The most viable Asset Voting

[Clay Shentrup]

I've been thinking about the most politically viable form of Asset Voting. There are a couple of questions that stick out in my mind, with regard to potential objections and logistics concerns.

One issue is, do you make the delegation optional? In that case, I think the best way is to piggyback on top of the existing mechanism which ensures certain voters only get sections of the ballot that are applicable to them. For instance, if I'm voting in Berkeley's District 3, I don't even receive the sections for the other districts. You could simply ask people, "do you want a transferable ballot for rent board, or a non-transferable ballot?" Then it's technically two different elections, but you can just sum the results quite easily. The cost is just that you have to ask people.

Another approach would be to have two sets of candidates. You can vote for "John Doe (without transferability)" or "John Doe (with transferability)". Seems weird though.

You could also add a simple checkbox "Do you want your candidates in this race to be able to transfer your vote to other candidates?"

It's not clear whether making this part optional raises more concerns than it addresses.

Secondly, should one go with the simplest implementation, where you only get one vote? An alternative would be to make it like cumulative voting, where you get as many votes as winners, but can assign more than one to the same candidate. That may be more sellable because it has more precedent. And then you're just tacking on the part about transferability as a "little local modification—pay no attention to it, nothing to see here". :)

Simplicity and familiarity/precedent are clearly not always the same thing.

[Gabriel Bodeen]

In the spirit of the book "Nudge", I think SODA's design has it right.  It's like your third option, except it makes transferability the default option.  Then the only non-transferable votes would be from people who actually don't want it, which could help make most voters' votes more productive according to the voters' own opinions.

[Jameson Quinn]

SODA is a single-winner system (and, of course I'd agree, a good one); but mostly, asset is discussed in the context of multi-winner.

What corresponds to SODA in a multi-winner context is basically PAL representation. PAL has a few extra things too:

• By only listing the "local" candidates, and making far-away candidates available only as write-ins, PAL simplifies the ballot.
• Once the slate of winners is decided, PAL assigns representatives to voters, so that each district has one representative from each party, and thus all voters (except at most a Droop quota) can know who "their local representative" is, whom they helped elect.

I'd like to have a deeper discussion of PAL here. I believe that, as a good drop-in replacement for single-member districts which preserves ballot simplicity and local representativity, it is the best proposal for the US. 

Clay, I think you disagree with me on that. What are your issues with PAL?

Jameson

--
You received this message because you are subscribed to the Google Groups "The Center for Election Science" group.
To unsubscribe from this group and stop receiving emails from it, send an email to electionscien...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Clay Shentrup]

I think PAL has about zero chance of going anywhere in the USA, so I can't devote a lot of time to thinking about it's theoretical aspects.

I think SODA might be worth considering here, it's just that it's name is so bad that I can never remember what it actually entails.

[Jameson Quinn]

Yet you talk about RRV and the like. Why do you think those are more worth discussing than PAL?

2014-03-11 13:08 GMT-04:00 Clay Shentrup <cl...@electology.org>:

I think PAL has about zero chance of going anywhere in the USA, so I can't devote a lot of time to thinking about it's theoretical aspects.

I think SODA might be worth considering here, it's just that it's name is so bad that I can never remember what it actually entails.

--

[Jameson Quinn]

Let me make a slightly more thorough pitch.

When you bring up PR (with a US/UK/Canada audience), there are basically 4-6 possible objections:

1. But what about local representation?

1a) Local accountability

1b) right of petition

2. Complexity

2a) ballot complexity / voter cognitive burden

2b) system complexity / shut up, my brain hurts / this is a trick / there must be a catch

3. Multiparty gridlock

4. (politicians only) But what about "my" seat?

Here's how you might respond:

1a) Definitely: "my system is not closed list." Ideally: "my system is not open list." 

1b) "my system is mixed-member"; or "...multi-member"; or "...PAL-like." Or: "this is not important."

2a) "My system uses delegation"; or "...closed list"; or, to some extent, "...open list". At the very least least, "my system is not fully ranked (STV)."

2b) Since no PR system is simple enough to dodge this issue, the only real responses are to use simple examples, and if possible to develop trust in other ways.

3) Pretty much any PR system could be objected to in this way. You basically have to confront this one head-on, with historical/international examples where it's no problem. Luckily, this is a rare objection, something only poli-sci types tend to come up with.

4) Basically, PAL-like systems are the only way I know of to answer this.

