Which voting machines can handle which election systems?

(Skip to bottom line)   (Executive summary)

Both single-digit Range (permitted scores 0,1,2,3,4,5,6,7,8,9, and X) and Approval elections can be handled by any kind of voting machine capable of handling multiple plurality elections, i.e. every voting machine in the USA, without modification. The underlying reason for this is Jan Kok's mathematical transformation. An extension of this method can also be used (as is explained in more generality here) to handle two-digit range voting elections (i.e. range 0-99 rather than 0-9) on plurality machines, but only if the voter's "X (intentional blank)" option is sacrificed. Here are examples of how sample ballots would look on different voting machine types, and here is a demo range election you can participate in yourself (try it!) a quicker demo and a version of the quicker demo that demonstrates what range voting would be like on a plurality-style voting machine.

But wait...

Q. But, I asked some voting machine manufacturers "Do your machines support range voting?" and they said "No."

A. Try asking them this way instead:

  1. "Do your machines support ordinary plurality voting?" (Answer: "Yes.")
  2. "There is a theorem that, given a black box that does plurality voting, we can use it to do range voting. Do you dispute this theorem?"

Is this inconvenient?

Depends on the voting machine type and the election. With optical scan machines, the best and most common type in the 2004 USA, range voting is maximally easy and convenient. (Check this map to see how many Americans vote on what kinds of machines and where.)

With range ballots run on ordinary plurality voting machines, slots (e.g. "levers" on New York style machines) get "chewed up" 10 times faster than with plain plurality voting. (Assuming 10 allowed scores. With 100 allowed scores, 20 times faster. With L allowed socres for 1<L<10, L times faster.) Consequently if it is a large enough election or if there are too many elections, you'd run out of levers and more machines would be needed (or you'd have to use range voting with fewer levels). Which would not be so fun. On the other hand with say, optical scan machines, you'd instead "chew up those little pen-fill blobs on the piece of paper" which seems far less of a limitation. (Indeed, range voting on such machines actually strikes us as more convenient and clear than plurality voting.) Ditto with many kinds of punch card machines. So we are not saying range voting is wonderful to run on all plurality machines. We are saying, it can be done on every plurality machine in the USA, with varying levels of convenience or inconvenience depending on the machine type and the election. (The most inconvenience occurs on certain kinds of punch card machines, but fortunately these are rare and their use-fraction is declining.) We do not especially recommend running range elections in this style. We would much prefer to have voting machines specifically designed for range voting. However, because range voting can be done on plurality machines as a stopgap measure, that makes it a lot more smoothly adoptable than many other forms of voting, for example IRV and Condorcet, which simply cannot be done on many kinds of plurality machines.

What we recommend

  1. For use with dumb plurality machines, and in Iowa'08 (pen & paper ballots with manual totalling), we advocate 0-9 and X single-digit range voting.
  2. For rare especially horrible types of plurality machines (perhaps levers or, worst of all, all-punches-in-a-row "butterfly ballot" punch card types) on which 0-9 is too nasty, we advocate 0-1 and X, e.g. "trinary approval." But there should be a law demanding these machines all be phased out by some deadline. This will not be a serious issue because most ballot-initiative states (i.e. the states in which we plan to make a play to get Range Voting after Iowa '08 hopefully adopts range voting) do not use these worst kinds of plurality machines, and by that time (2010) even fewer states and counties will. You can confirm this by checking this 2004 machine-types map of the USA and the list of ballot-initiative states.
  3. For use with future machines specifically designed for range voting, we advocate 0-99 and X (or perhaps even 0-99999 and X), with multiplication of the 0-9 results by 11 to get compatibility with coexisting 0-99 machines.
  4. This all will have the advantage that it permits easy migration to the range system without needing to buy all new machines, but at the same time provides continual incentive to upgrade to 0-99 machines actually designed for range voting, since some voters will want the extra fineness (about 5% in our polls used votes like 98 and 73, and about 30% employed numbers like 75 and 35, both with an 0-100 permitted score range).
  5. The smaller ranges like 0-9 and 0-1 will be mandated to be phased out by a certain date, resulting in uniformity with all machines at the maximum range.

What kinds of voting machines are the best?

