November 05, 2002

IRV Isn't All That Great

Here's a good explanation of how IRV is just screwy:

Say we have two extremist candidates and a centrist. For instance, say we have Pat Buchanan, Colin Powell, and Ralph Nader.

Now, say people rank them all:

  • 5 million people vote only for Powell.
  • 8 million left-wing people rank Nader first, Powell second.
  • 7 million right-wing people rank Buchanan first, Powell second.

What's common sense? First, could a situation even vaguely similar to this happen? I think so. It's not a ridiculous scenario. Second, just stare at that for a second and tell me who you think should win.

Ok, now look at it this way. 13 million people prefer Powell to Buchanan. 12 million people prefer Powell to Nader. Only 8 million people voted for Nader. Only 7 million people voted for Buchanan. (This is how Condorcet voting looks at things.) Shouldn't Powell win?

If a voting system asks voters to rank candidates, isn't the implication that those rankings should be paid attention to?

IRV voting would eliminate Powell first, since he has the least first-place votes. Nader would win. Posted by Curt at November 5, 2002 02:46 AM