In the end, if you want to have good answers for 1a, 1b, 2a, and 4, you'll end up designing something that's a lot like PAL. Asset does well against many of these objections, but the more you try to make it compatible with existing institutions and laws, the more you are pushed towards something like PAL.

Jameson

[Jameson Quinn]

In my previous pitch, I didn't fully explain the situation with Asset.

Basically, there are two ways to go with Asset. You can either do full-fledged asset (either irrevocable or revocable), in which voting weights and number of representatives are both fluid; or you can try to guarantee that it will elect a fixed number of equal-weight representatives. There is a lot to recommend the former system theoretically, but constitutionally, it's a no-go. So, we're left with the latter.

And that's harder than it looks. If you just set a quota and have a free-for-all, there is good reason to believe some seats will go empty. For instance, two similar candidates with over half a quota of votes each could get into a chicken dilemma, each insisting that the other should be the one to cede their votes.

I've thought about this. There are various ways to go — including eliminations, shrinking quotas, fixed delegation orders, enforceable pledges, and more. Most of these have problems, as anybody familiar with IRV's issues could probably guess. And whatever you end up with, it is a long way from the clean simplicity of pure asset. I think people who haven't gone through the exercise of designing an asset-based system like this would be surprised at how far.

PAL is one attempt to square that circle. It may not be the best possible answer. But if you don't think it is, then you should propose something else.

Of course, you can also give up on asset. There's plenty of systems — RRV, STV, BTV, etc. — which have no delegation aspect. But then you have pretty unavoidable problems of ballot complexity. And there's also party list systems, which I guess you could call semi-delegated; but even the open-list ones of those have issues of accountability.

So yeah; I wish I could make PAL simpler. But I don't think that there's any option that is constitutional and doesn't have that problem. So I think PAL is very much worth considering.

Jameson

[Clay Shentrup]

On Tuesday, March 11, 2014 10:16:53 AM UTC-7, Jameson Quinn wrote:

Yet you talk about RRV and the like. Why do you think those are more worth discussing than PAL?

RRV's complexity is purely in the tabulation. Voters know that they just score the candidates and get n winners.

Your description of PAL includes some notion of "local" candidates "having a representative from each party for each district" and other fundamental changes whose purpose even I can grasp.

Spending some significant time in real world political activism (going to local political club meetings, pitching ideas to your elected officials) will really benefit you.

[Jameson Quinn]

2014-03-12 0:26 GMT-04:00 Clay Shentrup <cl...@electology.org>:

On Tuesday, March 11, 2014 10:16:53 AM UTC-7, Jameson Quinn wrote:

Yet you talk about RRV and the like. Why do you think those are more worth discussing than PAL?

RRV's complexity is purely in the tabulation. Voters know that they just score the candidates

...which is more complex than just voting for one, as in the current system and in PAL...

 

and get n winners.

...which is arguably more complex than having your own representative. If you want to call your representative to support some bill, or need help with an immigration issue, whom do you contact?

Your description of PAL includes some notion of "local" candidates

... which voters are used to ...

 

"having a representative from each party for each district" and other fundamental changes whose purpose even I can grasp.

Spending some significant time in real world political activism (going to local political club meetings, pitching ideas to your elected officials) will really benefit you.

In this, you sound like Rob Richie. I'm not saying that you're as wrong as he is, just that whenever someone talks about how other people are going to react to something, it's going to be some mixture of insight/experience and projection/oversimplification, and the person making the claim is, almost by definition, likely to underestimate how much of the latter there is. Thus, while there's certainly a place for this kind of political intuition, when RR starts to fall back to blustering in this vein, we can tell it's only because he doesn't have a logical leg to stand on.

Concretely, I have done the kinds of real world activism you suggest to some extent, though I'm sure less than you have.

Jameson

[Jameson Quinn]

I'd like to clarify my motives for being a bit insistent in this discussion. I do honestly think that PAL, or something like it, is likely to be the most viable proposal for the US in the mid term. And I'm prepared to defend that statement. But even if I'm wrong, I think that by talking through this disagreement, we'll refine our arguments. 

In other words: ideally, one of us would change our mind, and we'd end up agreeing about PAL. But even if that doesn't happen, I think you (Clay) will end up better able to effectively promote your preferred PR method, and I'll end up better able to effectively promote PAL. I think that's worth our time.

Jameson

[William Waugh]

Tell me again, what is PAL?

[Frank]

I was about to ask the same question. Same with SODA. A web search was not helpful.

