I have an even better version of this which I was planning to write a post on at some point - just have each district run an only-vote-for-your-first-choice election, but rather than selecting only the plurality winner as if they represented 100% of the constituents, select everyone who got any significant fraction of the vote, and send them to the legislature with voting power proportional to the number of votes they got.
This would mean:
1. Gerrymandering is completely impossible, and the result is guaranteed to be proportional to people's preferences (e.g., whatever percentage of people vote for Party A, that will be the percentage Party A gets in the legislature). It is even better than proportional representation in this respect, since PR is still discontinuous due to each party having a discrete number of representatives. It is also more proportional than your system, since minorities in a district still get representatives, rather than just being able to reduce the majority's voting power.
2. Everyone is represented. You will always have a legislator from your district who represents you, not just one who was voted for by the majority of your district who you may disagree with.
3. Every vote matters equally, regardless of what district you're in, since every votes gives one representative the same amount of extra voting power. Participation is therefore encouraged everywhere, with no need to focus only on competitive districts. Parties are encouraged to maximize their support even in districts where one party is guaranteed to have the majority.
4. Unlike regular PR, local interests can still be taken into account because there are still local districts. Everyone still has a small number of representatives who are specifically meant to be their representatives.
5. People still vote for specific candidates, rather than parties, unlike in regular PR. This gives people the chance to make sure the parties are putting forward the best candidates possible and sometimes break party lines.
Since it's not just one, this means you will have to send many more people (and thus pay more salaries, etc.), which is probably worth it. However, if we allow that, we run into another problem:
The number of people sent creates a kind of threshold for how many votes a candidate must receive. For example, if there are a hundred candidates but only ten people are sent, voters who support unpopular candidates face a dilemma: vote honestly and potentially waste their vote, or vote strategically for a more popular candidate. This can be problematic for new or upcoming parties and politicians, as they must convince voters that enough people will support them to ensure their votes are not wasted.
We can solve this by allowing voters to rank their candidates. First, we count all the votes and eliminate the candidate with the fewest votes, redistributing those votes to their second-choice candidates. We then eliminate the next candidate with the least votes and continue this process until only ten candidates remain. The voting strength of the remaining ten is proportional to the share of votes they received after all these rounds.
I've previously designed a version of this system with two differences:
1) It used a linear voting mechanism that was overly complicated, and I wasn’t happy with it.
2) It allowed for a variable number of seats depending on how much over the threshold a candidate went. I felt that having a presence in the capital held value beyond voting power (e.g., networking, sharing viewpoints, etc.). For instance, in a race where ten candidates each received slightly more than 9% of the vote and one candidate received slightly less than 9%, that candidate would still be sent, resulting in 11 representatives. Conversely, in a race where nine candidates received 11.111% of the vote each and one candidate received only a single vote, that candidate would not be sent, and only nine representatives would be elected instead.
In any case, feel free to still write that post, and feel free to incorporate this revision.
Quick edit: I wondered why this hadn't been done before, but it's probably because the math would be too cumbersome, having to count up the fractions everytime a vote is called in the senate/house, especially since sometimes people don't show up so you have to recalculate what percentage of the vote a bill should get to pass. However, we now live in a time of computers, so if we all give the representatives a little ipad with a voting adjustment program running on it, this shouldn't be a problem anymore.
Yeah, so technically speaking, the stuff that I said about proportionality isn't 100% exact because there has to be some threshold in practice to avoid having an obscene number of representatives. The "significant fraction" I mentioned was meant to be some very small threshold, something like 1%, put in place to prevent the legislature from having a ridiculous number of members. PR countries already introduce a threshold - usually 5% - larger than what you would normally need for 1 seat if you really allocated them proportionally. Personally, I don't like the 5% threshold, and I think it should be lower. With a threshold as small as 1%, there would be almost no distortion from pure proportionality (consider that the threshold would not have a uniform effect as it does in PR - it might hurt one super-minority party in one district but allow them to get through in others, so that overall, it balances out, as compared to PR thresholds which just exclude a small party from the legislature altogether).
The basic argument for why my system is more proportional is that, without a threshold, it's exactly proportional, whereas no other system is, even PR. Once a threshold is introduced to keep the size of the legislature down to a practical amount, it will introduce some distortions, but it would introduce them to any system, not just mine. So the practical version with the threshold is still less distorted than any other system.
