[EM] IRV / RCv advances

Abd ul-Rahman Lomax abd at lomaxdesign.com
Thu Jul 19 09:39:50 PDT 2018

Great question, thanks.

On 7/19/2018 10:17 AM, Kristofer Munsterhjelm wrote:
> One thing I don't get about this phrasing of Asset is how the problem 
> of scale is handled.
> If the number of people initially voted for is small enough that they 
> can all physically meet, then Asset is easy: everybody who gets at 
> least one vote meets and then negotiate to distribute vote assets 
> until the vote assets have been concentrated on sufficiently few 
> candidates to fit the final council. If the council votes are 
> unweighted, one might additionally add a constraint that every 
> candidate needs more than a quota's worth of votes.
Actually, what I've been proposing is far simpler. A quota is set, and 
once a candidate is chosen by that number of electors, the candidate is 
elected. It is not essential that the election process be complete by 
some deadline. Some seats might remain vacant, and under some conditions 
and rules, some seats might not be filled until the next election. An 
Assembly can -- and often does -- function with fewer than the full 
number of seats. However, two points should be considered: the Assembly, 
like traditional assemblies, makes its own rules and it can create 
observer seats with limited voting power. I prefer to not make that part 
of the voting system. The system creates seats and there might be some 
unassigned votes, I call "dregs." Ideally, these are considered, but 
they are controlled by the owners of the unassigned votes.

The other consideration is that an elector may delegete votes to a 
candidate who is already elected, and in delegating such votes, the 
elector is consenting to assignment of those votes by the selected 
candidate. Or the elector may hold back the votes until an effective 
candidate is found, to be chosen directly. I prefer in this system that 
votes be transfered in named blocks (named after the original elector 
who received them). Thus every vote becomes traceable to the elector, 
and thus the actual secret ballot voter knows whom their vote elected. 
This is quite simple to do. Fractional votes might also be part of the 
dregs. Again, I'd suggest that those transferring votes maintain the 
identify of all voting blocks. Simple to do with computers, and it 
creates a traceable and verifiable path from voters to seats. (As to the 
original votes, those would be identified by precinct and, again, there 
are quite good reasons to suggest that this identification remain.)

While Asset is not "district-based," as I've described it, in practice 
people will mostly prefer to select someone local to represent them. 
Asset allows them to do that, if they can agree locally with other 
electors, while some may prefer to choose some special representative 
for their point of view or identity for the entire jurisdiction.

> But now suppose there are a million voters and everybody chooses one 
> of their friends as the one they trust.

Which is quite what I'd suggest. I.e., I suggest that initial voting be 
for someone who is not only considered qualified, but with whom on may 
actually, in real life, communicate.

> The set of everybody who's somebody's friend is unlikely to be small 
> enough that they can physically meet.

It is not necessary for everyone to meet. Indeed, meeting may not be 
particularly useful until amalgamation is at a very high level.

So, a million voters. Not stated, how many seats? Let's set the quota at 
20,000. So this is 50 seats maximum. There may be electors with only one 
vote. My sense is that most of "small electors" will simply assign that 
vote to their favorite and leave it at that.
> How does Asset proceed from this point? Is it just like delegable 
> proxy, where the "initial candidates" name other candidates in turn 
> until the set has been narrowed down enough?
I have suggested that electors may use delegable proxy to rapidly 
identify efficient transfers. But that would be voluntary, not official. 
Officially, an elector logs into a web site and records vote transfers. 
These are all public and traceable. The actual election is public, but 
only a very simple choice is being exercised by each elector, it is not 
necessary to meet in person, because deliberative process is not required.

The exact details may obviously vary. But let's imagine that votes have 
assigned and amalgamated to a level where half the seats plus one have 
been assigned. If the vote required to pass a substantive measure is 
based on a majority of the total possible seats (thus actually 
representing a majority of the total body of voters), then the Assembly 
may begin functioning, but requiring unanimity for action. As more seats 
are elected, that would decline. If all seats are elected, or if 
fractional voting is allowed (a tweak), the standard to meet for passage 
would be a simple majority. Motions of privilege and the like would be 
based on actual elected and present seats.

So we have 25 seats left, and 500,000 voters not yet represented. 
Whenever 20,000 voters agree, another seat is elected. How many electors 
are involved. If the average number of votes per elector by this time is 
200, groups of congenial electors could meet, and it would be about 100 
electors, very doable. They do not need to agree with each other, in 
toto, only in blocks enjoying, within the block, unanimity of choice. 
There might be a big convention, toward the end.

By the way, who pays for the expenses of electors? My suggestion is that 
elector be a fully volunteer position, and that electors pay their own 
expenses. However, people may donate to a fund for elector expenses, 
assigning this to electors. If spent for legitimate expenses, this would 
be tax-free. Anything beyond that would be taxable income, and some 
electors might make it a full-time job, similar to present lobbyists. 
They would serve to communicate between seats and the general public, 
and especially those who voted for them.

Asset Assemblies, I predict, will self-organize with much more 
efficiency than we might expect.

However, this is one reason why I suggest NGO implementation of Asset 
before using it for governmental purposes. That's to develop experience 
in how it works -- and how it might not work.

More information about the Election-Methods mailing list