Ranked Choice Voting

[Hohn]

You've probably heard of the idea. Instead of voting for a candidate, you rank all of them in order. This capture information about how you feel about ALL of the candidates, and allows for instant runoffs.

This advocacy group is pushing it: FairVote

Now, the idea seems to have merit and I think it would go a LONG way to improving things-- primary elections in particular. But is there a downside that the Fairvote group is overlooking?

I'm pondering that question, and I put it to you for analysis as well.

 

[Bill of Rights]

To paraphrase a quote attributed to Josef Stalin, it matters not who counts the votes, nor how they are cast. What matters is who counts the votes.

Until the voting machines give a paper trail that allows for accounting of who received votes, rather than just the reporting of the aggregate totals, we will never have verifiably fair elections. Of note: I do not mean that anyone can find out how *I* voted or how *you* voted without the voter telling them, what I mean is that if one precinct reports that candidate A got only 100 votes in that precinct, and 101 or more people can be found to show their voting receipt that they voted for "A", election fraud is proven.

Blessings,
Bill

You've probably heard of the idea. Instead of voting for a candidate, you rank all of them in order. This capture information about how you feel about ALL of the candidates, and allows for instant runoffs.

This advocacy group is pushing it: FairVote

Now, the idea seems to have merit and I think it would go a LONG way to improving things-- primary elections in particular. But is there a downside that the Fairvote group is overlooking?

I'm pondering that question, and I put it to you for analysis as well.

 

[SSGSAD]

To paraphrase a quote attributed to Josef Stalin, it matters not who counts the votes, nor how they are cast. What matters is who counts the votes.

Until the voting machines give a paper trail that allows for accounting of who received votes, rather than just the reporting of the aggregate totals, we will never have verifiably fair elections. Of note: I do not mean that anyone can find out how *I* voted or how *you* voted without the voter telling them, what I mean is that if one precinct reports that candidate A got only 100 votes in that precinct, and 101 or more people can be found to show their voting receipt that they voted for "A", election fraud is proven.

Blessings,
Bill

Then this will NEVER happen, in our lifetime, or as long as "present day" "l", have anything to do with elections .....

ie, VOTER ID ..... JMHO ..... YMMV .....

 

[Pyroponce]

Instant runoff voting is a good idea, which means it will never happen. I would say that IRV is a necessary but insufficient condition of transparent elections. It would need to be combined with things like paper trails, voter ID, closed primaries, more congresspersons, etc. The establishment will never let it come to pass obviously because they will never give up anything that could threaten their power.

 

[T.Lex]

You've probably heard of the idea. Instead of voting for a candidate, you rank all of them in order.

Well now. I think we know who has multiple accounts at INGO now. *coff* jamil *coff* A little astroturfing never hurt anyone.



That may be too much of an inside joke.

 

[mikefraz]

The idea is just crazy enough to work. Which means it will
never happen.

 

[Hohn]

The main appeal to me of RCV/IRV is that it reflects reality of how all the rest of our decision making in life generally goes.

If I go to the store and they don't have my favorite variety of coffee, I can't buy it, so I go to my next strongest preference.

Our current voting system is like saying that if the store doesn't have the milk you prefer, you can't buy any food at all. That's nonsense. Nobody lives their life that way. In almost everything, we have more choices that a simple binary vote. It is not a choice of A or B, but rather a choice of A, then B, then C.


My job involves a good deal of stats work. That means I have to deal with variables of different kinds. And depending on the kinds of variables, you can get either a lot more or a less less information.

Worst of all is a "nominal" variable. The only value is the name, it tells me nothing about kind, quantity, or relative importance. It's not really even a variable, it's more of a label.

Next most useful is the categorical variable. Instead of a big blom of random names (like a nominal variable), I can at least associate things with each other by kind. This is useful for things like Customer A vs Customer B, or comparing one region to another.

Then we come to Ordinal variables. This allows me to not only group things together, but to rank them. This is yet another degree of information not present in the other kinds of variables.

Finally we get to the BEST OF ALL-- continuous variables. I can not only associate things together, or rank them, but I can *quantify* the relative differences beyond mere ranking.



Consider how this plays out in some cases. An hypothetical employer grades each employee on the 3-point scale based on whether they are "average/acceptable" range (2), HiPot/above average (1) or below average/remedial (3). Let's say there are quotas and that only the top 20% can be "1", the lowest 20% are "3", and everyone in the middle are "2" (60%).

