[CharlesColeman]

> Approval voting does way better, and it finally has been put into practice in local elections (in Fargo ND), and is being considered for a big city (St. Louis MO).

That looks a lot more understandable and straightforward than ranked-choice voting, which is a big plus to me. Vote counting systems should be so straightforward that even a simpleton can understand them, as the more complicated they are the less I think they'll be trusted.

Even with such a system, I think I'd still like there to be a final runoff election for important offices. It gives an opportunity for correction if a crazy somehow unexpectedly comes out ahead, and it will give a majority endorsement to someone.

[robbrown451]

The proposition in St. Louis does have a runoff. Seems like a lot of expense.

One system I've been looking at recently is "For and Against", which is also easy to understand. You get to vote for one candidate and against one. Against votes count as a -1, although in theory you could "phase the system in" by having the against votes count as -.5 or the like, the first election with the new system. That gives parties and other structures a bit more time to adapt.

For and against is interesting in that it simply cancels out the vote splitting effect of plurality. You no longer have an incentive to hold nominations to eliminate candidates, since similar candidates will split against votes.

[CharlesColeman]

> The proposition in St. Louis does have a runoff. Seems like a lot of expense.

Honestly, I think there's too much focus on expense nowadays. We shouldn't feel forced to pursue the best cheapest system, or feel limited only to systems that are no more costly than the ones we have today.

If an idea results in a more trusted and effective electoral system, I think it would be worth it even if it costs significantly more.

[ClayShentrup]

For and against still gives you the vote splitting problem. Approval voting is much superior.

[robbrown451]

Hi Clay,

I'm not specifically advocating for For and Against (at least not compared to Approval), but I think it has interesting properties that can be used effectively for explaining vote splitting and Duverger's Law to those who don't quite grasp it. (I also think the fact that it could theoretically be phased in gradually is very interesting, particularly to those that worry about politicians who were elected with plurality being reluctant to endorse a different method)

To the best I can test it, though, (I've been working on a graphically based voting simulator), For and Against comes up with remarkably good results compared to Approval, sometimes better, sometimes worse. Let me know if you want to see the graphical voting simulator, I'd be interested in your ideas about the assumptions made (there are lots of ways to tune it). I guess the way I'd put it is "yes it still has vote splitting, but that vote splitting doesn't result in Duverger's Law")

Interestingly, For and Against is less expressive than Approval if there are five or more candidates, but more expressive if there are only 3. (e.g. you can say "Nader>Gore>Bush", rather than just "Nader=Gore>Bush" or "Nader>Gore=Bush")

But again, I think the main value of For and Against is using it to explain how vote splitting results in a duopoly, ultimately opening people's minds to Approval or for that matter STAR (which I see you are the inventor of?).

If you weight it toward "against" (e.g. count the against votes as -1 but the for votes as only 0.1), it has the exact opposite effect as plurality, in the sense that parties would want to have many candidates running, to dilute the against votes directed at their candidates. If you count them equally, though, the vote splitting effects effectively cancel each other out. The tunability of the method, as well as the way its strategy sort of directly derives out of Plurality strategy, has a lot of pedagogical value to me. Approval addresses vote splitting and Duverger's Law, but I personally find it harder to clearly explain how it actually accomplishes that.