On Thursday, March 27, 2014, William Waugh <google_wil...@spamgourmet.com> wrote:

Tell me again, what is PAL?

--

You received this message because you are subscribed to the Google Groups "The Center for Election Science" group.
To unsubscribe from this group and stop receiving emails from it, send an email to electionscien...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--
P.S.: I prefer to be reached on BitMessage at BM-2D8txNiU7b84d2tgqvJQdgBog6A69oDAx6

[Jameson Quinn]

Sadly, it seems Electorama is down, or I'd point you there for system definitions. I've emailed Rob Lanphier to see if he can fix it.

So: SODA is "simple optionally-delegated voting":

1. Candidates predeclare strict delegation preferences.

2. Voters vote approval ballots, but if you vote for one candidate only, your vote is considered delegated, and that candidate will be able to (effectively) add approvals to your ballot if they are not able to win.

3. In descending order of number of votes held, candidates assign the approvals on the ballots delegated to them. (Descending order, because that way, in a chicken dilemma situation, the stronger candidate will get the first chance to not share votes, so by the time the weaker candidate chooses, they will have a clear choice of sharing their votes with the strong candidate they prefer or letting one they like less win.)

PAL is a similar PR system. You can see an explanation (with graphics!) here.

Jameson

2014-03-27 20:54 GMT-04:00 William Waugh <google_wil...@spamgourmet.com>:

Tell me again, what is PAL?

--

[Clay Shentrup]

For what it's worth, I think SODA is too complicated to be practical.

[Jameson Quinn]

About SODA, I basically agree with you, Clay. That is to say: even though SODA is actually simpler than approval for voters,* Approval is clearly the first step, for its simplicity.

However, I don't think that argument works against PAL representation. There is no true PR system which the average voter would understand. STV, RRV, open list... all of these involve technical details and quota calculations which most voters will tune out on. So I think that we can't hold out for perfect simplicity if we want PR.

Of course, we still have to listen to voters, and complexity is one of the things they often complain about. But, as I said in an earlier message, that usually amounts to one of two underlying concerns. 

For some voters, it's a matter of trust. If a system seems too complex, they are concerned that the details may hide an ideological agenda; for instance, that it might give an unfair disadvantage to large or small parties. I think that the way to deal with these trust issues is to face them head on. Try to understand where people coming from; if they're a major-party supporter, you might say "I know that sounds like a lot of details, and I can go through that more slowly if you want me to. But the important thing is that the whole system works to ensure that things are proportional, that smaller parties are getting as close to their fair share of representation as we can give them, but not more." (Similarly, you'd spin it in the other direction for a small-party audience.)

For other voters, the concern is not with the complexity of vote-counting, but with the difficulty of voting itself. In this sense, PAL is ideal; since it's a delegated system, you can simply vote for your favorite candidate. That's actually a lot simpler than something like STV or RRV.

And in my experience, talking to regular people, complexity is only one concern. A lot of people are concerned with whether they'll still have "their" local representative. In this regard, PAL is actually much better than the current system. Since each district gets one representative per party, each** voter can easily find, not just a local representative, but a sympathetic local representative whom they helped elect. Of course, that means that each representative's territory covers multiple districts. But pretty much everyone would rather have a representative from their own party with constituents spread over a slightly larger area, than one from an opposing party just because that's what 51% of their close neighbors wanted.

Clay, I'd really like to hear how you respond to these arguments.

Cheers,

Jameson

*SODA is simpler for voters because they can simply vote for one and go home. I can actually prove a property about this: assuming that all voters agree with their candidate's predeclared ranking, and a CW exists, then there is a (delegated) strong Nash equilibrium in which the CW wins. In other words, under these (strong) assumptions, SODA beats Gibbard-Satterthwaite, in a way that is impossible for any system without a predeclaration step. I am not arguing that these assumptions will actually hold in all real elections, but I do think it is plausible that would-be strategic voters will not be able to foresee how they would be broken, and so will be unable to find a strategy that beats honest delegation.

**Of course, there could still be up to a Droop quota of voters without effective representation. In most cases, though, it will be even less than that.

2014-03-27 23:36 GMT-04:00 Clay Shentrup <cl...@electology.org>:

For what it's worth, I think SODA is too complicated to be practical.

--

[Clay Shentrup]

On Friday, March 28, 2014 8:43:34 AM UTC-7, Jameson Quinn wrote:

However, I don't think that argument works against PAL representation. There is no true PR system which the average voter would understand. STV, RRV, open list... all of these involve technical details and quota calculations which most voters will tune out on.

