Compromise voting systems


Three systems that compromise between certain other voting systems.

  1. Choose-range-or-IRV. You can either cast a Score Voting ballot or an Instant Runoff Voting ballot. Then the Score Voting ballots are fixed while the IRV ballots are processed normally until one winner remains.
  2. (STAR+Score)/2. Like STAR but here the Runoff ballots are appended to the regular ballots. So in the Runoff step among A and B, your A=3 B=2 ballot is still there, but there is also an A=5 B=0 ballot derived from your vote.
  3. Approval IRV. Like IRV but you can give candidates tied rankings like A=B=C>D>E=F. The vote weight of each layer (in this case ABC is layer one, D is layer two, and EF is layer three) is equal to the fraction of candidates in the layer above it that are eliminated, times the weight of that higher layer, so if B was eliminated then D would have 1/3 weight while A and C would still have 1 weight. Oh, and candidates ranked in last place always have a vote weight of zero because you hate those people.

Let me know what you think. I mostly invented these just to invent them. Also note that 2 can be fine tuned to create an uncountable infinity of systems. (The score ballot has weight x, and the Runoff ballot has weight 1-x. Heck, x=-1 or -0.5 looks like an interesting system too…)


Off course for a serious proposal I would value simplicity and “easily explainable” but I have often thought about combining different system to get some if the goods, less of the bads and maybe more honest voters and more fair outcomes for multiwinner systems. So just for fun:
Let voters assign 321 for 3 out of 10 of these options:

  • score nationwide partys
  • approval regional 3seat-districts
  • asset vote open candidate-list (nationwide)
  • IRV
  • make parliament gender-proportional
  • make parliament age-group proportional
  • make parliament 33% new, 33% some experience and 33% very experienced
  • make parliament educiational proportional
  • make parliament religious proportional


I think any system used for a single winner should meet Frohnmayer balance. That is, for every possible vote in the system, there must be another possible vote that would cancel the first vote. I consider that when a system meets this constraint, that constitutes evidence that it provides the same power to each voter as to each other voter. Favorite betrayal seems to happen because voters are seeking power instead of merit. If every vote has the same power, there will no incentive to seek power.


Since you are speaking of systems that would permit the voter to choose from among more than one ballot type, here’s my entry.

The voter can choose an Approval ballot, an IRV ballot, an anti-IRV ballot, or a comprehensive ballot. Ballots of the first three types would be converted to the latter before the tally.

A comprehensive ballot consists of so many ranks as the voter chooses (unless it is the result of a conversion, in which case the conversion algorithm would make the choice). The order of the ranks matters.

Each rank starts with a verb support or oppose, followed by a list of candidates.

The tally proceeds in rounds. Each round seeks to eliminate the worst candidate. The last candidate standing wins.

I do not propose this for multiwinner elections. I have not thought about what reweighting would have to be done to turn it into a PR method.

Each round of the tally begins with counters initialized to zero, one for each candidate who is still in the running.

The round then examines the ballots. With respect to each ballot, it looks through the ranks in their order, stopping when finding a rank of which any of the candidates it names is still in the running. If no such rank is found, the ballot is exhausted, unfortunately. Otherwise, if the settled-on rank signifies support, the counters of the named candidates are incremented. If opposition, they are decremented.

Once the round has so treated all the ballots, it eliminates the candidate with the lowest sum on his or her counter.