And yeah, if you want to reduce the distortions created by the threshold even further, you can introduce ranked choice voting and just say that if your first choice doesn't meet the threshold, then your highest-ranked candidate that got enough first-choice votes to pass the threshold will be counted as representing you in the legislature. Then the only way for your vote not to count is if you don't rank a single candidate who clears the threshold. The only question is whether it's worth it to complicated the system for such a small gain in proportionality.
As for the fixed threshold vs. fixed number of legislators issue, I was originally thinking of it in terms of a fixed threshold, since it seems unnecessary to send some nobody who got 0.01% of the vote to the legislature only to have completely negligible voting power just because there weren't enough serious candidates who ranked higher. There is one issue with this, which is that you could hypothetically get a crazy result, like "101 candidates each get an equal number of votes, so no one passes the 1% threshold." To fix this, we could instead use a negative threshold - instead of fixing the minimum percentage of votes a candidate needs to get into the legislature, fix the maximum proportion of "wasted votes". Basically, instead of having a 1% threshold and only letting candidates who got at least 1% of the vote in, have a 99% threshold that all winners combined have to pass, and keep adding the next-biggest vote-getter until you surpass 99% of the vote.
> PR countries already introduce a threshold - usually 5% - larger than what you would normally need for 1 seat if you really allocated them proportionally
Well, no. A lot of countries don't have threshold over the amount of seats e.g. the Netherlands. In fact the Netherlands performs better than your 1% since they have more than 100 seats (150 seats).
But... that's still a threshold in practice, since you need at least one seat and if you don't get a percentage of the population equal to that seat, you've wasted your vote. It also has the additional problem of creating a ton of parties, right now there are 15 parties in the house (and 15 in the senate, not even exactly the same 15). This makes forming a coalition really slow, and the Netherlands has some of the slowest government formation in the world. If you set a higher threshold but used single transferable vote (the system I discussed, which isn't exactly the same as ranked choice voting), you can have quick coalition formation without having people feel the pressure to vote strategically.
If you're worried it's too complex you can allow people to just cross off one party, and leave the ranking as an extra option to those who want it.
I have an even better version of this which I was planning to write a post on at some point - just have each district run an only-vote-for-your-first-choice election, but rather than selecting only the plurality winner as if they represented 100% of the constituents, select everyone who got any significant fraction of the vote, and send them to the legislature with voting power proportional to the number of votes they got.
This would mean:
1. Gerrymandering is completely impossible, and the result is guaranteed to be proportional to people's preferences (e.g., whatever percentage of people vote for Party A, that will be the percentage Party A gets in the legislature). It is even better than proportional representation in this respect, since PR is still discontinuous due to each party having a discrete number of representatives. It is also more proportional than your system, since minorities in a district still get representatives, rather than just being able to reduce the majority's voting power.
2. Everyone is represented. You will always have a legislator from your district who represents you, not just one who was voted for by the majority of your district who you may disagree with.
3. Every vote matters equally, regardless of what district you're in, since every votes gives one representative the same amount of extra voting power. Participation is therefore encouraged everywhere, with no need to focus only on competitive districts. Parties are encouraged to maximize their support even in districts where one party is guaranteed to have the majority.
4. Unlike regular PR, local interests can still be taken into account because there are still local districts. Everyone still has a small number of representatives who are specifically meant to be their representatives.
5. People still vote for specific candidates, rather than parties, unlike in regular PR. This gives people the chance to make sure the parties are putting forward the best candidates possible and sometimes break party lines.
Since it's not just one, this means you will have to send many more people (and thus pay more salaries, etc.), which is probably worth it. However, if we allow that, we run into another problem:
The number of people sent creates a kind of threshold for how many votes a candidate must receive. For example, if there are a hundred candidates but only ten people are sent, voters who support unpopular candidates face a dilemma: vote honestly and potentially waste their vote, or vote strategically for a more popular candidate. This can be problematic for new or upcoming parties and politicians, as they must convince voters that enough people will support them to ensure their votes are not wasted.
We can solve this by allowing voters to rank their candidates. First, we count all the votes and eliminate the candidate with the fewest votes, redistributing those votes to their second-choice candidates. We then eliminate the next candidate with the least votes and continue this process until only ten candidates remain. The voting strength of the remaining ten is proportional to the share of votes they received after all these rounds.