There are different kinds of complexity. A few types that stand out are:

• tabulation effort required by administrators.
• clarity of the algorithm
• complexity of determining how one "should" vote
• complexity of understanding the overall nature of the voting process.
  

E.g. say you use the strategy-free Score Voting system where two probability distributions are selected, then the winner is picked by implementing the distribution with a higher "expected total score". Because it's free of strategy, it's trivial for a voter to decide how to fill out a ballot. Just give sincere scores. But the algorithm is very complex to understand. The game theoretical underpinnings will seem abstract to most voters. Although, tabulation is still quite simple.

Contrast to IRV, which is about as algorithmically/conceptually complex (some might argue even less so), but much more labor-intensive for administrators.

With the third item, we can see a marked difference between Score and Approval Voting. With Score, a voter who thinks a candidate is mediocre can just give that candidate a middle score, and be done with it. With Approval, he may vacillate, unsure which side of the line that candidate should fall on. So even though Approval is simpler in the first two respects, it can take voters longer to complete, and can be perceived as "harder" to use.

As for the fourth item, consider MMP. To American voters, it may seem kind of weird to have this connection between a party list and a local district representative. The actual voting is straightforward—you pick a local candidate, and you pick a party. But the idea that your party gets docked a "list seat" for every local candidate who wins from that party—that's a little abstract.

Contrast that with RRV. The algorithm may be complex, but the nature of the process is easy to understand. A group of candidates goes in, and a subset of winners comes out. You rate them to indicate your support. Even the complex algorithm itself can be glossed over as, "the candidates with the most points win, but your vote loses strength based on how much it affected the race so far".

You recently said something about PAL:

• By only listing the "local" candidates, and making far-away candidates available only as write-ins, PAL simplifies the ballot.
• Once the slate of winners is decided, PAL assigns representatives to voters, so that each district has one representative from each party, and thus all voters (except at most a Droop quota) can know who "their local representative" is, whom they helped elect.

This immediately seems esoteric. There are "local" candidates..but you can vote for non-local candidates as write-ins? Huh? I don't get it. This literally seems weird to me and scares me off as some overly contrived and arbitrary scheme, right up front. And I'm a voting theory geek! This is kind of complexity that I think is most concerning. It's more like MMP in that it makes the very nature of voting more abstract.

 

So I think that we can't hold out for perfect simplicity if we want PR.

True. Though I think Asset Voting is amazingly simple. "Optional" Asset Voting (Optional Transfer Voting, as I've referred to it) is maybe even better, because it gives voters a choice of whether they want their votes to be transferable.

I can't respond in much more detail without understanding how PAL works.

[Jameson Quinn]

Thanks for the response, Clay. This is exactly the kind of conversation I think is productive to have about PR.

2014-03-29 17:09 GMT-04:00 Clay Shentrup <cl...@electology.org>:

On Friday, March 28, 2014 8:43:34 AM UTC-7, Jameson Quinn wrote:

However, I don't think that argument works against PAL representation. There is no true PR system which the average voter would understand. STV, RRV, open list... all of these involve technical details and quota calculations which most voters will tune out on.

There are different kinds of complexity. A few types that stand out are:

• tabulation effort required by administrators.
• clarity of the algorithm
• complexity of determining how one "should" vote
• complexity of understanding the overall nature of the voting process.
  

I think this is a good list, thanks. 

E.g. say you use the strategy-free Score Voting system where two probability distributions are selected, then the winner is picked by implementing the distribution with a higher "expected total score". Because it's free of strategy, it's trivial for a voter to decide how to fill out a ballot. Just give sincere scores. But the algorithm is very complex to understand. The game theoretical underpinnings will seem abstract to most voters. Although, tabulation is still quite simple. 

I doubt that there are any PR systems which are this simple for tabulations, or this provably strategy-free.

 

Contrast to IRV, which is about as algorithmically/conceptually complex (some might argue even less so), but much more labor-intensive for administrators.

With the third item, we can see a marked difference between Score and Approval Voting. With Score, a voter who thinks a candidate is mediocre can just give that candidate a middle score, and be done with it. With Approval, he may vacillate, unsure which side of the line that candidate should fall on. So even though Approval is simpler in the first two respects, it can take voters longer to complete, and can be perceived as "harder" to use.

