Rank choice voting unworkable for presidential elections


president is single seat, so it would be Score Voting. And that would be quite practical, and simpler ballots than IRV. If some states refuse to switch to Score ballots, you just count their ballots as Approval with only one choice filled out. Or maybe every candidate fills out a ballot and publishes it, and voting for a candidate means voting with that candidate’s Score preferences.

An advantage of this is you get rid of the need for a party primary. So e.g. Bernie supporters would have been able to vote for him and he might have beat both Hillary and Trump.

Allocation voting supposedly has the same benefit, but the ballot is more complicated.

Overall I don’t see any legitimate argument you can make that IRV is any good at all when it comes to practical matters. The only real points it can make are that it is more expressive than FPTP.


okay your first response, you haven’t made any argument at all that applies to only ranked or score and not neither or both, so I’m going to ignore that entire comment.


for your second point the only exceptions you made to the problem with your first (that your argument applies to both or neither), were flatly wrong.

a more expressive ballot is necessarily more complex, and vice versa. expressivity is a measure of how many different ways you can fill out a ballot. ballot complexity is a measure of what I just said in the previous sentence.

that is the ONLY argument that is actually different between the two you mentioned and you got it WRONG. score and allocation voting are more expressive and have more complex ballots than ranked choice.

now I have responded to all of your arguments.


again, the only substantive differences between score/allocation voting and rank voting, at this level, is ballot complexity/expressivity and tallying algorithm.

I guess I should have explained that ballot complexity and expressivity increase or decrease together, and that score/allocation is higher on both than ranked choice.

now if anyone wants to discuss those topics we can talk about the proper balance between complexity and expressivity, and that is a big topic area on which discussion is welcome. but the mathematics of greater or less than here is not in dispute among any competent mathematician.

the tallying algorithm and resultant fairness is also a big open area of discussion.

edit: I should have said less than or equal to and greater than or equal to. also I should clarify that a lot of apparent ballot complexity really must be shifted to tallying algorithm. for instance if a ranked choice ballot gets two people scored as second place, the tallying algorithm should be able to handle that without invalidating their ballot. and really there aren’t many sensible options for handling such a ballot and I believe in fact there is only one: transfer the vote to the candidate who otherwise has the most votes


Uh that first post is completely utterly wrong. Tell me how you expect to count a 15-candidate nationwide election for the most important office in the country under IRV.

  1. Count all the first place votes, then send those sub totals upward from precinct to state to the Grand Central Election Tallying Station (GCETS). Then GCETS figures out which candidate to eliminate, and does so. Then the precincts retally everything, and the process repeats.
    Problem: Too much risk of near tie nightmares causing massive chaos. Bush v. Gore recounts cause elections to take months. Remember that we have to do 14 rounds.
  2. Each precinct sends every physical ballot to the GCETS. But what if the trucks fail somehow? How do you know you reached all the ballots? (“Well, each precinct says how many votes they submitted.” But what if someone miscounted?)
  3. Virtually send the votes. But this poses a security risk. Not to mention how this uses a LOT of data. Remember, this is a nationwide scale of over 100 million voters. There is no way to do this without leading to either black box worries of bribery concerns.

How this is different with Score Voting
Score and Approval are additive methods. What we can do is have each precinct publish their sub totals, and those numbers simply need to be added and there is only one round, not 14. If there is a near tie situation, it only needs to be resolved once.

STAR voting
is similar to Score except precincts also need to publish a 15x15 table of the runoff margins. Like “A>B by 458 votes, B>C by 266 votes, and A>C by 1203 votes.” So this is a little more complicated but still doable. And we do not need to send each individual votes, only the total scores and the pairwise table for each precinct. And the end result can be computed and verified with simple addition.


Wow all of those 3 options are really horrible. It’s like you deliberately tried to think of the absolute worst way to tally votes, and then deliberately applied that to IRV and IRV alone.

Tallying ranked choice is a solved problem. There are municipalities and entire states that have done so numerous times. It’s very simple: you total up by unique ordering. Those are your totals. Not per candidate, per ordering. This is utterly trivial to do, even with millions of votes, on this new-fangled thing we call a “computer”. It’s VERY good at counting. Indeed, you could just annonymize the votes (read: remove the names), and send really all the ballots, and my frickin’ laptop could tally that in a few seconds.

