Our group uses Systemic Consensus: Negative Score Voting w. control options and consensus logic.

[Doug w]

Greetings from Germany! I'm contributing with yunity.org, a project aiming to release a multi-sharing and saving platform. Adjectives describing this wonderful group of humans are not limited to: distributed, pluralistic, epicurist, practically moneyless, nomadic, unincorporated and anarchic. That might sound a bit chaotic. It is.

Without elected delegates or a 'judiciary' function (read: enforcement) and with 'legislative' and 'executive' functions evenly shared by everyone, people have to want to do things. What about when contributors disagree? This is the type of gig that you might expect to go down the (unanimous) consensus/voting-with-veto/leave-no-one-behind path or the autonomous/advice process/I'll-do-what-I-want-I'm-an-anarchist path. However there is exploration between the collectivist and individualist extremes. For the past 6 months we've been trying out Systemic Consensus (originally Austrian [lang:de] 'Systemisches Konsensieren'). In brief,

Ballots accept score votes (cardinal ratings) and always have two control options to make sure everyone has an option they can accept, these are:

• 'The passive solution' - maintain the status quo (details case dependent)
• 'Further solutions/repeat cycle' - essentially 'None of the above'. The vote is rescheduled for some future time and there is a return to forming more proposals.

Score voting proceeds in a negative fashion - voters score options from 'accept' (0) to 'fully resist' (n, usually 10 in our case). The option with least votes (the least resisted/most accepted option) is selected. In the case of a tie, approval voting or normal score voting can be used.

The rationale for negative voting is that making decisions people aren't happy with is bad because these people,

• may not help → frustration/burn-out
• may actively resist → conflict
• may potentially leave the group → fragmentation

... and it is thus most important to firstly assess how much people don't want to do something, before assessing how much people want to do something.

OK, so there you have it! Now I have been critically thinking about this system for some time with how/what could be made better. However, before I launch into my conclusions and flavor the conversation, I'd be interested to get feedback from people outside this bubble. Thought? Comments? Concerns? Questions? Fire them at me.

Cheers,
Doug

(p.s. I note that this framework was mentioned briefly on this mailing list some years ago. You can take a look at one of the project results here and our most current documentation on the subject here (lang:en))

[William Waugh]

If the ranges of two score voting systems can be mapped to each other as y=mx+b, those systems are strategically equivalent. Your choice of putting the top of the scale at nominal zero may have a psychological effect on some voters, but it does not change the nature of the voting system itself viewed as a game.

On Saturday, October 29, 2016 at 6:50:53 PM UTC-4, Doug w wrote https://groups.google.com/d/msg/electionscience/AzYFf2qEFw8/Wrajn8pIAgAJ

[Doug w]

Hi William,

Thank you for putting that in succinct words, my maths is good enough to describe things so easily. The psychological effect presumably plays a significant factor in our group where there are social connections between most of those involved. Perhaps you are right, that on a large enough scale where people don't know each other and are thus more likely to vote strategically/dishonestly that this 'game' is identical to positive score voting with the same strategy of 'semi-honest exaggeration'.

Thanks for feedback!

[Jameson Quinn]

You're describing a situation where the ideal is consensus. In that kind of situation, I'd definitely want some kind of veto threshold. For instance: if any option besides status quo gets the minimum vote (-10) from over 1/3 of voters, it cannot be passed; and score voting applies only among options not thus eliminated. This would allow more-honest voting, because one could vote truly unacceptable options down without having to strategically vote everything else up.

Why do I pick 1/3? It's a compromise between the desire for consensus/supermajority and the paralysis that can result. In my experience with consensus organizations, it's dangerous to put this number too low (paralysis) or too high (lack of buy-in). I'd say 10% to 50% is about the reasonable range, and so 1/3 is about the middle of that range.

--
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 electionscience+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Doug w]

Hi Jameson,

I'm really not keen on veto at all! I find it to be a dominative and polarizing motif. An approach I think is morepromising that I've been thinking about a lot these past hours is of making Support distinct from Resistance in the vote schema. I think this schema concept has similarities to your PAR schema. For example,

| Score  | Acceptance = 1-Resistance   | Support   |
| --     | --                          | --        |
| +3     | 1.00                        | 1.00      |
| +2     | 1.00                        | 0.67      |
| +1     | 1.00                        | 0.33      |
| 0      | 1.00                        | 0.00      |
| -1     | 0.67                        | 0.00      |
| -2     | 0.33                        | 0.00      |
| -3     | 0.00                        | 0.00      |

(^ That might render horribly...)

