Hacker News Reader

A maximally representative district would be a large multi-representative district that elects 5 representative using ranked choice or approval voting.

Then you get the property of representation you're looking for without having to draw fine-grained arbitrary distinctions on a map.

https://www.fairvote.org/fair_representation#what_is_fair_vo...

I assume you mean that all the candidates run against each other simultaneously, and the top 5 all win a seat? I don't think that has the properties you are claiming. At least not with approval voting.

Let's spherical cow this for a second and say that 51% of the population would like candidates from the Foo party, and 49% would like candidates from the Bar party. What prevents the 51% from electing five Foo candidates? When a "fair" split would be three Foo candidates and two Bar candidates.

In Approval voting, one would expect five Foo candidates with 51% votes, and five Bar candidates with 49% vote.

Ranked choice might fix this, but the actual results would depend entirely on the vote resolution mechanism.

The idea you are looking for is Single Transferable Vote aka STV: https://www.fairvote.org/rcv#how_rcv_works

I'm a fan of STAR (Score Then Automatic Runoff) though I'm not sure the details on how that looks with multiple positions up for grabs. https://www.starvoting.us/

I'm familiar with the concepts.

http://zesty.ca/voting/sim/

Instant runoff is listed as Hare, and it has some very... interesting results. I don't think it fixes any of the issues people have with first-to-the-post voting, and sometimes makes those issues even worse.

Would you please click the link and read about the Fair Voting Act in its entirety? I believe it addresses every issue you have raised so far.

I'm confused. I haven't actually brought up any issues except those raised in the link I provided, where IRV has some serious issues. So allow me to remedy it by providing specifics.

Your link clearly says, in heading: "RCV for Single-Winner Offices (also known as Instant Runoff Voting / IRV)". So unless they are not actually using IRV, then the link I supplied is relevant. It's specifically this part which is broken, quoted from your link:

The candidate with the fewest votes is eliminated, and voters who picked that candidate as ‘number 1’ will have their votes count for their next choice.

So unless a candidate gets a good-enough first-vote showing to not be last, they have zero chance of winning. Even if 100% of the population picks them as their second candidate, and the remaining first-pick votes are spread over 100 other candidates, it doesn't matter they will lose without first-pick votes. And that's what results in graphs like the "square" graph in my link, where a centrist population ends up with a candidate that doesn't make sense.

The heart of the matter is that part of the vote information is ignored until other events happen. But by the time that triggering event happens, the information may no longer be used. In my example, the fact that 100% of people pick a candidate as their second choice is flat-out ignored until candidates start being eliminated. By the time that second choice is looked at, the relevant candidate is no longer in the running.

I have not seen similar multi-candidate simulations, but STV seems to have the same issue regarding ignoring information. Unpopular first vote candidates will be eliminated, even if they're extremely popular second vote candidates.

For single-winner elections, approval voting is both simpler to administer (because there's no ranking) and has better outcomes (because all information is taken into account at the beginning). However, a more complicated method would need to be used for multi-candidate races if "fairness" of representation is a target.

Hi, I'm from Australia, where we actually use IRV and STV. I dislike IRV's lack of monotonicity too, but for STV, which is the top-of-chain proposal, your objection

> Unpopular first vote candidates will be eliminated, even if they're extremely popular second vote candidates.

doesn't seem to be a major problem. Groups do well on second and subsequent choices from a combination of being popular in their own right and more-popular groups saying "put us first and then them second".

Our federal electoral commission publishes every ballot ordering for our Senate elections now. A guy named David Bahry has a "preference explorer" which lets you get a high-level overview for yourself: https://pappubahry.com/pseph/aus_2019/

There is a lot more to the Fair Representation Act than just RCV. Lots and lots of detail on the site, including some mathematics on multi winner elections.

Make it proportional to the vote. If it's 51/49, bar still gets 2. Also makes third parties feasible that way.

Personally, I think fully proportional meaning getting rid of districts is the best idea. The entire state is one district, parties publish a ranked list well before the election, and based on the votes everyone got, the top n people from the party's list get elected.

We still have a problem though: how many seats does each state get? Do we still continue to apportion seats each state gets every ten years on a census?

> We still have a problem though: how many seats does each state get? Do we still continue to apportion seats each state gets every ten years on a census?

This seems to work and nobody has an issue with the high-level apportionment.

The only criticism I've heard of it is the 435 cap, which is kind of arbitrary, and now one rep represents way too many people.

What would the number of representatives look like if it was fluid and we said the unit is the smallest state that only gets one representative is the yard stick?

A quick search shows Wyoming is the smallest state by population with 578,759 so if we decreased the apportionment from seven hundred thousand something to that 578,759, how many more seats would we need?

If a state with a population of 578,759 gets one seat does a state with population of 578,759 * 2 - 1 get two seats or one seat?

round up or down as needed. 330 million / 578000 is 570 representatives, give or take. To put this in perspective, the current French National Assembly is 577 members, the Italian Chamber of Deputies is 630 members, the UK House of Commons is 650 members, and the German Bundestag is 709 members.

