JAN KOK's BRILLIANTLY SIMPLE IDEA SHOWING HOW TO RUN SINGLE DIGIT RANGE-VOTING ELECTIONS ON TODAY'S PLURALITY VOTING MACHINES ----Warren D. Smith July 2005----------------------------------- SINGLE DIGIT RANGE VOTING (DEFINITION): I. You provide as your vote in an N-candidate election, an N-tuple of numbers, each entry consisting of a single digit {0,1,2,3,4,5,6,7,8,9} or an INTENTIONAL-BLANK. II. The candidate with the highest average score is elected (where blanks are not incorporated into the averaging). Example: 3 voters and 3 candidates: voter#1: (9, 9, 0) voter#2: (5, *, 9) where *=BLANK voter#3: (2, 2, 9) average scores: (16/3=5.33, 11/2=5.50, 18/3=6.00) and the third candidate wins. JAN KOK's IDEA: Kok pointed out that on any kind of voting machine capable of handling multiple parallel plurality-elections (i.e. all the voting machines in the USA) it also is possible to make them handle single-digit range voting. You transform the election Joe vs Dick vs Jane to three parallel plurality (pseudo)elections JoeElection: Joe0 vs Joe1 vs .. Joe9, DickElection: Dick0 vs Dick1 vs .. Dick9, JaneElection: Jane0 vs Jane1 vs .. Jane9. The results of these three 10-candidate plurality elections -- i.e. the total count for Joe9, the total count for Joe8, etc are easily used to find the result of the single 3-candidate single-digit-range election. With, say, 5000 voters, the total amount of work to do that is tiny compared to the amount of work the voters did to vote. This transformation works fine except for the fact that it multiplies the number of "candidates" by 10, which may strain some kinds of voting machine. The same kind of transformation could also be used to do Approval Voting on plurality machines, namely Joe vs Dick vs Jane (approval election) is transformed to three 2-"candidate" plurality elections Disapprove_Joe vs Approve_Joe, Disapprove_Dick vs Approve_Dick, Disapprove_Jane vs Approve_Jane. Note that these transformations both also permit the improved variant forms of range and approval voting with "you get to leave it blank" votes for candidates you feel ignorant about, and we could have (or not have) an "intentional blank" button. DEFINITION OF "DUMB PLURALITY MACHINE": It computes the total number of votes, and the N totals for each of the N candidates, and it prevents the voter from entering (or at least refuses to count) an illegal "overvote." ERROR DETECTION: Plurality machines that detect overvotes would detect illegal range votes. Plurality machines that detect undervotes would detect "unintentional blank" votes. THEOREM: Any dumb plurality machine can be used to perform a single-digit range voting election (with "X" no-opinion scores allowed, or forbidden, either is ok) if it is capable of handling multiple plurality races. EVALUATION: How well would all this work in practice? Well frankly I would prefer to have voting machines purposely designed for range voting, but it is nice to know that present voting machines can handle the job. But how well depends on the machine: LEVERS in New York State I use lever machines with about 300 binary levers... there is a map so the voter can figure out which levers to depress... Kok's scheme would work fine with them with an appropriate new "map" - and I think it would be just as easy and clear for the voters as it currently is - BUT considerably more levers would be consumed per contest, which might force typical elections to use more than 1 voting machine, while they presently use one. For that reason approval would be better than range voting in practice on such machines. OPTICAL SCAN I think Kok's plan would work quite excellently with many kinds of optical scan ballots (and many kinds of punch card ballots too). In fact there would seem to be little or no advantage for approval versus range on such machines. A range election would be just as easy for a voter to understand and do as a plurality election. BUTTERFLY BALLOT PUNCH CARD (ALA PALM BEACH FLORIDA 2000) I think Kok's scheme would be in big trouble on this sad kind of voting machine. A 10-candidate contest would become a 100 "candidate" 10-contest election which would require a large number, not just one, of those puke-inducing butterfly cards, really inconveniencing voters although in principle it would work. "D.R.E." MACHINES Which contain computers, touch screens, etc, obviously can be programmed to handle practically any kind of voting... as well as practically any kind of election fraud... PEN-AND-PAPER HUMAN-COUNTED BALLOTS LIKE IN IOWA Both AV and single-digit RV are easy to handle, no Kok-transformation required.