This is an ordinal (rank) system. But what is the difference between the lowest "1" and the highest "2"? It's exactly one person (in the whole stack). But it is also an entire category.

If this company has 100 employees, then the difference between the #20 and #21 employees (the lowest #1 and the highest #2) is exactly the same between #1 and #79 (the highest #1 and the lowest #2). The system is so crude that the difference between 20 people in one category and 60 in another cannot be measured-- it's just one category down.



What's this to do with voting?

Our vote system now is categorical. For partisans, there are basically two categories: D and R. For independents, there are still practically just two categories. Because there are only two categories, the huge range of voter preference within category is never captured. The system is "blind" to that internal variation, just like the employer's ranking system can't tell the difference between employee ranked #21 and one ranked #79, because they are both "Twos."

Primary elections are categorical also. For a closed primary, the category is not "party" (there is no categorical distinction when everything's in the same category). The two categories rather are: 1) the one candidate I most prefer and 2) everyone else.



Going to ranked choice voting moves from categorical variable to Ordinal variable, and this captures a TON more information about whom the voters actually want and what they care about.


But-- IS THAT A GOOD THING? Is it a good thing to move us closer to a direct democracy and further diminish the role of parties?

 

[Twangbanger]

Didn't we do a little mini-trial of this here on INGO a while back, and the results weren't that different from what the other system produced? In other words, the experiment returned a Republican candidate who (by popular approbation) had no chance in the general election? (Can someone *cough* dig up the threads and see who won?)

 

[jamil]

I think I wrote this same post a few months ago. I am actually a little freaked right now. Well. Not really. Anyway. Great post.

The main appeal to me of RCV/IRV is that it reflects reality of how all the rest of our decision making in life generally goes.

If I go to the store and they don't have my favorite variety of coffee, I can't buy it, so I go to my next strongest preference.

Our current voting system is like saying that if the store doesn't have the milk you prefer, you can't buy any food at all. That's nonsense. Nobody lives their life that way. In almost everything, we have more choices that a simple binary vote. It is not a choice of A or B, but rather a choice of A, then B, then C.


My job involves a good deal of stats work. That means I have to deal with variables of different kinds. And depending on the kinds of variables, you can get either a lot more or a less less information.

Worst of all is a "nominal" variable. The only value is the name, it tells me nothing about kind, quantity, or relative importance. It's not really even a variable, it's more of a label.

Next most useful is the categorical variable. Instead of a big blom of random names (like a nominal variable), I can at least associate things with each other by kind. This is useful for things like Customer A vs Customer B, or comparing one region to another.

Then we come to Ordinal variables. This allows me to not only group things together, but to rank them. This is yet another degree of information not present in the other kinds of variables.

Finally we get to the BEST OF ALL-- continuous variables. I can not only associate things together, or rank them, but I can *quantify* the relative differences beyond mere ranking.



Consider how this plays out in some cases. An hypothetical employer grades each employee on the 3-point scale based on whether they are "average/acceptable" range (2), HiPot/above average (1) or below average/remedial (3). Let's say there are quotas and that only the top 20% can be "1", the lowest 20% are "3", and everyone in the middle are "2" (60%).

This is an ordinal (rank) system. But what is the difference between the lowest "1" and the highest "2"? It's exactly one person (in the whole stack). But it is also an entire category.

If this company has 100 employees, then the difference between the #20 and #21 employees (the lowest #1 and the highest #2) is exactly the same between #1 and #79 (the highest #1 and the lowest #2). The system is so crude that the difference between 20 people in one category and 60 in another cannot be measured-- it's just one category down.



What's this to do with voting?

Our vote system now is categorical. For partisans, there are basically two categories: D and R. For independents, there are still practically just two categories. Because there are only two categories, the huge range of voter preference within category is never captured. The system is "blind" to that internal variation, just like the employer's ranking system can't tell the difference between employee ranked #21 and one ranked #79, because they are both "Twos."

Primary elections are categorical also. For a closed primary, the category is not "party" (there is no categorical distinction when everything's in the same category). The two categories rather are: 1) the one candidate I most prefer and 2) everyone else.



Going to ranked choice voting moves from categorical variable to Ordinal variable, and this captures a TON more information about whom the voters actually want and what they care about.


But-- IS THAT A GOOD THING? Is it a good thing to move us closer to a direct democracy and further diminish the role of parties?

