I'm sitting across a table from Lynne of Left in Lowell, listening to results being announced district by district. About half of the 40 districts have been tallied so far. Lynne is liveblogging the results.
What you're claiming makes no sense. Let's say that Kerry has exactly 50% on the first round, as in your example. As weaker candidates are eliminated and those ballots are reassigned to other candidates, the overall pool of ballots does not increase. Kerry can't lose any votes, since he's the strongest candidate. It's possible that some ballots are exhausted (that is, they have no more ranked choices), in which case the pool shrinks, and Kerry's 50% becomes >50% even if he gets no more ballots, and he wins. If no ballots are exhausted, rounds will continue until at least one new ballot gets reassigned to Kerry, at which point he wins. But there's no way to push him below his percentage on the first round, so he can't lose. It's simply impossible.
Let's modify this a little bit, for a more illustrative example: Kerry gets 50%-1 votes on the first round. Now it is mathematically possible for Bush to win in a subsequent round ... but only if all other ballots vote for him, and none of them rank Kerry higher. So the presence of ballots with Nader #1, Kerry #2, can only help Kerry, not hurt him.
The way IRV plays out in practice matches common sense, so the explanation to voters is simple. You seem to have looked at mathematical analyses that misrepresented what IRV is, and have twisted your perception of how it works to something much more complex, but also incorrect.
There is abolutely no way that moving some votes from category 2 to category 3 would help Bush win. It's ABSOLUTELY IMPOSSIBLE.
Add some other candidates and other votes and whatever else you want, but the simple fact is, there is no practical way that someone's vote would help Bush win if their ballot meets these two simple conditions:
They do vote for Kerry somewhere on the ballot (not necessarily #1)
they don't vote for Bush
It's very simple.
I've seen a lot of crackpot critiques of IRV, but all of them pretend that IRV requires you to to "rank all the candidates in order of preference," which is ridiculous BS invented by people who wanted a simple way to analyze it mathematically to compare to other voting systems, without paying attention to reality.
If you "rank all the candidates in order" that means you're voting for candidates you oppose! The solution to any such problems is simple: don't vote for candidates you oppose!. Period.
We have IRV as part of a proportional representation system here in Cambridge, for city council elections. We had 18 candidates running last year. A majority of voters voted for fewer than 10 of them. Fewer than 1 in 10 voters actually "ranked all the candidates". Voters for the most part understand that you shouldn't vote for the candidates you don't want to vote for.
Look at the number of contributors. There were actually only threereal house parties, nationwide. Pretty much all the rest were single contributors (except for one who got a second contribution for a total of $100, and two others who got second contributions but raised $50 or less in total).
Let's take your hypothetical Kerry/Bush/Nader election, with Nader supporters backing Kerry over Bush. If Bush gets >50% first round votes, that means there weren't enough Kerry+Nader voters to beat him. It doesn't matter how many of them vote for Nader #1 vs. Kerry #1. If, on the other hand, Bush gets <50% first round votes, Nader is eliminated, and all the Nader votes switch to Kerry.
In other words, Nader voters who prefer Kerry over Bush can't possibly swing the election to Bush by failing to give Kerry their #1 votes. If Bush has enough to win in round one, they're irrelevant. If Bush doesn't have enough to win in round one, their #2 votes will be counted.
On election day 2004, I was at the Election Protection call center in Broward County, Florida, in charge of coordinating response to reports of voting machine incidents. Anyone who called 866-OUR-VOTE in Broward and reported a voting machine problem was directled to me unless I was too busy with other calls.
In the past few years I've also been a poll-watcher for progressive campaigns in several special elections, and/or been a precinct captain training and coordinating poll watchers, and I've participated in a recount for a progressive write-in candidate for Democratic State Committee (we gained 3 votes in the recount, putting our candidate in an exact tie with her opponent - though the Boston part of the district did not count write-in votes and we know our candidate got some there, while her opponent probably didn't; the opponent won the coin toss).
... and yet, I still think we need more national media attention.
Now, I'm not pushing the "2004 was stolen" idea as a certainty. I think it's possible that Kerry got more votes in Ohio, but not very likely. I think it's certain that urban voting was suppressed, and that there was no accurate count of the Ohio vote, and that this is a big problem even if Bush really did win that state. (I left out of the things I did last year: standing at a bus stop in downtown Cincinnati, registering mostly poor people to vote)
MoveOn recently sent all its members a ballot to vote for their top three priorities for the next year. "Elections that can't be bought" and "Accurate elections" are two of the items on the list. Please, everyone, vote for both of those!
As for me, this year I'm working for John Bonifaz for secretary of state of Massachusetts. Wherever you are in the country, this race should matter to you. Here's someone who's been fighting the election reform fight, down in the trenches, more than any of us, someone who has shown real national leadership. Put him in charge of elections in an influential state, and election reform will get a major push nationally. National leadership is in his platform, the Voters Bill of Rights.
