Saturday, November 27, 2010

How Strictly Come Dancing teaches us that AV would be a mildly good thing

Nate Silver of FiveThirtyEight fame has written an article for the New York Times about what he sees as the flaws in the voting system used by Dancing with the Stars (the American carbon-copy of Strictly Come Dancing). It's quite an amusing piece for its earnestness and attention to the finest detail, although given my devotion to the Eurovision Song Contest I may not be in the best position to make that observation! But if you've ever wondered how Ann Widdecombe finds it so easy to survive in Strictly despite being placed bottom by the judges week after week, Nate has the answer for you -

"Suppose, for instance, that late in the season, when there are five couples left, four of the five teams receive 9’s across the board from the judges, and the final couple instead receives straight 7’s. In terms of the way the judges normally vote, that is a rather clear verdict: the low-scoring couple has had an inferior performance, and should be eliminated.

But in reality the low-scoring team would need to receive only 24 percent of the votes from the home audience — just barely better than the 20 percent they would get if the audience voted completely at random — to be guaranteed passage into the next round. It doesn’t matter if 24 percent of the audience thought they were the best-performing couple — and the other 76 percent thought they were the worst one! They would still advance to the next episode."

On the latter point, isn't that one of the obvious fatal flaws with any first-past-the-post voting system? Perhaps that irony wouldn't seem so obvious to an American political commentator more used to two-horse races, but it just so happens that is precisely the problem with FPTP that a 'Yes' to AV would remedy, even if it wouldn't address the far greater problem of disproportionality.

Silver goes on to make a series of detailed suggestions about how the voting arrangements on Dancing with the Stars could be improved, such as encouraging the judges to use the full range of possible scores between 1 and 10, rather than clustering most of the contestants between 6 and 10. That's fine in theory, but if the American show is anything like Strictly, the studio audience would probably start a riot if the weakest couples were routinely being given 1s and 2s.

The real problem with the show's voting system has always been the phenonemon of a couple placed in the middle of the leaderboard by the judges finding themselves being abruptly eliminated, simply because the public have a greater incentive to vote for couples at the bottom of the pile who are perceived to be in greater danger. At least this year with the scrapping of the dance-off we've been spared the tedious weekly ritual of the judges sanctimoniously announcing that "it is a travesty that you're in the bottom two, rather than X, Y or Z", neatly ignoring the fact it was partly the said judges' over-the-top criticisms of X, Y and Z that motivated the public to pick up the phone and save them. The obvious solution to this problem is surely to withhold the judges' scores until after the public have voted. I can't see that would detract from the show very much - The X Factor gets by quite happily without the judges scoring each performance out of 10.

But Strictly is just such a peculiar programme. Whatever the horrors of X Factor, at least it's a talent show in the truest sense of people being there on the basis of their talent. The Strictly philosophy is to randomly round up a group of people who for the most part, quite naturally, can't dance - and then get a smug Australian expert to scream abuse at them for weeks on end about their inability to dance. The producers pick contestants for their fame and popularity, not their dancing potential - and then the judges and the show's more humourless devotees work themselves into apoplexy because other people mysteriously treat it as a popularity contest, not as a "serious dancing competition". Bizarre.

UPDATE : Having thought about this some more, I've realised that either the American show must have a slightly different voting system, or else Silver must have misunderstood it. In Strictly, only the judges' rankings of the couples matter, not the raw scores. If anything, that makes it even easier for Widdecombe to survive.


  1. In the scenario you mention, where the top 4 couples were on 9s (which in the American version, with 3 judges, would mean they'd score 27) and the bottom couple was on 21 (3 x 7), the top 4 couples would have a score of 5 while the bottom couple would have 1. In order to secure a place in the next week, the bottom couple would have to come highest in the popular vote to get a total score of 6. If they came second, they'd equal the 5, but the highest public vote counts so one of those on 5 would have 6.

    In Strictly, we've seen cases when it's got down to the final 3, where the top placed couple with the judges has gone out because the bottom placed couple are popular with the public. It happened with Lisa Snowdon, Zoe Ball, Lesley Garrett, Aled Jones and Ali Bastian.

    While I loved all of those dancers, the people who displaced them, Tom Chambers, Darren Gough, Christopher Parker, Julian Clary and Chris Hollins, were all incredibly entertaining, and/or endearing. In the case of Tom, Darren and Chris, they were popular series winners.

    The parallels to AV are not exact, but it does give more power to the voter and ensures that the overall winner has the support of more than half the electorate. No, it's not proportional, but it's better than the system we have at the moment. It means at least that half the people will feel some level of positivity towards their representative, which is not always the case at the moment.

  2. Agreed, Caron - I certainly won't vote for AV with any great enthusiasm, but the more I've thought about it, the more I've come to the conclusion it does represent a slight improvement on FPTP.