Hacker News Reader

Comment by AnthonyMouse | Hacker News Reader

A switch to proportional voting could require constitutional changes in most places. Also, proportional voting introduces all kinds of unnecessary problems like not having districts and thereby constituents having their votes diluted and not having a specific representative. Which is one things the American system gets right.

What we really need is a switch from "first past the post" to approval/score/range voting, which would dissolve the two party system by eliminating spoilers and thereby making third parties viable.

Score/range voting is "that thing the Olympics uses"; approval voting is "that thing the Olympics uses if the only possible scores a judge can give are 0 or 1".

As far as I can tell the biggest impediment to this is a lot of people proposing alternative systems that aren't as good (e.g. IRV) and then no change is made because proponents of change are divided on which change to make.

View on HN

> A switch to proportional voting could require constitutional changes in most places.

Less places than you might think. It especially wouldn't for Congress (so long as it was proportional by a method similar to STV within state delegations, or within some subset of state delegations), as the only barrier there is a statutory prohibition on at-large districts for delegations greater than 1, adopted to head off the use of FPTP at-large districts to systematically deny representation not minorities.

> Also, proportional voting introduces all kinds of unnecessary problems like not having districts

Party list proportional would do that, but most other proportional systems would not.

> and thereby constituents having their votes diluted

Proportionality doesn't dilute votes; the more proportional a system is the more efficiently it allocated votes to give them maximum effect.

> and not having a specific representative

Proportional systems don't need to have that effect, either, no matter how you measure it (and FPTP—and any other simple singlr-member district system—definitely has that effect for supporters of the nonwinning party in a district, who are effectively unrepresented entirely.)

> Proportionality doesn't dilute votes

It causes your vote to be mixed with a larger number of other votes, which is the definition of dilution. By giving the candidates more alternative constituents to win over in order to stay in office, your vote is less important to them and there are more ways they can screw you over and still stay in office.

> the more proportional a system is the more efficiently it allocated votes to give them maximum effect.

That isn't inherently true. Suppose you have a state with two districts. 60% of the state are religious conservatives, 30% are libertarians, 10% are socialists. With proportional representation the socialists get disenfranchised because they don't get their own candidate. Even the libertarians might get disenfranchised if the religious conservatives manage to get two representatives. But with range voting, you end up with a candidate who has to make all the interests in their district as happy as possible because anyone who comes along and can make them happier would defeat them in an election -- so nobody gets disenfranchised because every victor needs to make everybody happier than anybody else.

Proportional systems also fail in the same way but worse when forming governing coalitions. Suppose you have a White party and a Black party and the White party voters are 70% of the population. Then proportional representation gives them 70% of the representatives, they form the majority coalition and the Black party representatives lose every vote.

Compare this to range voting where if you so much as have a White party, they lose even in many majority-White districts against a moderate candidate who can earn the support of both White and Black constituents. It promotes the election of moderates with broad appeal over the election of extremists who fight to gain a majority coalition that can steamroll over anyone in the outgroup.

> constituents having their votes diluted

Wut? Literally every vote would be worth the same under proportional voting, that’s kind of the point isn’t it?

Of course that means that people whose vote is currently outrageously relevant will have their vote dilluted.

"Diluted" means "spread out". AnthonyMouse's point is that voters would no longer vote for a particular representative. Their vote would be "spread out" across the whole party list.

(dragonwriter disagrees https://news.ycombinator.com/item?id=25549671, but I'm not sure what he/she is referring to)

There is this thing, where you vote for party, but give ratings for the representatives on party list. So basically if a guy is on the list down bellow but a lot of voters pick to prioritize him, then he moves to top. This happenes during ellections couple of months ago i LT, where guy was on 141 position, but still got a seat in parlament.

Edit: that party got 50 seats total.

> What we really need is a switch from "first past the post" to approval/score/range voting, which would dissolve the two party system by eliminating spoilers and thereby making third parties viable.

Systems like approval voting or STV when used with single member districts do not reduce the two party duopoly. To achieve this you must use multi member districts, which has the effect of reducing the threshold to gain a seat. The best case study for this is maybe Australia, which has been running systems like this for a century. In their Senate they elect using very large multi member districts and have a representation from a number of parties, whereas in their House they use single member districts. And the result there is a two party duopoly.

For example at the moment in their Senate (multi member districts) currently there are seven parties and one independent:

  Liberal (31)
  Labor (26)
  Greens (9)
  National (5)
  One Nation (2)
  Centre Alliance (1)
  Lambie Network (1)
  Independent (1)

On the other hand the Australian House uses single member districts and third parties do much worse. For example look at how the Greens have a whopping 11% of the seats in the Senate above but an absurd 0.6% in the House below. This absurd 20x discrepancy for the Greens in proportion of seats between the Senate and House is the impact of whether you use multi member vs single member districts with RCV. And the size of the multi member districts will determine the size of the impact.

