User:Randomstaplers/Article 1

Originally titled Let's Simulate IRV!

I've got a feeling that those pushing IRV (also known as RCV, or the CGP Grey 'alternative vote') at any cost, 'cause we need voting reform now!' have never actually seen IRV simulated, outside of charts provided by Grey. Because if one were to see visual aid on who wins and loses, one might be reluctant to adopt IRV.

Votesim

See also: Running IEVS on your computer!

The simplest computer simulation, as far as I know, only requires grade school geometry to understand. It was invented by Ka-Ping Yee in 2005, no relation to any voter advocacy group... so far. You can see his results here. Votesim was soon added to Warren Smith's IEVS.

How does it work? Well, Mark Frohnmayer rewrote Votesim again, this time by adding animations. His video pretty clearly describes Votesim's operation.

A few haphazard votesim results

But first: A word on models

See also: Psychology and Voter UI

All models break down in some way, but that doesn't mean their results aren't useful. The world, for instance, doesn't look anything like what is shown on mapping websites, but it doesn't stop us from using them to navigate the world. This model assumes everyone is honest, so Borda looks a lot like score, because it cannot differentiate the difference between ranked ballots and cardinal ballots, a UI problem which can surface when real people vote.

It is also biased towards Llull winners (which Condorcet rediscovered in the 1700s, who knew?)

The threshold for approval voting does not correspond with polling.

Results?

16 Candidates

With a standard deviation of 50.

• This is the ideal model. Condorcet/Llull, and one voter are identical to this.

• First-past-the-post.

• Approval, but see what I said above on models first.

• Range/Score.

---

Take notes on these two voting methods. We'll come back to them later.

• Top-2-Runoff, essentially FPTP done twice.

• IRV.

5 Candidates

With a standard deviation of 100.

• This is the ideal model.

• Top-2-Runoff.

• IRV.

So why is IRV controversial?

Opinions ahead!

• There are situations where IRV produces worse results than Top-2. And in the multi-candidate simulation, IRV doesn't produce significantly better results than Top-2. What's the point in advocating for IRV in a Top-2 state, when there are better voting systems?

• IRV is absolutist. You support candidate 1, and only 1, until their elimination. Then support for candidate 2, etc. If a compromise candidate is eliminated, well...

Let's say [Red], [Yellow], and [Green] are running in a ranked election.

[Red] does well, but doesn't reach a majority, and causes [Yellow] to lose...

[Yellow] is evenly split. (sic, but you can't determine how the vote will go)

So Green wins instead. Even though polling shows you would probably prefer Yellow, as the optimum candidate for society.

• Laboriously transferring votes when a candidate is eliminated? No easy way to count in precincts? Less transparency is the last thing we need when people don't even trust the simple FPTP system.

• IRV has been repealed before. The fact some of us are still advocating for it is insanity. (Insert your own word here, I got tired...) Oh, and if the big parties catch on to IRV's flaws, they might use the voting method as an excuse to never reform voting again!

At the end of the day...

I hope we take a more cautious approach to voting reform and not just jump on the IRV bandwagon.

There are better voting methods, some of which don't produce worse results than T2R. Let's not shoot ourselves in the foot here, advocating for any voting system that could do worse than doing FPTP twice.

See Also

• Running votesim-based programs
• Psychology and Voter UI