I've answered this in a separate thread, because I think you're bringing up important issues, but it's liable to distract us from the question of which PR system is most viable, which is what I want to focus on in this thread. 

As for the fourth item, consider MMP. To American voters, it may seem kind of weird to have this connection between a party list and a local district representative. The actual voting is straightforward—you pick a local candidate, and you pick a party. But the idea that your party gets docked a "list seat" for every local candidate who wins from that party—that's a little abstract.

Contrast that with RRV. The algorithm may be complex, but the nature of the process is easy to understand. A group of candidates goes in, and a subset of winners comes out. You rate them to indicate your support. Even the complex algorithm itself can be glossed over as, "the candidates with the most points win, but your vote loses strength based on how much it affected the race so far".

You recently said something about PAL:

• By only listing the "local" candidates, and making far-away candidates available only as write-ins, PAL simplifies the ballot.
• Once the slate of winners is decided, PAL assigns representatives to voters, so that each district has one representative from each party, and thus all voters (except at most a Droop quota) can know who "their local representative" is, whom they helped elect.

OK. Let me try to explain PAL in simple terms, as I'd present it to a voter. Anything in (parentheses) is something I wouldn't explain unless specifically asked, but still in simple terms that I think would work with an ordinary voter.

The ballot looks pretty much the same as the one you recognize: basically, one local candidate from each party. (Some parties, especially smaller ones, might decide to run the same candidate in several districts; while in other parties, there might be more than one candidate who got enough signatures locally to get on the ballot.)

You vote for the candidate you like the best. If there's a candidate from across the state who you like better, even though they're not on your ballot, you're free to write them in.

Anybody who gets enough votes to earn a seat is elected. Then, to fill the rest of the seats, the weakest candidates are eliminated, and their votes are passed on — first, to the "same faction" from the same party, as defined by the candidate themself and as listed on the ballot; then to the rest of the candidates in their party; then to other parties in the order he or she pre-declared. This continues until all the seats are filled. That guarantees that a large majority of the votes will help elect a representative they supported directly or indirectly; for instance, in a state with 9 seats, at least 90%.

Now, in order to ensure that voters can know who "their own" representative is, each winner is assigned a multi-district territory, so that each district ends up in the territory of one representative for each winning party. (As much as possible, the assignment process also maximizes the amount of support they got from inside their territory, and equalizes the total support for the party as a whole in each representative's territory.)

Say you lived in a strongly-Democratic area of a 9-representative state where the statewide totals were 32% Republican, 32% Democrat, 21% Libertarian, 11% Green, and 4% other. The winners would then be 3 Republicans, 3 Democrats, 2 Libertarians, and 1 Green. Your district would then be represented by one Democrat, whose territory would probably be your district and one neighboring one; one Republican, whose territory would probably include 3 or 4 districts overall; one Libertarian, whose territory would probably be 4 or 5 districts; and the Green, whose territory would be the whole 9-district state. If you supported any of those parties besides the Democrats, you'd still be able to write to your representative, and expect them to be sympathetic to most of your views. And for local issues without a partisan divide, you could go to any or all of the four; at least the Democrat, in particular, would be sure to be from somewhere nearby.

 

This immediately seems esoteric. There are "local" candidates..but you can vote for non-local candidates as write-ins? Huh? I don't get it. This literally seems weird to me and scares me off as some overly contrived and arbitrary scheme, right up front. And I'm a voting theory geek! This is kind of complexity that I think is most concerning. It's more like MMP in that it makes the very nature of voting more abstract.

I hope my explanation above addressed some of these concerns. If it didn't, I'd challenge you to choose your favorite PR system, and explain it at at the level you'd like the average voter to understand it. (Of course, I'm not claiming that's an easy task, and I promise I won't be waiting to pounce or nitpick.)

 

So I think that we can't hold out for perfect simplicity if we want PR.

True. Though I think Asset Voting is amazingly simple. "Optional" Asset Voting (Optional Transfer Voting, as I've referred to it) is maybe even better, because it gives voters a choice of whether they want their votes to be transferable.

That's true. Asset is great. But there are constitutional issues. I think we need at least one good non-asset proposal in our toolbox, and that's what I'm trying to focus on here. 

Thanks,

Jameson

[Frank]

I think I am a bit behind in this conversation. I am not sure how PR is necessarily better than the traditional "American model", for lack of another term, of representation. Meanwhile, or maybe because of that lack of surety (sp?), asset voting (at least SODA and PAL variants) strike Me as needlessly complex. What is the most compelling argument to be made in favor of asset voting? What I have read so on the subject far seems fraught with uncertainty from a Voters perspective.