The more evenly you split the territory up into districts, the more you force candidates to really have a national agenda that takes every portion of the population into consideration. You get that by having districts that put together various different combinations of people, that are small enough that each group is likely to count in a number of districts, and large enough that each district has multiple groups. Each district then becomes a different combination that candidates must appeal to. If you do this well, candidates have to appeal to a very broad set of districts.
If you go too far to either extreme - districts too large, or too small - candidates can run more polarized campaigns, focusing on fewer groups. That would work well for a proportional representation parliamentary system, but it sucks in a first past the post system, especially one with a seperate executive branch where we're trying to elect on executive.
Districts also have the advantage of balancing turnout and giving each area it's correct proportional strength. Candidates can choose where to campaign based on the assumption that each district's voting strength is equal (if districts are equal - in the current system, they're not, but at least we know their voting strengths, and it should be proportional to population). This is based on the idea that there's a community of interest based on geography, and I sense you're challenging that idea, but that's definitely something I'm willing to support. Despite all our mobility, there is definitely a strong community of interest based on geography, and it has aspects that make it better for democracy than other kinds of communities of interest.
That's exactly my point. It is much more efficient for a candidate to target a million-person city than it is to target more than ten 100,000-person cities. It's not just canvassing, it's also media, community organizations, everything.
It's even worse when you're talking 6 or 7 major cities - those cost a campaign the same resources as it would take to reach hundreds of smaller ones as effectively.
In other words, exactly what you said.
You don't think that's a problem?
I think it helps us to force candidates to spread their appeal broadly, and districts help do that. To the extent that we have districts now (the states), we get it partially - you can't make up for fewer votes in Missouri by getting more votes in New York, for example. But within states, we have the same issue, and as I said, states are mostly too big to be good districts.
I'm using cities as an obvious example, but the same problems apply to all sorts of population clusters - ethnic, ideological, differences in wealth, age, class, etc. The more you split the territory up into districts, the less candidates can pick and choose the groups they want and ignore everyone else.
It's really uncanny just how many news articles and blog posts there are about this race that don't actually say what the date of the primary is. I have yet to find one that does. I'm talking everything from local blogs, to USA Today - they all talk about the primary, they just don't bother to say when it is. Some must assume all their readers know, some probably assume their readers don't care, but it's so universal, it's creepy.
Add to the strangeness that neither Tester nor Morrison's web sites say what the primary date is, in any obvious place. With the low turnout in primaries, you'd think they'd want to keep reminding their supporters (who visit their sites) of the primary date. But no.
The Maine/Nebraska system is not bad. It does suffer from the problem that Congressional districts can be gerrymandered, which I pointed out in my comment near the top here. But despite that, using CDs as electoral districts may still be better than using states. Or about the same: NPR reported a survey of the numbers per CD in 2000 showed that if every state used the ME/NE system, the electoral result would have been... exactly the same as it actually was.
Proportional allocation of electoral votes, on the other hand, is a disastrous idea, and I'm glad Colorado rejected it.
It has two major problems.
1. There's no clear way to allocate votes among multiple candidates. You can come up with a formula that more or less makes sense for two candidates, but even that gets hairy when you have a small number of EVs to allocate. What about three candidates competing in a state that has 7 EVs? Or 6 EVs? What formula can you come up with that really makes sense to voters?
2. It drastically underrepresents large states. Let's compare Montana, worth 3 EVs, with Wisconsin, worth 10 EVs, for example. Currently, candidates pay a lot of attention to Wisconsin because if they win, they gain 10 EVs. But if Wisconsin allocated proportionally, the difference between getting 48% or 52% there would likely only be worth a swing of 1 EV. Effectively, Wisconsin would now be worth 1 EV, or 1/3 as much as Montana. If Wisconsin passed such a measure, candidates wouldn't bother to campaign there.
For states currently considered "safe", proportional (if it could be made to work sensibly, see point #1) would indeed bring them from being worth 0 EVs (because the result is predetermined) to being worth 1, 2, or 3 EVs. But in trade for that small gain, those states would permanently be small states, even if different candidates or demographic changes or political swings turned them into swing states. They'd still only be worth 1-3 EVs. California might be worth a few more, but certainly nothing close to in proportion with its population.
Did you read my original comment? I'm not defending the current system, I'm pointing out its problems and suggesting what direction the solutions should go in. And I'm suggesting that a flat popular vote is the wrong direction, and would make things worse, not better. So just repeating what the problems are with the current system, is no rebuttal to what I said.
Or to put it another way...
Current system: Has big flaws.
Flat popular vote: Even worse.
I decry going to flat popular vote, and suggest another direction to go. You answer me by pointing out that the current system has big flaws. Yes, we knew that already.
That's an illustration of one of the problems with the current system: States make for bad districts. Particularly, many of them are much too large. The larger your districts, the higher the chance that a lot of the population will live in "safe" districts rather than swing districts.
