July 16th 2017
As I was driving to work on Friday, listening to some music, it struck me that The Man With The Golden Gun is a truly terrible song. Lulu's giving it all she's got - "He charges a million a SHHHHOT!" - but the clumsy phallic imagery is really painful. It starts off badly with "He's got a powerful weapon", reaches possibly its lowest ebb of unsubtlety with "Who will he bang?", and is simply baffling with "It comes just before the kill" (I don't want to be crude, but if that line means anything then surely it means something that is generally perceived as... disappointing?). Anyways, the song's a bad 'un. This got me thinking. TMWTGG generally ranks towards the bottom of lists of Bond films, but it's always been one of my favourites: in Christopher Lee it boasts the finest actor ever to play a Bond villain (with honourable mention to Christopher Walken, Javier Bardem & Christoph Waltz); Roger Moore is on (mostly) fine form; the mirror scene is great; the sets are truly iconic - not only Scaramanga's island, but also that wonderful listing ship with the interior built horizontally - and it has possibly the greatest car stunt in the series. Swanee whistle aside.OK, Nick Nack wouldn't pass muster these days, and Britt Ekland's Mary Goodnight is more than usually subservient, even for a Bond girl, but if you come to a Bond film looking for political correctness - or even, let's face it, basic decency - you're going to be limiting your choice quite severely. Look, at least this one doesn't have Sean Connery pretending to be Japanese. And I get that Sheriff JW Pepper - as featured in the clip above - is not everyone's favourite addition to the franchise. But, for me, there's more than enough that's great in this film to make up for the stuff that isn't. So, how come it's so unpopular? Which brings me back to where I started. Driving into work, listening to Lulu warbling about golden shots & glittering ends, I hatched a theory: the quality of the theme song plays a large part in how fondly a Bond film is remembered. Does this work for the other films? My immediate thoughts were that it did: Skyfall is, to my mind, largely overrated; the theme song won an Oscar. Die Another Day & Quantum of Solace are the two worst Bond films of recent times; their theme songs are correspondingly bad. Maybe I was onto something. But could I use maths to prove it? Of course I could. Or, rather, of course I couldn't. What I wanted to prove is that the quality (or otherwise) of the theme song affects the way that people feel about the film as a whole. There are two problems with trying to prove that: the first is that quality is necessarily subjective (some of you may, even now, be screaming at your screens that the reason people don't like TMWTGG is that it stinks); the bigger problem is that correlation doesn't imply causation. To give an example from a book that Ant lent me called How Not To Be Wrong, in the twentieth century it became clear that smoking was strongly correlated with getting lung cancer - that is, smokers and lung cancer sufferers tended to be the same people - but this didn't necessarily mean that smoking caused lung cancer. At least one prominent statistician argued that, in fact, lung cancer caused smoking. Sounds pretty stupid, sure, but still quite mathematically sound. (That last sentence is how I describe myself on my CV). Anyway, let's take a quick break for a correlation joke.
Back to Bond. I can't prove that people are conned into thinking that TMWTGG is bad and Skyfall is excellent because of their respective songs, but I can try to prove that there's a correlation between the quality of the song and the quality of the film. I think that would be quite interesting, because there's no real reason to believe that there would be (apart, perhaps, from the argument that a good director / production team would strive for excellence in everything they did, and therefore the level of quality would be similar across the board). As an aside, I've mentioned this to some friends, and my friend Jen made the case that, because a lot of Bond films blend together in people's minds, they might generally remember the song better than the film itself. Could be something in that. To get round the issue of subjectivity as best I can, I took the internet's favourite film ranking system as my data set for film quality: step forwards, Rotten Tomatoes. I won't list all 24 films here, but it kicks off with Skyfall, Casino Royale & Goldfinger, and ends with TMWTGG, Octopussy & A View to a Kill. Poor Roger Moore. The song ranking was harder, because I'm not aware of a Rotten Tomatoes equivalent for music, but I had a quick google and pulled out lists from sites called Digital Spy, Consequence of Sound, and High Snobiety. Actually, I've decided to ignore the final list because it's clearly nonsense: I'll come back to that, but for now it's enough for you to know that they put Die Another Day in fourth place, and On Her Majesty's Secret Service last (for comparison, Digital Spy put OHMSS top). None of the lists included Dr. No, the first James Bond film, because it didn't really have a theme song other than the James Bond theme. Fair enough. Using the Digital Spy list, I made this graph comparing the ranking of songs against the ranking of films.
If the film and song quality were entirely uncorrelated you would expect the dots to be scattered at random; if they were perfectly correlated then the dots would be perfectly in line from bottom left to top right. If, as seems unlikely, the quality of the song were totally the opposite to the quality of the film ("inversely correlated" or "negatively correlated") the dots would be in a line from top left to bottom right. I think it's clear from the graph that, while the correlation is far from perfect, the dots are a bit more bottom-left-to-top-right than they would be if there were no connection between song and film quality. Fortunately, maths can tell us by how much. It's called a correlation coefficient - stamping all over my preferred terminology, bottom-left-to-top-rightness - and it indicates how unrandom the dots are. If the coefficient is zero then there's no connection at all; if it's 1 the connection is perfect and positive; if it's -1 then the connection is perfect and negative. I know, I know, the tension is too much. You haven't been this excited about hearing a number since last week's lottery draw. So I won't keep you in further suspense. The correlation coefficient is 0.537. So, while it's not as strong as I thought it might be, it's still a clear indication of a correlation between how much people like the films and how much they - or, at least, Digital Spy - like the songs. The coefficient is in fact lower for Consequence of Sound (0.452) and, intriguingly, practically zero for High Snobiety (0.075). Which suggests that the good people - or, rather, idiots - at HS make no connection at all between song and film quality. What about the correlation between song quality and song quality? This sounds like a trick question - the song quality must be directly correlated with the song quality, because they're exactly the same thing; it's like saying that tall people are tall and short people are, wait for it, short - but I've got three different rankings of song quality, and I want to know how well they tie together. Do Digital Spy, CoS & noted idiots HS all agree with each other? Well, you already know that they don't, but I want to put in some more numbers, so you're either gonna have to let me get on with it or you're gonna have to, I don't know, stop reading this page. And miss out on all those lovely, juicy numbers. Like a fool. It turns out that DS & CoS have pretty similar taste: the correlation between their two lists is 0.770. What about High Snobiety? Amazingly, the correlation between their list and Digital Spy's is a lowly 0.178 (and HS vs CoS is 0.356). So, if you weren't impressed by that 0.537 song/film match-up earlier - and you weren't impressed, were you? I can tell - just compare it to that song/song figure of 0.178, and admit how wrong you were not to be impressed. It's OK, I understand. I forgive you. So, there you go: all the film-based stats you could ever have wished for, and some. What a lot of fun we've had. Now go and watch The Man With The Golden Gun with the sound off, and tell me it's not great.
what am I listening to?
Anthology - Ocean Colour Scene
what am I reading?
How Not to Be Wrong - Jordan Ellenberg