--

You received this message because you are subscribed to the Google Groups "The Center for Election Science" group.
To unsubscribe from this group and stop receiving emails from it, send an email to electionscien...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Toby Pereira]

I was just looking at the link to PAL that you gave. I think it seems mostly quite good. I think the explanation isn't always that clear in places though. I don't think it's that clear what the graphics in 3 and 4 mean. Also, it says that every winning party gets a representative in each district (in graphic 4) before it's made clear (i.e. in writing) that a representative can have more than one district. So you'll have readers thinking "What I a party only gets one representative?" for quite a while before they find out the answer. It took me a little while to realise what graphic 4 meant and I'm still not sure what graphic 3 means.

 

Also, although every voter will have a representative, if there is only one candidate elected from their favoured party, it's not necessarily going to be a local representative since they could be based anywhere. Presumably since it's a secret ballot, it's up to voters to determine who their representative is? Given that, people might choose different representatives to contact on different issues. Not that this is necessarily a bad thing, but it means that the idea of one representative per person isn't that solid under this system. Also, although people might vote for a local candidate, candidates' own preferences aren't necessarily that likely to be local, so the idea of voting for a local candidate is broken as soon as your candidate is eliminated and your vote transferred. Possibly more of an observation than a criticism, but I think it's worth mentioning given that it's supposed to be about local representatives.

 

While parties with many candidates elected would have them spread geographically so that no individual would have too many people to represent, if you had an independent that got a lot of support nationally, then because they have no party, they would presumably cover all the districts and they might have a disproportionately large number of voters who consider them to be their representative.

 

Is the voting system just STV except with your candidate's ranking rather than your own?

[Warren D Smith]

On 3/27/14, Jameson Quinn <jameso...@gmail.com> wrote:
> Sadly, it seems Electorama is down, or I'd point you there for system
> definitions. I've emailed Rob Lanphier to see if he can fix it.
>
> So: SODA is "simple optionally-delegated voting":
>
> 1. Candidates predeclare strict delegation preferences.
> 2. Voters vote approval ballots, but if you vote for one candidate only,
> your vote is considered delegated, and that candidate will be able to
> (effectively) add approvals to your ballot if they are not able to win.
> 3. In descending order of number of votes held, candidates assign the
> approvals on the ballots delegated to them. (Descending order, because that
> way, in a chicken dilemma situation, the stronger candidate will get the
> first chance to not share votes, so by the time the weaker candidate
> chooses, they will have a clear choice of sharing their votes with the
> strong candidate they prefer or letting one they like less win.)

--Jameson here was forced to explain SODA in a concise way, because
his usual ultra-long impossible-to-read explanation was "sadly" offline.
Notice, he did it tolerably well. If this were edited without lengthening it,
it would become even better.

> PAL is a similar PR system. You can see an explanation (with

> graphics!) here<http://qr.ae/GFCI4>

--Unfortunately, here Jameson was able to refer to an ultra-long explanation
involving numerous extremely annoying and stupid pictures of birds,
gorillas, and
rainbows, therefore he did so instead of saying something concise
and simple we could actually comprehend without a strong urge to vomit.

--My strong suggestion to Jameson is, if he is willing to devote
the energy to write these ultra-long and annoying screeds,
then he should use some of that energy to write short easy to follow
explanations. These can be linked to longer explanations of fine points
(where only motivated readers will follow those links).

--If you explain a voting method with a 500-page book, but are incapable
of explaining it with a half-page self-contained explanation, then
you've failed. Even though the book is much more "impressive."

--Jameson thus, by devoting way more effort and wasting way more time
and paper, actually does a worse job of making anybody like his voting
systems.

[Jameson Quinn]

Responding to Frank:

2014-03-30 15:46 GMT-04:00 Frank <frankdm...@gmail.com>:

I think I am a bit behind in this conversation. I am not sure how PR is necessarily better than the traditional "American model", for lack of another term, of representation.

What do you mean by "American model"? Do you mean:

• A presidential (non-parliamentary) system? Though most PR countries are parliamentary, this is basically a coincidence; there's no a priori reason for these to go together.
• Local representation? But there are PR systems which give that, to a greater or lesser degree.
• Gerrymandering? Blech; of course you don't mean to say that's good.