  Labor (68)
  Liberal (61)
  National (15)
  Liberal National (1)
  Greens (1)
  KAP (1)
  Centre Alliance (1)
  Independent (3)

As demonstrated here the Greens are clearly very popular with people but they just can't win in the duopoly reinforcing single-member districts. I would rather see us work to achieve an outcome like the the Australian Senate than the Australian House.

WRT your concern about who is your representative in the multi member case. It can be argued now ~40% of people who vote for the loser in our current system don’t have a representative, and a more proportional system would clearly improve this. So I think, if say we had RCV with five member districts, you could either just contact whichever member you prefer, contact them all, or they could introduce joint offices to work together, or a system could be set up to assign them to “sub district” that would be akin to our current districts. For example assign them based on who gets the most first preference votes in a district. Or rotate them every six months between sub districts. This might even reduce partisanship.

> Systems like approval voting or STV when used with single member districts do not reduce the two party duopoly.

What do you mean?

Suppose you have a single member district with four parties running: Democrats, Republicans, Greens and Libertarians. First past the post polling shows 48% Democrat, 48% Republican, 2% Green and 2% Libertarian. Your preferred candidate is polling at 2%.

You're then going to vote for one of the major parties because you know perfectly well that your preferred candidate is not going to win and you also know that you prefer one of the major party candidates over the other one.

By contrast, with approval voting, all four candidates get around 50% approval because there are no spoilers and no reason not to approve of the candidate you actually want, so one of the third parties can plausibly edge out both of the major parties for the highest approval and win the district. They also have a much better case that they could win which means they can get into the debates etc.

Also, approval voting makes it plausible for independents to win when they tailor their platform to the district, which just might destroy the political parties whatsoever -- and that would be most excellent.

> It can be argued now ~40% of people who vote for the loser in our current system don’t have a representative, and a more proportional system would clearly improve this.

No one disputes that the existing system is terrible. But approval voting still fixes that, because a candidate with 60% approval loses to one with 70% approval, so someone who can make more of the district happy wins. And there is no way to give full consideration to 100.0% of the voters short of direct democracy, which has its own set of problems.

To your point, in my state, reformers have been able to change our laws to allow RCV, a much easier lift.

> ...we really need ... approval/score/range voting

Agreed. That's been my long time position.

That said, per The Democracy Nerd podcast, I'm newly curious about multi-member seats. Which might be a smaller lift in a lot of places.

https://xraypod.com/show/democracy-nerd/saving-democracy-in-...

"... Kristin Eberhard from Sightline Institute discusses her proposal to eliminate the state Senate, along with other ideas to save democracy in Oregon."

(Sorry, no timestamp.)

Eberhard claims that prior to eliminating multimember seats, Illinois' legislature was far more productive, constructive.

FWIW, I'm strongly in favor or unicameral legislatures at the state level, and I'm chewing on the notion for the federal level.

The purpose of a bicameral legislature is to have a process for placing them in office that sufficiently differs as to allows the bodies to represent distinct interests in the legislature.

The US Senate was originally appointed by the state legislatures and intended to represent the interests of the state governments at the federal level -- representation that they no longer have, leading to a massive federal takeover of government. Compare the ratio of tax revenues collected by the state vs. federal governments before and after the 17th amendment. There was a drastic, immediate, permanent shift.

Does this mean you support bicameral?

In your future perfect PR world: Still bicameral? PR just for House? Replace House & Senate with a parliament?

Or do you still want to preserve the vetocracy with a Senate (House of Lords)?

What do state, county, city governments look like?

>Also, proportional voting introduces all kinds of unnecessary problems like not having districts and thereby constituents having their votes diluted and not having a specific representative.

I'd rather phrase it, the proportional systems have quite large districts (or alternatively, much more of representatives per district) to achieve the proportional allocation. But I get what you mean.

However, it is very well possible to design variations to any traditional party list proportional system where (at least nominally) one representative is identified with each current-sized and located district.

For example, take a classic proportional system, and use it to select the number of representatives per party list. In proportional methods, usually the list is either preordered by the party or order within the list depends on the individual total popularity of candidates in all districts. So, to achieve nominal geographic representation, one can assign a district to each selected representative. E.G. each district is nominally represented by the one of those elected who had the largest share of votes in that district and is not yet assigned to some other district by the same rule (if one complains that one may have their representative elected by people out of one's district, this implicitly happens in such proportional systems anyway).

Alternatively, one could achieve remarkable amount of geographic identification by relaxing the "one representative <-> one district" requirement. For example, require that each party list can put forward only one candidate per (equally populous) district; a candidate gets personal votes only in that district, but the number of seats given to list is allocated proportionally over all districts, as usual in proportional systems. Then the order within list is assigned by the personal votes of each candidate. (This has the added complication -- or benefit -- that districts with low overall turnout may be left without a specifically named representative and the district with high turnout for all candidates standing there may get them all elected; nevertheless, all votes count towards the party list total sum, no matter the district.)

Reply on news.ycombinator.com