I've previously designed a version of this system with two differences:
1) It used a linear voting mechanism that was overly complicated, and I wasn’t happy with it.
2) It allowed for a variable number of seats depending on how much over the threshold a candidate went. I felt that having a presence in the capital held value beyond voting power (e.g., networking, sharing viewpoints, etc.). For instance, in a race where ten candidates each received slightly more than 9% of the vote and one candidate received slightly less than 9%, that candidate would still be sent, resulting in 11 representatives. Conversely, in a race where nine candidates received 11.111% of the vote each and one candidate received only a single vote, that candidate would not be sent, and only nine representatives would be elected instead.
In any case, feel free to still write that post, and feel free to incorporate this revision.
Quick edit: I wondered why this hadn't been done before, but it's probably because the math would be too cumbersome, having to count up the fractions everytime a vote is called in the senate/house, especially since sometimes people don't show up so you have to recalculate what percentage of the vote a bill should get to pass. However, we now live in a time of computers, so if we all give the representatives a little ipad with a voting adjustment program running on it, this shouldn't be a problem anymore.
Yeah, so technically speaking, the stuff that I said about proportionality isn't 100% exact because there has to be some threshold in practice to avoid having an obscene number of representatives. The "significant fraction" I mentioned was meant to be some very small threshold, something like 1%, put in place to prevent the legislature from having a ridiculous number of members. PR countries already introduce a threshold - usually 5% - larger than what you would normally need for 1 seat if you really allocated them proportionally. Personally, I don't like the 5% threshold, and I think it should be lower. With a threshold as small as 1%, there would be almost no distortion from pure proportionality (consider that the threshold would not have a uniform effect as it does in PR - it might hurt one super-minority party in one district but allow them to get through in others, so that overall, it balances out, as compared to PR thresholds which just exclude a small party from the legislature altogether).
The basic argument for why my system is more proportional is that, without a threshold, it's exactly proportional, whereas no other system is, even PR. Once a threshold is introduced to keep the size of the legislature down to a practical amount, it will introduce some distortions, but it would introduce them to any system, not just mine. So the practical version with the threshold is still less distorted than any other system.
And yeah, if you want to reduce the distortions created by the threshold even further, you can introduce ranked choice voting and just say that if your first choice doesn't meet the threshold, then your highest-ranked candidate that got enough first-choice votes to pass the threshold will be counted as representing you in the legislature. Then the only way for your vote not to count is if you don't rank a single candidate who clears the threshold. The only question is whether it's worth it to complicated the system for such a small gain in proportionality.
As for the fixed threshold vs. fixed number of legislators issue, I was originally thinking of it in terms of a fixed threshold, since it seems unnecessary to send some nobody who got 0.01% of the vote to the legislature only to have completely negligible voting power just because there weren't enough serious candidates who ranked higher. There is one issue with this, which is that you could hypothetically get a crazy result, like "101 candidates each get an equal number of votes, so no one passes the 1% threshold." To fix this, we could instead use a negative threshold - instead of fixing the minimum percentage of votes a candidate needs to get into the legislature, fix the maximum proportion of "wasted votes". Basically, instead of having a 1% threshold and only letting candidates who got at least 1% of the vote in, have a 99% threshold that all winners combined have to pass, and keep adding the next-biggest vote-getter until you surpass 99% of the vote.
> PR countries already introduce a threshold - usually 5% - larger than what you would normally need for 1 seat if you really allocated them proportionally
Well, no. A lot of countries don't have threshold over the amount of seats e.g. the Netherlands. In fact the Netherlands performs better than your 1% since they have more than 100 seats (150 seats).
But... that's still a threshold in practice, since you need at least one seat and if you don't get a percentage of the population equal to that seat, you've wasted your vote. It also has the additional problem of creating a ton of parties, right now there are 15 parties in the house (and 15 in the senate, not even exactly the same 15). This makes forming a coalition really slow, and the Netherlands has some of the slowest government formation in the world. If you set a higher threshold but used single transferable vote (the system I discussed, which isn't exactly the same as ranked choice voting), you can have quick coalition formation without having people feel the pressure to vote strategically.
If you're worried it's too complex you can allow people to just cross off one party, and leave the ranking as an extra option to those who want it.