Amongst economists, mathematicians and psephologists, Arrow’s Impossibility Theorem is famous. In layman’s term it says “there is no such thing as a perfect voting system”.
Specifically, it talks about methods of counting — first-past-the-post vs preferential and so on. If a voting system has to choose from among at least three options from at least two people, no method can satisfy all five of these desirable properties:
- Non-dictatorship: The counting method must allow more than one voter’s opinion to count.
- Universality: The method should account for all preferences in yielding a ranking of preferences.
- Independence of Irrelevant Alternatives: Changes in the ranking of irrelevant options should not effect the ranking of relevant options.
- Monotonicity: Putting an option lower should not increase their chances of winning and vice versa.
- Citizen Sovereignty: Every possible outcome should be reachable by some combination of votes. There must be no outcomes which are unachievable by any combination of votes.
Arrow’s Theorem is the foundation of a lot of work, only a small percentage of which I pretend to understand. The practical upshot is that the voting system chosen is selected by a combination of tradition and reference to outside criteria.
Which leads me to drag a software analogy into view, as I often do these days. Computer Science according to Abelson and Sussman is probably the study of processes — how they are described, how they are performed, how they may be analysed and how they may be performed. In theoretical terms this leads off into a fascinating and arduous borderland betwixt computer scientist and mathematician. In the practical world it leads to the concept of non-functional characteristics, or the “ilities”.
A non-functional characteristic of software is one that doesn’t go directly to its purpose. In accounting software, the ability to correctly perform double-entry operations is a functional characteristic. It is directly connected to the purpose of the software. How many double-entry operations it came perform per second is a non-functional characteristic. It is not directly part of the software’s function, but it can be measured, specified and optimised nonetheless.
Generally these are called “ilities” because that’s the common postfix: scalability, reliability, availability, introspectability, observability, recoverability and so on and so forth.
I would say that a non-functional characteristic affecting Arrow’s list from the outside is “trustability”. If it is possible to demonstrate easy subversion of the count, all other properties are meaningless. The theoretically minded might say this is a subset of the non-dictatorship property, but I think it is a non-functional property from outside the theorem.
I raise this point because the USA never seems to see the end of the matter. Right now argument is brewing about whether Diebold voting machines positively and deliberately affected the New Hapshire primaries in favour of Hillary Clinton; or at the expense of Ron Paul. I will not go into whether I think either is the case. Instead I will point out that nobody can trust the system. There’s no definitive way to distinguish between hypotheses of why the votes were as they were.
Part of the reason is the secret ballot. But part of the reason is that the system is not trustable through mutual inspection. The most useful property of Australian electoral systems is that mutual distrust is harnessed through scrutineering. The next most useful is the non-partisan nature of the AEC and state Electoral Offices. These contribute to the trustability of the system and help to underscore the essential purpose of democracy, which to me is the ability to change Government without resort to coups.
In Australia when there is a poll divergent from the election, we laugh at the inaccuracy of the pollsters. In the USA this leads to fears of tampering. The difference is that we trust the integrity of our voting system enough to mock the pollsters.
One of these days I hope the USA realises that the time has come for root-and-branch electoral reform — or at least consider contracting the AEC. They need independent electoral commissions, a better voting method and multi-party scrutineering. Independent redistributions would help too, but it seems less likely than other reforms. Because if this cycle of distrust continues to worsen, resorting to coup will become a more certain way of changing government. And because if the USA can’t trust its own democracy, how can we?
Yes, the scrutineering seems so bloody obvious, doesn’t it?
I wonder why it is so hard. I don’t mean the fact that the US doesn’t have the independent electoral commission – that will be hard to change. I mean the voting machine drama. Why the fancy machines that they can never get right?
Why not just have an ordinary computer with a printer? You’d vote with a mouse on a ballot “paper” on the screen and get a print out to deposit in the ballot box in the ordinary way.
Same as doing it by pencil except that it would be instantly counted as well. If there was a discrepancy, the paper would take precedent. It could reduce informals to nil, too.
I’ve been wondering this for years. It seems so simple. Am I overlooking something?
Mike Pepperday wrote:
It seems so simple. Am I overlooking something?
Yes. Americans are more democratic than Aussies. On election day we only vote for our MHR and a bunch of senators – and maybe a referendum or two.
On an American election day our cousins can be voting for president, senator, representative, governor, state senator, state representative, mayor, local councillor, judges, sheriff and dogcatcher to name but a few. Oh, and a bewildering number of ‘propositions’ at both the state and county level.
Also polling booths are more poorly staffed in the US than Australia – here voting is compulsory so authorities make it easy as possible to vote. The upshot is that automation is forced on most polling booths. [I say 'most' because elections are run at a local or state level so terms and conditions vary.] And while the Australian method of counting is more transparent if you talk to Americans many believe their process is less ‘contaminated’ than ours seeing it’s untouched by human hands.
There seems to be massive inertia in the US electoral system since the end of the Cold War. These days Americans are more likely to emphasise that they are a republic and not a democracy. I can’t see a reform of the magnitude of their 1965 Voting Act for a long time to come. And with any proposed change we must ask the question _Cuo bono?_ “Who benefits?” The existing politicians owe their positions to the existing system.
In practice most votes wind up being hand-counted anyway, and the discrepancy between counts just makes things worse. You need a trustworthy system in the first place.
I see that point of view, but the secret sauce of our system is scrutineers. Mutual distrust ensures that almost any potential corruption is uncovered. In the USA you get allegations of incompetence and corruption for entire districts or even states. In Australia the worst it got at the last election was arguments about who brought How To Votes into the polling area.
Jacques,
There are a couple of things wrong with US electoral processes(they don’t deserve being called a system!).
1. The partisan nature of the electoral processes right down to precinct level – there are a few places that have non-partisan electoral processes like Iowa, and
2. FPTP particularly when used in conjunction with insecure computer voting – in the US computer voting systems are little more than overpriced abacuses because they do not keep a record of each individual vote as is required anywhere preferential voting is used. Have a look at ACTEC site to see how computerised voting systems should be handled.
As a result of the partisanship gerrymanders are common and electoral machinery may be in the hands of the party bosses – it therefore requires landslides to change governments.
Sy,
Americans are not more democratic – it just that they tend to do everything on the same day, and their constitutional changes are not by referendum!
The nicest thing I can say about the US voting system is its an improvement over the UK’s FPTP voting system.
Really? I’m less familiar with the UK system. What makes it worse?