So at best, you're comparing PR to a single-member district system with a good voting system (say, approval) and with non-gerrymandered redistricting process. The latter is way easier to get wrong, but OK, let's say you manage to obtain it. What are the pros and cons of each?

In PR, you have under 1/6 of of citizens unrepresented, while in single-member districts, you have close to 1/2. On the other hand, with single-member districts, there will be fewer "fringe" representatives; you could argue that this is good because forces people to make larger political coalitions.

I don't think the case against single-member districts is as airtight as the one against plurality. But I'd still clearly prefer PR.

Jameson

[Jameson Quinn]

2014-03-30 17:07 GMT-04:00 Warren D Smith <warre...@gmail.com>:

On 3/27/14, Jameson Quinn <jameso...@gmail.com> wrote:
> Sadly, it seems Electorama is down, or I'd point you there for system
> definitions. I've emailed Rob Lanphier to see if he can fix it.
>
> So: SODA is "simple optionally-delegated voting":
>
> 1. Candidates predeclare strict delegation preferences.
> 2. Voters vote approval ballots, but if you vote for one candidate only,
> your vote is considered delegated, and that candidate will be able to
> (effectively) add approvals to your ballot if they are not able to win.
> 3. In descending order of number of votes held, candidates assign the
> approvals on the ballots delegated to them. (Descending order, because that
> way, in a chicken dilemma situation, the stronger candidate will get the
> first chance to not share votes, so by the time the weaker candidate
> chooses, they will have a clear choice of sharing their votes with the
> strong candidate they prefer or letting one they like less win.)

--Jameson here was forced to explain SODA in a concise way, because
his usual ultra-long impossible-to-read explanation was "sadly" offline.
Notice, he did it tolerably well.  If this were edited without lengthening it,
it would become even better.

Thanks. Any specific suggestions?

 

> PAL is a similar PR system. You can see an explanation (with

> graphics!) here<http://qr.ae/GFCI4>

--Unfortunately, here Jameson was able to refer to an ultra-long explanation
involving numerous extremely annoying and stupid pictures of birds,
gorillas, and
rainbows,

The birds and gorillas aren't mine; they're part of the question, not the answer. The "rainbows" are a standard way of showing a legislative breakdown, going back as far as the French revolution.

[Jameson Quinn]

2014-03-30 15:56 GMT-04:00 Toby Pereira <tdp...@yahoo.co.uk>:

I was just looking at the link to PAL that you gave. I think it seems mostly quite good. I think the explanation isn't always that clear in places though. I don't think it's that clear what the graphics in 3 and 4 mean. Also, it says that every winning party gets a representative in each district (in graphic 4) before it's made clear (i.e. in writing) that a representative can have more than one district.

I fixed it up to a new version, but then realized that Piktochart no longer allows export to a graphics file without subscription. I'd happily pay (it's a good tool for the job) but I don't want a monthly subscription. I'll take a screenshots the next time I have a nice tall monitor but meanwhile, oh well.

 

So you'll have readers thinking "What I a party only gets one representative?" for quite a while before they find out the answer. It took me a little while to realise what graphic 4 meant and I'm still not sure what graphic 3 means.

 

Also, although every voter will have a representative, if there is only one candidate elected from their favoured party, it's not necessarily going to be a local representative since they could be based anywhere. Presumably since it's a secret ballot, it's up to voters to determine who their representative is? Given that, people might choose different representatives to contact on different issues. Not that this is necessarily a bad thing, but it means that the idea of one representative per person isn't that solid under this system. Also, although people might vote for a local candidate, candidates' own preferences aren't necessarily that likely to be local, so the idea of voting for a local candidate is broken as soon as your candidate is eliminated and your vote transferred. Possibly more of an observation than a criticism, but I think it's worth mentioning given that it's supposed to be about local representatives.

 

While parties with many candidates elected would have them spread geographically so that no individual would have too many people to represent, if you had an independent that got a lot of support nationally, then because they have no party, they would presumably cover all the districts and they might have a disproportionately large number of voters who consider them to be their representative.

All fair points.

 

 

Is the voting system just STV except with your candidate's ranking rather than your own?

Well, almost. The difference is that the candidate is only allowed 2 levels of support within their own party, and 1 level per party for other parties. So A1's preferences (candidate 1 from party A) could be something like {A3, A5}>{A2, A4}>{B1, B2,...,B5}>{C...}>....

When several surviving candidates are tied for first on a ballot, that ballot is divided evenly between those candidates.