" 1. Virtually send the votes. But this poses a security risk. Not to mention how this uses a LOT of data. Remember, this is a nationwide scale of over 100 million voters. There is no way to do this without leading to either black box worries of bribery concerns."

you mean like send it from the voting machines to the tallying centers? I’m pretty sure they already do this. Like, that’s how they count votes NOW. (by the way, they should use cryptographic signing to protect against data tampering. but that’s another topic.)


the population of the u.s. is 326,766,748. If every single person voted, and ranked all 15 candidates, the total amount of data the raw votes would be - untallied - would be a little over 4 gigabytes. a city government like milwaukee has fiberoptic backbone, that then feeds into a local gigabit network. at 1 gigabit/second, that would take 4*8 = 32 seconds to transfer.

but of course you wouldn’t do that. you’d transfer the totals for each ordering. so the amount of data would be way way less.



There is no Federal body for counting votes. We have 50 state bodies that count votes. If we created a federal body that counted all the votes it would undermine the credibility of elections. Imagine a contested election where an unpopular president wins. And he oversees the branch that just happened to say he narrowly won.

Prencinct summability in a federation of 50 States is a must.


“Prencinct summability in a federation of 50 States is a must.”

Of course, but why mention it? Nobody’s disputing that, and nobody’s discussing any methods that aren’t precinct summable.


Rank choice voting is not precinct summable


maybe we’re talking about two different things.

by precinct summable i mean that you can total up how many votes each ranking order gets, and display that online so people can double check.

and then the higher level tabulations can just add those numbers together with all the other precincts. and it’s addition so its associative and cumulative. – the order of addition doesn’t matter, nor does the number of levels of heirarchical addition (parenthesis, so to speak)

i thing that’s crucial.

i don’t think the ability to give a single number for each candidate, and that’s all the data you need, is at crucial. it’s rather an arbitrary thing - why not 2 numbers, or 3? what’s can you do with 1 that you can’t do with 2?




A 8 candidate race, excluding write ins. Has 40,000 permutations you’re going to list how many ballots are in reach if those 40,000 possible orders?

Then you consider absentee ballots have to be counted in IRV. In approval voting a candidate can win by such a large % to make absentee ballots moot but in IRV it’s impossible to discount a vote.

Irv would be a train wreck for a presidential election


More FUD.

Firstly 40,000 is nothing.

Secondly if the permutations outnumber the ballots, then clearly they are not all in use. Worst case, you have the number of ballots, and this is very very unlikely. Far more likely it’s a small fraction of that. And 40 candidates? Really? Maybe try something a bit more realistic.

You clearly have an axe to grind that’s biasing your judgement and causing you to make bad arguments.


But if there is any way for the fact that someone voted E > H > I > R > V > B > A > D > U > S > X > Y > Z to get exposed to the public, then we could have people bribe or threaten that you must vote exactly that way or else you and your family will be hunted down and destroyed.
You say anonymize the ballots. But then you figure out that H won, but you have to leak the decryption key to the public or else people may fear that someone in the ballot counting process artificially made “H” actually equal Dr. Fascist instead of Ima Goodperson, the person who was supposed to be labeled H.

My point is, you cannot have a transparent, secure, and bribe-free election under IRV although you can with Score Voting.


Currently in the US we only seem to use it (IRV) for elections covering a smaller area. Scaling this up to having to count votes from all over the country is a nightmare. Bringing in physical ballots from every precinct, including Alaska and Hawaii, to the GCETS is a nightmare. Doing it digitally runs into concerns with Russian hacking. Attempting to count via the fourth estate leads to bribery issues.


firstly what you saidwould apply to ALL tallying methods.

Secondly you don’t understand how asymmetric encryption work. it’s asymmetric. the public key and private key are different, and you can give away the public key without giving away the private key. and whay you decrypt with one you need the other to decrypt. this is used for making secure connections over the internet, and for authenticating hosts. for instance you see that little “https” up in the url bar? That tell’s you that your computer has the public key for this site, and the host has the private key. your browser knows that the host is really forum.electionscience.org and not some imposter site, because forum.electionscience.org is the only person that can prove it has the private key, by encrypting an arbitrary string with it that can then be decrypted by the public key and be back to the original string. this technology is so well established that we use it for credit card transactions over the web all the time.


I love how wonderfully vague your posts are. What specifically did I say that applies to all methods? How specifically does asymmetric encryption apply here?