[EM] The Equal Vote Coalition and robla

[Kristofer Munsterhjelm]

On 2025-04-19 04:32, Rob Lanphier via Election-Methods wrote:
> The more I think about STAR vs strictly CWC-compliant systems, and the 
> more I try to get ChatGPT and other LLMs to make the best case they can 
> for you, I can't get a convincing case that the distinction is a reason 
> to divide into warring factions over.

As a Condorcetist, here's my take on this:

Approval is the best bang for the buck method due to its simplicity. 
STAR is slightly more complex but mitigates some of Approval's problems, 
at least unless people start cloning candidates left and right.

The more important simplicity is, the better Approval and STAR are 
compared to proper Condorcet. The more important risk-aversion and 
theoretical guarantees are, the better the latter is compared to the former.

Not being an American, I don't know the threshold after which people get 
confused because the method is too complex. But I can imagine a society 
where anything more than summing values is seen as too weird; in such a 
society, it makes complete sense to support Approval because you're not 
going to get Condorcet no matter what. And I can imagine a society 
that's comfortable with Meek STV; in such a society, you can go right to 
Ranked Pairs, no problem. These are extremes, but the "is it too 
complex?" position seems to be the most clear division between Condorcet 
and Approval.

The obvious risk for Approval is the Burr dilemma. Like center squeeze 
in IRV, it doesn't have to happen often. If it leads to clearly unfair 
results (like Burlington), then people could easily go for a repeal.

STAR is a good compromise in the sense that it's unlikely that you'll 
get a Burr dilemma, at least without cloning. But it's vulnerable to 
cloning and is very much a pragmatist method, not passing many criteria 
and still having the formal "what should I rate my second favorite" problem.

Proper Condorcet methods mitigate or solve these problems. In addition, 
a rank-order ballot is unambiguous for honest voters. However, the 
methods are more complex.

I would probably say: as long as the activists don't step on each 
other's toes, do whatever. I personally think that Approval is still a 
little too risky, but STAR and approval plus runoff (St. Louis style) 
should be safe enough. Even if their practical performance might be 
reduced due to clones, it shouldn't outright blow up in people's faces.

> p.p.p.s. Some of you may have noticed I've turned on "Reply-all munging" 
> for the mailing list.  I'm pretty sure the increased use of DKIM is the 
> reason why I've missed out on many messages from this list.  Munging 
> helps with DKIM compliance and thus should make delivery more robust, 
> but it also means simple replies are more likely to go to the mailing 
> list rather than just the sender, even if you mean to just send a 
> private reply to the sender.  Be careful out there!

It does mess a bit with my mail reader. If you have the time, could you 
check if it's possible to use ARC? 
https://proton.me/blog/what-is-authenticated-received-chain-arc
It should make munging unnecessary.

-km