Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Mathematicians looking to make elections in the US more representative

nature.com

Post reply on HN
Print view View original

131–136 of 136 posts

Re: Mathematicians looking to make elections in the US more representative

#131

Post by joveian »

jamesonquinn wrote: post #126

As others have noted, proportional representation (PR) is the best solution to gerrymandering. (Not arguing that we shouldn't also support second-best solutions like nonpartisan redistricting and court challenges to the worst gerrymanders.) The three PR methods in common use are STV-with-multimember-districts; mixed-member proportional (MMP); and open (or, yuck, closed) list systems. All of these (except closed lists…

Do you have a more detailed desription about how this method works? Your links do not describe it in much detail but it is interesting.

Even without the details my sense is that this method suffers from some very serious problems. Picking the two candidates with the highest initial votes seems like a bad idea; this can remove all of the most widely supported candidates if there are enough similar candidates (i.e. if 70% of the voters are split between 10 candidates and 30% between two candidates, but the 70% consider the two that the 30% would elect to be least favored).

In general, strategic voting is likely to have a huge effect. I don't understand what the write-in option does, but if it allows you to vote for candidates in a different district than that would allow votes to be used to disloge more influential members of an opposing party against the wishes of the district that candidate actually represents.

In any case, always interesting to learn about possible voting methods :).

IMO, limiting the effectiveness of strategic voting is one of the most important things a voting method should do, while also not electing anyone who would heavily lose in two candidate races vs. any of the other candidates. I don't think it needs to be strictly Condorcet, but the further away from that a method gets (at least when geographic representation is involed) the less reasonable it seems in a lot of ways IMO.

I've been thinking about the possibility of a parallel system where the grographical representation part is non-partisan. I think this might work if geographical representatives are allowed and expected to allocate their vote on a given issue proportionally to how the people they represent would vote on that issue.

Re: Mathematicians looking to make elections in the US more representative

#132

Post by thaumasiotes »

ouid wrote: post #130
thaumasiotes wrote: post #102

Earlier quoted context omitted.

Think about it for more than half a second. Moving a Democrat into a Democratic district makes that district's Democratic seat more secure. At the same time, moving that Democrat out of your district makes your Republican seat more secure. You can't do one without the other.

Probability of victory is highly non-linear in the proportion of voters. It's so non-linear that it makes sense to treat it as a phase transition. A district is secure when the outcome is certain. There are two distinct phases where the outcome is "secure" and the border between these two phases is, because of the non-linearity, very small. That is what secure means to most people. The specific language that the OC,…

As I just pointed out to you, it can never happen to one party without simultaneously happening to the other party. Packing Democrats into an already-secure Democratic district adds Democratic voter proportion to their district and adds Republican voter proportion to yours.

If you don't believe that this can be in the interest of either party (since it's happening "to" both of them), you don't have much left to complain about.

Re: Mathematicians looking to make elections in the US more representative

#133

Post by taejo »

travmatt wrote: post #91
Torwald wrote: post #90

Earlier quoted context omitted.

Could this be mitigated by having a system with more than two parties?

Multi party systems tend to favor the most ideologically driven actors in the system. An example of this is Israel. Furthermore, American politics isn't two ideologies. Each party is composed of multiple groups that compete for dominance. Republicans have business interests, evangelicals, etc., while democrats have progressives, neoliberal, etc.

I'm not sure I follow. American politics isn't two ideologies, but it does have ideologies. Is it better to bundle up these dozens of ideologies and interests into two parties than to have ideologically driven parties?

Re: Mathematicians looking to make elections in the US more representative

#134

Post by a_imho »

TheCoelacanth wrote: post #129
a_imho wrote: post #68

Sortition >> Voting

I think this is a really interesting idea, and I would love to see it tried in a modern government. There are a lot of potential problems, but possibly less than those involved with voting.

From what I understand, the US legal system has the jury duty as a form of sortition, who make life or death decisions.

It is a solution to make the governing bodies more inclusive and representative, which is the title of the article. No winner takes all, no fake news, no votes buying, no suspicious campaign donations, probably no 2 party system, less oligarchies, corruption and nepotism. What is not to like? Wikipedia [1] does list a couple of points, all debatable and/or easily countered imo.

I also happen to like how seamlessly it integrates with the current representative model and remains scalable as opposed to more direct approaches.

[1] https://en.wikipedia.org/wiki/Sortition#Disadvantages

from the average voter pov, who benefit little to none from the current establishment in this presumption

Re: Mathematicians looking to make elections in the US more representative

#135

Post by TheCoelacanth »

a_imho wrote: post #134
TheCoelacanth wrote: post #129

Earlier quoted context omitted.

I think this is a really interesting idea, and I would love to see it tried in a modern government. There are a lot of potential problems, but possibly less than those involved with voting.

From what I understand, the US legal system has the jury duty as a form of sortition, who make life or death decisions. It is a solution to make the governing bodies more inclusive and representative, which is the title of the article. No winner takes all, no fake news, no votes buying, no suspicious campaign donations, probably no 2 party system, less oligarchies, corruption and nepotism. What is not to like ? Wikip…

The big difference that I see from juries, is that juries only choose from a fairly limited selection of possibilities, and they have a few specifically identified professionals (the judge and the lawyers for each side) who guide them through the process. With an actual government, there are only a few constraints and everything else is open to them. There are also thousands of interested parties trying to influence them instead of just two. That would make their job much more challenging than being a juror. It could potentially end up handing more power to lobbyists because of the inexperience of the people in power and resulting in more corruption because there are no longer have to hide it to ensure re-election.

Re: Mathematicians looking to make elections in the US more representative

#136

Post by a_imho »

TheCoelacanth wrote: post #135
a_imho wrote: post #134

Earlier quoted context omitted.

From what I understand, the US legal system has the jury duty as a form of sortition, who make life or death decisions. It is a solution to make the governing bodies more inclusive and representative, which is the title of the article. No winner takes all, no fake news, no votes buying, no suspicious campaign donations, probably no 2 party system, less oligarchies, corruption and nepotism. What is not to like ? Wikip…

The big difference that I see from juries, is that juries only choose from a fairly limited selection of possibilities, and they have a few specifically identified professionals (the judge and the lawyers for each side) who guide them through the process. With an actual government, there are only a few constraints and everything else is open to them. There are also thousands of interested parties trying to influence…

I mean it is already practiced in some form, this is not a whole new concept.

Of course, there are discussions to be held about the implementation details, including the governments reach and the checks and regulations. A lot can be argued about what political experience brings to the table (imo bad things for the very large part, but that is a ugly can of worms). Personally, I would trust my neighbor more to do The Right Thing, than any career politician.

Nevertheless, a sampling based method is much more representative and more in line with the rule of the people idea.

Post reply on HN