This is intended to be a compromise between ideological/factional expressiveness and excessive wheeling and dealing within parties. Coherent factions (such as, in today's Democrats, the Progressive Caucus vs. Blue Dogs) would have the power to keep votes within the factions as much as possible, but it would be hard for somebody to get elected based on pure wheeling and dealing, without direct support from the voters. I think unrestricted STV as you describe would have too much chance for the latter.

[Frank]

That's what I meant: single Winner, what You call "single Member", yes. (It was late here when I typed up that message.) Now, I am certainly intrigued about PR, especially variants which give local representation because, if nothing else, local representation can give Some a sense of "My Representative, Someone I am more likely to run into on the street and be able to interact with Them and hold Them accountable". (Yes, I know in a sufficiently large district actualizing that interaction is unlikely; I just feel such a feature may help to assuage the concerns of Those not wanting to deal with a Representative too geographically distant from Constituents.)

As far as 1/6 versus 1/2, I keep remembering something I heard during the post-election litigation in 2000: democracy is an approximation at best. However, all else being equal, having "only" 1/6 unrepresented would certainly be preferable to having 1/2 unrepresented. Yet, the prospect of "filtering out the fringe" until They can articulate a persuasive argument also strikes Me as a moderating influence.

I guess I just need to read up some more on the idea of PR and see what methods exist for achieving its benefits (increased accuracy in reflecting the Voters) while retaining the perceived advantages (i.e., locality, moderation) of single Winner. [granted, if Anyone here can point Me in the direction of such, cutting down My research time, I would be most appreciative.]

[Jameson Quinn]

2014-03-30 15:56 GMT-04:00 Toby Pereira <tdp...@yahoo.co.uk>:

I was just looking at the link to PAL that you gave....

I redid the graphic. 

[Toby Pereira]

OK, thanks. What I found most confusing was section 3. Because the outer bits aren't exactly aligned with the inner bits (not sure if it's intentional), I wasn't sure if they were all part of the same wedge. Especially because the "dark fringes" - indicating that a candidate is elected - are sometimes an entirely different colour - e.g. white on top of green. But I understand what it means now. I'm still not convinced it's the most easy-to-understand way of representing what you're saying though. Like Warren, I'm not a massive fan of the rainbows anyway, but I think they'd be more workable if you did away with mini-rainbows. I know it would take up more space, but I think each rainbow would be better off the same size as the final one. That way, instead of having a strange different colour on the outside of the wedge indicating that a candidate has been elected, you could simply write "elected" in the wedge. Or maybe if you're sticking with the mini-rainbows, you don't need each candidate to have a unique "elected" colour. All the outer fringes could be black and I think it would be clearer. Then you could just write "a black outer fringe indicates a candidate has been elected."

 

So on the STV aspect, a candidate can have two levels of ranking inside their own party, and just one for each other party, and also a separate rank for each independent - is that right? And does a voter who chooses not to delegate just have one level of approval available to them? One possible problem I have is that people who vote for independents might be at a disadvantage. I might support a certain candidate, and while in general I might trust their judgement on other candidates (as I agree with them politically), they might not be particularly aware of all the other independent candidates that I support, especially if they are dotted around the country and not that well known generally.

 

A different system that I thought of is that you have an approval ballot that lists all your local candidates and also has tick boxes for "approve all candidates from [party name]". You can also write in the name of any candidates nationally, or use a pre-prepared printed ballot paper that lists every candidate you approve (which you can make online), which you take into the polling station and get endorsed. It's not an asset system so slightly off-topic, but how do you think that would compare with PAL? It would still be very simple for voters who don't pick non-local candidates, although they do risk wasting their vote if they pick one candidate who then doesn't get elected. But that's why I would add the box for all candidates from each party.

 

[William Waugh]

In the thread starting at https://groups.google.com/forum/?fromgroups=#!topic/electionscience/FKx0A4ma-Uo

Jameson Quinn says "There is no true PR system [that] the average voter would understand."

I beg to differ.  The PR system proposed by Jim Mueller seems to me very easy for the average voter to understand.  Those candidates with sufficient count of signatures can get on the ballot.  You vote for your favorite candidate from among those.  That candidate takes office and represents you.  Each member of the legislature wields a weighted vote according to the count of voters she represents (i. e., those who voted for her).  Is this not simple?  And as for PR, it seems to me that it provides a finer grain of PR than any "voting system" that elects equal-weighted legislators.

William Waugh