull it back to retrieve them. Or the other way around.
- Doesn't answer the question of what to do if there's a tie, but neither do any of the current systems.
- This is only effective for elections where the voter shows up at the polls to vote, it does not work for absentee elections. But you can, if the poll workers can be trusted, vote what the paper ballots said. If that's an issue, they use a separate polling booth (and the boxes for that) for the absentee ballots.
The nice thing about this is it requires no expensive hardware, you can do the whole thing even in places with no electricity, all you need are reliable scales - and we've had those for thousands of years - uniform boxes to hold the ballots, and some tokens which can be counted and weighed. The weight gives you a fast total, counting gives you a verified voter count if you need it. The boxes have seals so the contents can't be tampered with. If the tokens are not something a voter would have (specialized token or something other than a standard coin) then the voter can't stuff the ballot box.
If each polling place only receives as many tokens as there are voters, there's no way for the polling place to forge vote counts. Can't hide unused tokens because all the scale totals have to balance. If there's no question over the vote, you don't even have to empty them until it's time to hold the next election.
It's cheap to run. I've bought plastic tubs at Home Depot, the ones that are 3 cubic feet with lid cost about $4. Small containers might be as little as $1 or less each. Padlock for each box is probably $1, or if we use one-way cable ties, they're about a penny each. Tokens might cost as little as 1c each in huge, bulk quantities, or as much as 25c if you were using, say, quarters. But if you were using coins, you'd get them back after the election was finalized. Maybe need $10 for a blanket to cover each question's ballot boxes so they can't be seen by the voter, and maybe another $20 to cover each voting booth so we can't see what the voter is choosing. Maybe $30 for each scale, and if you set up the controls right, you can reduce it to a single scale for the whole polling place.
So you don't need $5,000 voting machines for each voting station, you need $6 boxes for each choice in each question, plus a few dollars for the tokens, the scale or scales, the covers over the question boxes so the voter can't see how other people voted, and the cover over the voter so the election officials can't see how any individual voter cast their ballot. And you can simply use the same cheap equipment over and over again at each election.
Oh yeah, the covers on the ballot boxes have to be such that tokens can go in but fingers can't so tokens can't be removed or people can't reach in and count the prior ones. Must also be designed that tokens can't come out until the seal is broken and the ballot box is opened. The boxes might have a lining so that you can't hear tokens striking each other, or put a metal plate on the bottom so you can hear the coin strike something metal - the bottom or another coin - so that no matter which way prior voters have decided, even if nobody before voted for that choice. (Otherwise voter #2, if their token happens to strike the one from voter #1, knows that voter #1 voted the same way.)
In fact, to have elections where the voter couldn't steal tokens or add extras, you could even do this with something similar to an old-fashioned lever voting machine, where you push buttons to release tokens for each office or ballot question, and they would dump into the boxes when you threw the lever. That would then allow you to implement multiple-choice questions too. You can use very simple mechanical components that are not very likely to jam or that can be reset quickly if they do. (This kind of reliability has a long history, such as the extreme reliability under even horrible conditions of the AK-47 has made Dr. Kalishnikov's rifle the weapon of choice and virtually standard equipment of all militias and military organizations that don't use American weapons, and some that do. Building a cheap mechanical device (like the AK-47) that can provide reliability under extreme field conditions was possible even for the Soviet Union, it can't be that hard to do here for elections which usually are held in climate-controlled buildings.)
Why nobody has gone to something as simple as this, I don't know. Maybe I'm the first to think of it. In which case, I call it "The Robinson Method" of counting votes. (Hey, I thought of it, I get to name it!)
Maybe there's something wrong with this scheme I don't see. It's
- allows for secret ballot;
- the election is auditable at any time because you still have the tokens in the boxes;
- scales nicely for any size election, large or small;
- highly resilient to voter fraud (at least as much as one with paper ballots, and perhaps more since you're using tokens);
- allows for fast counts, as fast as it takes, say to move the box to the scale, say 30 seconds after the end of the election;
- you can count the result by hand;
- you can count the result mechanically;
- works where there's no electricity;
- can be done with simple, hand held equipment and requires no machines;
- voters don't have to be able to read or figure out how to use a voting machine to vote;
- you can determine the winner of an election even if you can't count;
- you can count the result without even having to break the seals on the ballot box;
- requires no expensive, fragile and vulnerable-to-fraud electronic equipment;
- requires no new equipment at each election;
- reliable, can be operated such that it uses nothing that has moving parts (except the scales) so there's nothing that can break down;
- easy verifiability of the voter's intent; and
- it's cheap to do.
In fact, because you can weigh the boxes without even having to open them, I think it might actually be better than paper ballots. It seems like The Robinson Method has a clear advantage over any other I've ever heard of. It has a big upside to it and either no or few downsides.
As the original 'ballot' from over 100 years ago used balls, it is conceivable to return to using balls again, and a ballot goes back to its original meaning!
So if there's nothing wrong with this idea and a lot of things highly right with it, maybe someone can tell me why a method like this isn't being used?
Again, because of chronic spamming, I had to disable comments. If you have one, e-mail me at firstname.lastname@example.org and if it is even vaguely relevant I'll post it here even if I disagree with it.