Some selection criteria that come to mind that favor acceptance over support without unduly incentivising manipulation are:

• Discard options with Acceptance < Acceptance(max-x), choose remaining option with most support. (where x is some predefined allowance value)
• Select option with greatest Acceptance + Support/y (where y is some predefined weighting factor, decreasing the value of Support)
  

I'm only really concerned with a manipulative minority, because in this organization/similar, people won't do things they strongly disagree with whatever the outcome of a vote, potentially resulting in fragmentation (bad for everyone, manipulators included).

Fire me anything I should read that addresses Social Choice Theory in this field.

Best,
D

To unsubscribe from this group and stop receiving emails from it, send an email to electionscien...@googlegroups.com.

[William Waugh]

On Thursday, November 3, 2016 at 5:11:48 PM UTC-4, Doug w wrote:

[ https://groups.google.com/d/msg/electionscience/AzYFf2qEFw8/PDrdF0nMAwAJ ]

Some selection criteria that come to mind that favor acceptance over support without unduly incentivising manipulation are:

• Discard options with Acceptance < Acceptance(max-x), choose remaining option with most support. (where x is some predefined allowance value)
• Select option with greatest Acceptance + Support/y (where y is some predefined weighting factor, decreasing the value of Support)

To begin with your second option first, because it is easier to analyse, this is just Score Voting, since multiplication distributes over addition. Fixing on a value for y reduces the table to a set of scores that can be assigned. How they are arrived at becomes irrelevant to the meaning of an election. Here is another example where the available scores to use are not evenly distributed: https://1787regime.wordpress.com/2016/08/12/proposed-abfg-score-voting-system-for-single-winner-elections/

Your first option, on the other hand, is not pure and simple Score Voting, because it has more than one decision point. It seems somewhat more akin to Majority Judgment or modern Bucklin. I don't know how to predict its behavior. It's not immediately obvious to me whether it meets Frohnmayer balance (which you may or may not value for the kind of organization you describe).

[Doug w]

Thanks for the feedback William.

An important assumption that I'm making that perhaps I didn't explicitly allude to is significant player honesty. Specifically that a supermajority of players choose to vote honestly or at least semi-honestly (i.e. not scoring an option they accept as 'reject' and not scoring an option they reject as 'support'). I asked around and people reckoned we could trust this honest supermajority to be ~90 %. OK...
 

To begin with your second option first, because it is easier to analyse, this is just Score Voting, since multiplication distributes over addition. Fixing on a value for y reduces the table to a set of scores that can be assigned. How they are arrived at becomes irrelevant to the meaning of an election. Here is another example where the available scores to use are not evenly distributed: https://1787regime.wordpress.com/2016/08/12/proposed-abfg-score-voting-system-for-single-winner-elections/

The example you linked has symmetric score weighting (A:+50, B:+49, F:-49, G:-50) which is different to what I was trying to describe. Would asymmetric score weighting really make no difference to the game? A thought experiment: 10 players score options A and B using the following scores and score weighting under the instruction "-3 signifies complete rejection, 0 signifies complete acceptance and +3 signifies complete acceptance and full support"

| Score  | Score weight   |
| --     | --             |
| +3     | 1.33           |
| +2     | 1.22           |
| +1     | 1.11           |
| 0      | 1.00           |
| -1     | 0.67           |
| -2     | 0.33           |
| -3     | 0.00           |
 
9 players honestly score A as +3 and B as +1. The final player would honestly score A as +1 and B as +2 but attempts to manipulate the game by scoring A as -3 and B as +3. Total scores: A:12, B:11.33 the manipulation attempt fails, barely. Changing the value of 'y' would change the manipulation potential and inversely, the 'minoritization potential' (the likelihood of selecting an option honestly rejected by a minority).

Your first option, on the other hand, is not pure and simple Score Voting, because it has more than one decision point. It seems somewhat more akin to Majority Judgment or modern Bucklin. I don't know how to predict its behavior. It's not immediately obvious to me whether it meets Frohnmayer balance (which you may or may not value for the kind of organization you describe).

Again, the rationale is to encourage honest voting and select the most acceptable option whilst remaining resistant to a manipulative minority. As such, I believe the Frohnmayer balance (if you mean "my against cancels your for") is still in effect. [Aside: Kudos to Mr Frohnmayer for popularizing the concept, but 'Frohnmayer balance' seems unnecessarily jargonic... come across anything else?]


Cheers.