A long time ago Frances and I mused "what are our top ten favourite films?"
There were a lot of "top ten" programmes on British TV, and a friend gave
me a book of movie top tens, so I thought it would be interesting to try and do
my own list.
That was probably a mistake. This article took about a year to write, on and
off. It's not trivial! My sorry tale reveals some interesting analysis issues,
possibly, though hopefully not, of the psycho- sort :)
Things started well, lulling me into a false sense of security. With almost
no effort I came up with 5 "definite" entries, which would have to be in the
list regardless:
T2
The Italian Job (the proper 1968 one)
Die Hard
Paint Your Wagon
The Mummy (1999)
So that was the top 5 sorted. I did say this was my favourites, not those
deemed worthy by earnest film critics, didn't I?
So far so good. But then it got more difficult. A quick scan of my DVD
collection produced a long list of candidates, but no compelling reasons why one
or another must beat the others into the top ten:
The Hound of the Baskervilles (Basil Rathbone), Under Siege,Zulu, Speed, Goldfinger, Live and Let Die, Licence to Kill,
Dracula (Frances Ford Coppola), Star Trek IV (narrowly shading II, VI and IX, which are all
viewing regulars), Jurassic Park (the first one), The Dambusters,
The Fifth Element, Crimson Tide,
Evolution, Tremors, Groundhog Day, Wilt, Striptease,
The Avengers, Vertical Limit, Monsters, inc., Fire Down Below,
Back to the Future III
Great, so I now had a top 28! But I wanted a top ten ;) So how should I
narrow it down? That was the fatal mistake - I asked an analytical question, and
kicked off an entirely inappropriate analytical process.
I just couldn't select 5 out of the 23. According to my calculations there
are 33,649 different ways of selecting 5 films from 23, so it would be "unfair"
to select randomly. Equally, I could have turned it into a top 25 or 30, but
that would plainly be cheating.
Then I thought "Maybe I could select films to represent certain favourite
aspects". For example, my list should include a Bond film, and a Star Trek. It
should include something directed by each of Stephen Spielberg, James Cameron,
John McTiernan, Andrew Davies and Guy Hamilton. (Oh, I forgot Roland Emmerlich.
Better add Independence Day to the list!) It should include performances by Bruce
Willis, Keanu Reeves, Sandra Bullock, Michael Caine etc. etc. ....
Wait a minute! What about Russell Crowe and Tom Hanks? I added Apollo 13
and
Gladiator.
This was turning into another long list. And then I noticed that my
"shortlist" of 23 (now 26) films didn't include a Carry On or Ealing Comedy.
Surely there should be room in my top ten for the comic genius of Sid James, or
Alistair Sim, or George Cole? I added The Belles of Saint Trinians, Carry on
Cleo, The Big Job, The Ladykillers and Two Way Stretch to the list, resisting
the temptation of about 5 more.
There were now two dangers. My increasingly inaccurately named "short"list
was in danger of growing to rather more than 30 . Worse, I was becoming tempted to start a
"best fit" exercise which would be worse than random selection. For example,
Chain Reaction is a good film, but didn't make the top 28. However, it would
cover several of the "must include" requirements. Would it be "sensible" to
include it instead of one of the others?
Hang on! Where did "sensible" come from? Had I started my own "Chain
Reaction"?
Struggling, I did what most men would do. I asked my wife for advice. However
"don't overwork it" and "just choose your favourites" didn't help much. Some of
her suggestions just added to the shortlist. "That space one" (The Fifth
Element) was already on the list, but "that other space one" (Stargate) wasn't.
Others sounded suspiciously like her favourites, not mine. Fried Green Tomatoes
At The Whistle Stop Cafe did not get added to the list. (I was tempted by
The
Thomas Crown Affair, but she only really suggested it because she fancies Pierce Brosnan.)
I toyed with the idea of a scoring scheme, deriving the top ten by numerical
analysis of several different criteria. But that sounded suspiciously like
"overworking it". And what would I do if one of the top five didn't make it?
What about "how often do we watch them?". That isn't quite right either. For
one thing timing skews the statistics. There's often time, and we're often in
the mood for Evolution or The Italian Job, when we don't have the time or the
emotional strength for Zulu. But it's still a "favourite". Also this is "my"
list, and the frequency of watching would generate a joint list, something
subtly different (for example, no films about submarines, and no films with no female
characters. Damn! I forgot The Hunt for Red October - better add it to the
list.)
Length of exposure isn't a discriminator. The Italian Job has stood the test
of time, but Where Eagles Dare, which used to be a favourite, hasn't. And
some quite new films, such as Sahara, have become "favourites" quite
quickly.
Frances suggested that I imagine I'm in jail for a year and forced to watch
the same ten films over and over. Which ones would I not tire of? I suspect
such a punishment counts as "cruel and unusual", but I suppose it's only the
Desert Island Disks idea by another name. Sadly it didn't help. Most of my top
30-ish would be OK, although I might get more and more annoyed about the
mis-casting of Uma Thurmann in The Avengers. And again, this test produces a
skewed list, since I'd end up choosing the funny and uplifting over the serious
and black.
I thought about choosing one film to represent each genre in the shortlist.
That would be easy, if I could choose the right five genres. But after
"historical" and "classic comedy" I ran out of steam, unable to decide on
unambiguous classifications within the remaining comedy/sci-fi/action continuum.
I desperately needed some way to reduce the number of choices I had to make.
And suddenly it came to me - Knock-out cup instead of league. I could use a binomial
sort, and reduce my problem to a series of simple choices between pairs of
films, on the assumption that both were equally appropriate at the time.
The big problem with knock-out competitions is that the strongest competitors
may meet too early. I decided to ignore this, but roughly group "similar" films
in the shortlist to ensure some variety in the output. By coincidence I now had
a "shortlist" of 32, so I didn't have to worry about odd numbers.
So finally, after some hard choices, I reduced my list to another 5 films: Zulu,
Jurassic Park, Vertical Limit, Independence Day, Carry on Cleo.
So here, by an excessively complicated route, is my top ten, or at least my
top 5 and another 5:
T2
The Italian Job
Die Hard
Paint Your Wagon
The Mummy
Zulu
Jurassic Park
Vertical Limit
Independence Day
Carry on Cleo
Are these truly my top ten favourites? I don't know. Just don't ask me to put them in order!