I see others have replied to you, correctly; but, to throw in my two cents: multimember districts use some form of ranked choice or quadratic or sortition system. What you’re talking about are called “multimember at large” districts and are both illegal (falling awry of the VRA), in addition to being unconstitutional.

> The representative system works best when it's able to group people with similar interests and values and lets them elect someone who can share those interests.

This sounds plausible, but I've been doubting it the past few years. It seems like a big driver in polarization because the most extreme subset of those "shared values" groups can more easily take the driver seat.

I think there are cases where that happens - but I also think we can't underestimate how much reactionary sentiment has stoked this polarization outcome as well. Politicians have learned that "fear of the other guy" can be just as powerful of a motivator (if not more so) to get turnout. So they know that in some cases running an extreme "anti candidate" improves their chances of winning - if only because it energizes the base that is most afraid, and thus most likely to turn out.

For this issue in particular, I think ranked choice could be a huge help.

This sounds nice, but who gets to choose what attributes of populations are the right ones to group together into districts? You suggest grouping by population density, which certainly makes sense for grouping together certain political interests (like zoning). And yet by doing that, you've essentially maximally fragmented people by gender, which is another attribute that makes sense for grouping together certain political interests.

That's an extreme example of course, and I'm not suggesting to instead create single-gender districts, but I hope it illustrates my point.

I'm not sure I understand what you're saying. What does gender have to do with population density? I chose population density, because it was the only value that I could think of that closely correlated with culture and that was not a protected status (ie, not race, religion, sex, etc). If it's done algorithmically, it doesn't seem like a bad measure.

Anyhow, the "who gets to choose" thing is tough. Right now legislators get to choose, but the judges said that they are failing at that, so IIRC, now an academic from a state university got to make some map adjustment. Ideally, I'd have a committee of academics provide vetted map recommendations that elected officials choose from, and that a judge signs off on.

My interpretation was that you chose to group together areas with similar population density because there are important policy concerns directly related to population density. And indeed that's true: water rights, zoning, building codes, pollution and noise restrictions, and more ought to depend highly on the population density of the area, and it doesn't seem necessary or even moral for rural voters to have much say in how dense cities set those policies.

But my point is that there are plenty of other policies which don't (or shouldn't) vary based on population density, and thus grouping districts based only on population density might give very little influence to interest groups for those other policies. Civil rights are the extremely obvious example, but there are others.

What you’re probably looking for is an increase in the number of Representatives rather than a change in the formula of districting. I am at least.

Look at it this way, if it were 1 Rep per between 30K and 50K instead of 1 Rep per ~700K, there is a decent shot that somewhere in those several thousand Representatives you’ll have a few that are from just about every slice of American life, and elections are still every 2 years if one session of Congress just isn’t good enough for you. Even Wyoming would be sending between 12 and 19 Representatives to Congress rather than the lone Rep they send now, and would have more and various opinions and interests represented.

> but who gets to choose what attributes of populations are the right ones to group together into districts?

Let the voters decide. Allow them to group themselves into self-selecting affinity groups, i.e. have proportional representation.

How would that work with representative democracy? With direct democracy that's pretty straightfoward: one could vote with conservationists on land policies, libertarians on drug policies, progressives on welfare policies, etc. instead of having to vote for a representative who probably disagrees wildly with at least one of those views.

> How would that work with representative democracy?

With PR there would be a larger number of viable parties, for example in the last Dutch election, 13 parties got elected.

> instead of having to vote for a representative who probably disagrees wildly with at least one of those views

With 13 parties you'd probably be able to find one which is reasonably close to your own views.

Having a larger number of viable parties doesn’t seem like it would help that much when your district is huge and gets one representative elected with FPTP.

This wouldn't happen, because districts would be multi-member.

For extra proportionality, have top-up seats (as used in Germany or Scotland).

Competitive districts select representative candidates.

Safer districts select the candidate aligned with the majority party (competitive primaries are less common than competitive elections).

Michigan has term limits. The notional goal is to avoid people that become lifetime representatives; much of the actual outcome is that policy expertise is concentrated in lobbying groups, and well liked, effective legislators are forced out. More rules end up making it easier to work the system.

I don't think your first sentence necessarily holds true - at least it doesn't lately. We just don't see powerful "blue dog democrats" and moderate republicans holding the keys any more.

While moderate candidates have more independent appeal - it's voter turnout that wins elections, even in competitve districts. Appealing to emotions, especially appealing to fear has been extremely successful at getting turnout for particular voters (ie, anti-abortion voters for example). Maybe that will calm down or moderates will turn out more and end this trend (as they did with Biden) - but even competitive races can turn out extreme candidates.

Another possibility to consider, is that what we call "competitive districts" are really just "moderate districts" to begin with. Places the population isn't particularly threatened by or well-served by either party. In that case, it makes sense that they would be both more competitive, and more moderate.