Didn't we do a little mini-trial of this here on INGO a while back, and the results weren't that different from what the other system produced? In other words, the experiment returned a Republican candidate who (by popular approbation) had no chance in the general election? (Can someone *cough* dig up the threads and see who won?)

I conducted it. I think I still have the spreadsheet I used to fally it. Going from memory, Cruz won. Rand Paul came in second. Rubio third. Ben Carson 4th. Trump 5th.


how the hell did I miss this thread?

 

[david890]

But is there a downside that the Fairvote group is overlooking?

The Constitution, for one. We'd need to amend it from our current system, and I doubt ANY politician would do that.

Congress has a 10% favorability rating, yet 80% of those up for re-election win. They won't change a system that works for themselves, and few voters seem willing to oust an incumbent.

Regarding the FairVote plan, how does one become a candidate? If there are no restrictions, I can see 500 candidates for mayor of Indy. The one with the most money runs the most ads and gets the most name recognition. We pretty much have that right now (ask the average person who the two 3rd-party candidates are; I doubt 20% can name them).

"Instant runoffs" couldn't happen until ALL the votes are counted. So, there would have to be a 2nd day of voting. Not seeing a lot of people going for that, given the number who can't take off work to vote in the first place. Now, make election day a national holiday (or Sat & Sun) and we might see a slight increase in voting.

IMHO, I think the average voter would be as suspicious of any new scheme as they are with the current one. I'd think we'd need to run several test votes of a new system (not legally binding, of course) concurrently with the existing system for any meaningful comparison. What happens if a "test" FairVote shows 80% for a very popular candidate (e.g., James Taylor; everyone likes James), while our current popular vote shows just 60%, and that candidate still manages to lose the Electoral College vote? It would not be pretty.

 

[Hohn]

I don't see how the Constitution would need to be amended. RCV is totally compatible with the electoral college, and the constitution grants States the right to chose how their electors are chosen.

Candidate selection could be the exact same way it is now. You will out the paperwork and comply with state law. For local offices, you just fill out papers. For other offices, a certain minimum number of signatures may be required.

It's important to separate the primary for the general for process. Parties determine their candidates, and should determine that process also. States should determine how general elections are run.

 

[cosermann]

Ranked choice voting - for people who can't make decisions.

Electronic voting - as a former information systems auditor - no way!

 

[kehrhit]

To paraphrase a quote attributed to Josef Stalin, it matters not who counts the votes, nor how they are cast. What matters is who counts the votes.

Until the voting machines give a paper trail that allows for accounting of who received votes, rather than just the reporting of the aggregate totals, we will never have verifiably fair elections. Of note: I do not mean that anyone can find out how *I* voted or how *you* voted without the voter telling them, what I mean is that if one precinct reports that candidate A got only 100 votes in that precinct, and 101 or more people can be found to show their voting receipt that they voted for "A", election fraud is proven.

Blessings,
Bill

Exactly. The voting machine should spit out a receipt that allows you to go back and independently verify how your vote was recorded, and to validate all votes. However I agree this is unlikely to ever be implemented.

 

[jamil]

Ranked choice voting - for people who can't make decisions.

Electronic voting - as a former information systems auditor - no way!

For people who can't make decisions, ranked voting isn't gonna help them. If you can't decide between two candidates, how the hell would you ever decide the ranks for 15? Ranked voting just gives you some say about who wins if your favorite doesn't.

Electronic voting - I'm not real fond of it. Mechanical systems can be taken apart and inspected. Software is a different monster. You can hide all kinds of ways to subvert it. But if we have to have electronic voting, it needs to be open source so that it can be inspected.

Exactly. The voting machine should spit out a receipt that allows you to go back and independently verify how your vote was recorded, and to validate all votes. However I agree this is unlikely to ever be implemented.

The machine I voted on for 2014 was kind of close to that. It was a two step system. You make your selections electronically, but that machine doesn't tally anything. Its purpose is just to print out your choices. Then you take the card with your choices printed over to another machine, which takes the card and reads it. There's probably nothing preventing you from taking a photo of the card when it's printed, so you have a record of how you wanted to vote.

But having a record of your choices does not prove those choices were actually tallied. In this case, the machine that reads the cards could tally something different and you wouldn't know.

And that's the problem I had with that system, there's no way for a voter to know that his or her vote was tallied accurately, other than accepting the certification of the voting system. I'd have felt much better about the process if the code for the "tally" machine were open source.