People see this problem and their knee-jerk reaction is "abolish the system! let's have a national popular vote". They don't realize that could make the problem worse! The solution to having many of our districts be much too large, isn't to make them larger by collecting them all into one. If we do that we may find to our dismay that in some years, the whole country is a "safe" district, and nobody's vote seems to matter. Even if it's not safe, the margins of victory are so huge that people will often feel their vote doesn't matter, anyway.
No, the solution to this problem is to make smaller districts. If we can find a mechanism to do this that avoids gerrymandering, few of these districts will be stably "safe" from election to election, and we'll have a majority of the population living in swing districts. Further, the swing and safe districts would be distributed all around the country, meaning that even if you live in a safe one, there's likely one next door that isn't safe, so you'll still get candidates visiting your region, and you'll still be able to get involved locally to help win the election.
It is better for a politician to cater to a million person city than a 100,000 town because the million people are as american as the 100,000
But why is it better for a candidate to cater to a million person city than to more than ten 100,000-person towns? Because that's what happens when you have no districts. And that's what happens within many states now - which would be solved if they were split up into more districts.
I don't think you really understood what I was saying, or the reasoning behind it. It's true that I didn't write a full essay on why we should have districted voting instead of a flat nationwide vote, but I did include things you seem to have missed. So, point by point...
1. Low turnout in safe states: You're right. As I said, states make bad districts. The larger they get, the more of a danger that lots of people will feel their vote doesn't count because the outcome is not in question. It's great for the swing states and bad for the safe ones. The solution I recommended is smaller, more uniform districts. As I said, if we divided California into a bunch of smaller districts, a lot of them would be swing districts.
So I didn't ignore this problem, and your counterargument of simply restating it doesn't make sense. If you want to argue that my proposed solution is not a good one, say so directly.
I'll say why your proposed solution isn't a good one: If you make the whole country into one district, you run the danger of making the whole entire thing "safe" and giving everyone the impression that their vote doesn't count! People are a lot more likely to vote in smaller districts where margins of victory are smaller, and their vote seems to matter more.
That aside though, my main point #1 is that you're simply restating a problem I already acknowledged, and tried to solve. Saying "this is a problem" a second time, is just as likely an argument for what I'm saying, as it is against.
2. Less incentive to cheat
I don't buy it at all. If we have a flat popular vote, of course nobody will focus on "delivering" an individual state. But if there's cheating to be done now, there will be cheating to be done then, to deliver more votes. The stakes would be even higher in a close election: Cheating that might, in the current system, only tip a single state, might tip the whole thing. Or stakes might be lower, but that's only if the election isn't very close, in which case stakes are low in the current system as well. Overall, this argument is a wash - incentive to cheat doesn't change.
3. Rural already gets ignored. Not really. Mostly this is an effect of states being bad districts. But, for example, look at New Hampshire, or Maine, which are mostly rural. In a flat popular vote system, they wouldn't matter much.
The examples you cite actually bolster my argument and undercut yours. Within a single state, which has both urban and rural areas, if a candidate can win by concentrating on urban + suburban votes, they ignore the rural areas. If you made the entire country one district, this dynamic would play out nationally, since votes from Boston and Cincinnati are just as good as votes from the middle of Kansas.
On the other hand, if you split things up into districts and require candidates to win broadly, they have to pay attention to a lot more places: look at the Iowa caucuses as an example. A candidate has to be viable in most of Iowa's 99 counties to win the caucuses. You can't use loads of extra votes in Des Moines and Cedar Rapids to make up for 0% in the rest of the state, the way you could if a statewide vote were held. The clearest counterexample is Oregon, where statewide elections are usually Portland+Salem+Eugene vs. the rest of the state. Democrats need only campaign in those three cities, and if their turnout is better, they win.
It's not just urban vs. rural. Any small group risks being ignored at candidates' whim - ethnic minorities, social classes, etc. Basically, any group a candidate thinks isn't likely to support him/her, isn't worth the attention, because you can always make up for them with more votes from a more favorable group. The smaller the districts, the more likely each group is to have one or more districts where they really matter. As I keep saying, states are far from the ideal districts, and most are too large, but they're a heck of a lot better than the whole country as one district.
4. Weather patterns, etc. already affect turnout, yes. But they don't affect how much each district counts. Putting aside the other major flaw I listed in the current system (the overrepresentation of small states due to Senatorial electors), each district counts in proportion to its population. If Florida's having a hurricane while Oregon has mild weather, the lower turnout in Florida isn't going to make it count any less. In a flat national vote, Oregon's voting power would increase relative to Florida due that accident of weather. Just as one example.
Again, states make for bad districts. We'd do a lot better with smaller, uniform districts. My contention isn't that keeping the system as it is is the best thing to do. My contention is that turning the whole country into one district is the wrong direction to change things, and will make everything worse.