Since you asked, there is a general "consensus" (The word "consensus" is from Andrew Gumbel's book.) being reached that precinct-based "optical scan" voting machines are (as of 2005) the best kind available measured by a combination of the following:

  1. Value for money (i.e. they are cheap, as opposed to "DRE" machines which are by contrast extremely expensive)
  2. Simplicity, security, reliability, and durability (as opposed to DRE machines which are by contrast very complex and unreliable and fragile and subject to hacking – and whatever you do, do not be a total idiot by getting one with wireless communication capability)
  3. Small size, weight, and storage space requirements (as opposed to DRE machines which are by contrast very large; mechanical lever machines are very heavy)
  4. Recountable/auditable "paper trail" produced as inherent side-effect of use (as opposed to many DRE machines, and to lever machines; punch card machines are innaccurate and have problems with "hanging chads" and easy overvote-fraud; even those DRE machines which do produce paper trails do it with a printer which means big unreliability, complexity, and expense compared to voters filling in the paper themselves as in optical scan)
  5. Work well for use with range voting – not just plurality voting.
  6. More accurate (lower lost-vote rates) than punch card and full face DRE systems.
  7. Opscan machines process far more voters (so you need fewer of them). A 2006 study by NY Verified Voting estimated that a DRE machine can handle only 200 registered voters, while a single precinct based optical ballot scanner, with sufficient numbers of inexpensive marking booths, can serve up to 4000 registered voters. The point is that slow part of opscan voting – filling out the ballot – is done away from the machine. On the DRE the slow part is done at the machine.
Optical scan machines are currently the most popular type in the USA probably for exactly these reasons – but DRE machines are the most popular type among people trying to sell us these machines so they can make lots of money, also for exactly these reasons. (Read story of Florida's Miami-Dade County, which got suckered into buying DRE machines from ES&S and now wants to discard them all and switch to optical scan as the cheapest way out of the mess they got themselves into. Later note: It now appears as though Florida will be switching over to optical-scan statewide and abandoning all its DRE machines.) However, the USA is unfortunately an outrageous totally nonuniform crazy quilt of different machine types and standards.

What about Condorcet and IRV?

Sorry, considerably fewer kinds of present-day voting machines can handle them.

You might naively think something like "optical scan machines are just an optical input device connected to a computer. Therefore, we can program it to perform any kind of election. Unlike, say, mechanical-counter based lever machines, which can only be used for additive kinds of elections, such as range, approval, and plurality."

However, the truth is messier. See, the great advantage of mechanical lever machines is precisely that they do not contain computers and therefore tampering is fairly hard to do and easy to detect. (Low-tech is better, many voting advocates say, and we think with some justification...)

Now for this reason, we think many punch card machines and optical scan machines are intentionally designed not to have computers, i.e. to have counters, mechanical or electromechanical or hardwired electrical. Or if they do have computers, they are designed to be computers ultrahard to reprogram, e.g. soldered-in single chip computers with program in built-in ROM. (Perhaps somebody knows more than I do re that? Help appreciated.)

If you want to know the answers to these questions for any particular machine type, good luck. Basically, the voting machine manufacturers lie constantly and make up different stories depending who they think is listening. (Check this about how their creation of the myth of "independent testing authorities" for voting machines, and this about a lawsuit against Diebold Inc. for telling lies.) If they think the listener wants security and incredible tamperproofness, meet story number 1... So you generally cannot trust what they say. There are numerous examples of that.

For IRV, the situation is even worse still. Even if your machines have computers, and even if those computers are totally programmable, then those machines still cannot handle counting IRV votes, unless they are all connected together in a giant network, or unless just one machine has all the votes, i.e. all the ballots in the whole state are first shipped to a single central counting agency. (That is because, in IRV, one machine needs to know the totals from all the others just in order to count its own personal stash of votes. In contrast, in Range voting, a machine can total its own personal stash without knowing anything about the rest of the world.) Either way, this is asking for trouble! Think about this in an adversarial environment with people demanding recounts and tracking all votes back to their original precincts, and other bad people trying to screw up the network. IRV is an election administrator's nightmare waiting to happen.


The bottom line: We know range and approval can be handled even by machine types designed only for plurality voting, that cannot be reprogrammed, that have no computers, that are not all connected in a network, and that are located in individual precincts. Some machine types make this pleasant for the voter, other less, but all can do it. But Condorcet, Borda, and IRV cannot be handled unless there is reprogramming and in the IRV case unless there is networking or centralized counting instead of precincts (Deeper look).

That is a significant adoptability edge for range and approval and it alone should be enough to convince all serious voting-reformers to forget about Condorcet, Borda, and IRV. (Not convinced? The cost of buying new voting machines USA-wide would be of order $50 billion. The cost of buying such machines for just one typical county far exceeds the entire nationwide budgets all of the USA's "third parties" combined. Based on this monetary comparison: Who do you think has more power – voting machine stick-in-the-muds, or voting-method reformers?)


Return to main page