Sure, it isn't an absolute effect, I could have phrased that better. I think it's pretty clearly a directional effect that safer districts pick their representatives in the primary (which often have a candidate backed by the institutional power structure).

But that just reflects the fact that a given party has overwhelming support in those districts. It doesn't mean that the result is unrepresentative. So Phila. and NYC pick their mayors in the Democratic Party primary. The result isn't unrepresentative - there was no question that the mayor was going to be a Democrat.

Yeah, I'm not arguing that it isn't representative, I'm arguing that we shouldn't design for that outcome, because it shifts power in the direction of the party and away from the constituents (and it does so even in districts where the constituents are largely comfortable with the resulting representative).

A lot of that gets to the question of what are we optimizing our k-clusters for?

The idea of geography-based representation spurs from the ideal that a representative should be able to "shake hands" with everyone they represent, right? That a townhall with one's representative might be isomorphic with one's "neighborhood meeting"?

In those cases things like "compact" serve a purpose in going back towards that "neighborhood" ideal (which arguably has never existed in US practice).

But it's 2021 and is geography and the "neighborhood" ideal still useful for us in selecting representatives? When was the last time you shook hands with your representative? When was the last time you had a neighborhood meeting where the representative just swung by for an impromptu townhall?

Maybe we should find a better optimization for our k-means clustering that isn't geography because gerrymandering seems to imply that we'll never fix geographic clustering? With tools like TV and the internet, geography may not even matter like it did to 18th century Americans. A Zoom call isn't a handshake, of course, but we have more tools for virtual shared interest groups than ever before and don't necessarily need to remain tied to the vagaries of geography.

I don't necessarily have good answers for what those metrics/optimizations could be/should look like beyond geography. All I know is that it would be hard to impossible to find good ones under the current two party system and as with most things, the blame for a lot of our problems continues to be directly on the two party system.

The important thing about geography is that cultures tend to develop along with it. That matters a little less today than in the past, but someone in Fresno is still going to have more in common with someone from Los Angeles than someone from Louisiana.

We also already have systems of representation along other lines that feed into geographically-representative government - consumer groups, interest groups, lobbyists, etc.

> someone in Fresno is still going to have more in common with someone from Los Angeles than someone from Louisiana

I question this. Someone from San Francisco is likely to have much more in common with someone from Portland or even NYC or London than they would someone from Bakersfield or Shasta. You can see that in lots of different areas, from migration patterns to election results to what they do on Sunday morning to which media people consume.

We have not shaken hands with our representatives because we capped the size of the House of Representatives because they felt it would be awkward to have so many Representatives in a single hall, or some such. Hard af disagree for me. I want to know my Representative, and in the modern era, there is no reason why we can't have a house of 1,000s of members, or some such. Our congressional districts are much too big.

I'm not sure I agree with this. It seems to create a lot of safe seats but with just one single token representative for the entire group who doesn't have enough power to get an agenda passed.

If those voters were spread across all of the state's representatives then they would have to care what the group thinks, but when they're all crammed into a single token rep they can be more easily ignored.

That sounds bad, but it's really not. First of all, sometimes groups should be ignored. For example, the pig farming agriculture industry of NC is causing huge environmental destruction. If the rest of the state doesn't like the precedent it sets to let these companies contaminate our water supply - then we shouldn't have to all compromise our values just because people in that industry want special treatment and command a lot of money / swing votes in multiple districts. Secondly, it incentivizes disticts to elect representatives who are good at coaltion building and who are capable of compromising and winning over allies to actually get stuff done.

Who gets to decide which groups "should be ignored". The pig farmers certainly don't agree. The whole point of competitive districts is that the groups who should be ignored will be eliminated by competition. Your whole premise seems to set up all districts with functional lifetime appointments rather than elections.

In competitive districts, politicians are far more dependent on campaign funding. Competitive elections are extremely expensive. So the group that gets ignored is just the group that has the least money to throw at candidates. It's the average person that always loses in that case. That's not fair either.

I would certainly prefer term limits, but if a population thinks their representative does a good job representing their needs - why shouldn't they get re-elected? Nobody is stopping them from having a competitive primary. Remember, AOC only got elected because her district was so non-competitive for republicans, that the establishment didn't bother channeling massive money into the primary for the incumbent. So this setup would actually make it easier for non-establishment candidates to sneak in, and even incentivize the other party to run a non-traditional candidate to compete (ie, like how dems used to win with blue dog democrats in rural areas).

> In competitive districts, politicians are far more dependent on campaign funding.

If we're dreaming big here, how about we try fixing campaign funding too? To give a concrete suggestion, the US should pass a constitutional amendment which allows Congress to limit expenditure on political advertising (but not other forms of political speech). Here is one such approach:

https://www.movetoamend.org/amendment

> The representative system works best when it's able to group people with similar interests and values and lets them elect someone who can share those interests.

This sounds like it would essentially result in big differences in economic status and in a few decades pit people against each other. It would lead to "let's put retired NIMBYs over here and social housing over there and let each vote for representation of their extreme views". I've never seen it lead to good results.