Friday, 6 January 2012

Weekly summary: Episodes 351 to 355

It's been a mixed week for me, starting off quite well and (sans conundrum on Tuesday) staying very close to the combined David/Lily score.  In fact on Wednesday I just pipped them -- only the second time this series that I have managed to do so.  Things went downhill come Thursday, though, and the week finished on a disappointing note.  This may have been my worst week for conundrums, too, only managing to solve two of the five.

David + Lily7288747487

Sebastian made it to a fifth game, pushing me down a spot in the rankings.  Kerin has a chance of doing the same, too -- she just needs to win Monday's game.  I'm still clinging to that sixth position, but it looks shakier and shakier as the series winds on.  Note that Kerin has already scored enough to ensure that she'll eliminate Alex van der Kooij from the running even if she loses, so she makes it onto the rankings table even though Alex has technically played more games than her at this point.

Sam Gaffney518162676455380
Toby Baldwin594854434844296
Shaun Ellis435638594044280
Daniel Chua445956533652300
Sebastian Ham3949584045
Geoff Bailey55646563

Nick Terry71465535

Kerin White555646


Other statistics:

Full Monties

Missed Full Monties1

Tough Numbers1

Impossible Numbers


Contestants averaging over 30 points a game:

Sam Gaffney380663.33
Geoff Bailey247461.75
Jimmy Driscoll61161.00
Leanne Cox57157.00
Ryan Sutton57157.00
Nick Terry217454.25
Kerin White157*3*52.33
Tim Clay51151.00
Peter Crop152350.67
Daniel Chua300650.00
Lainie Mercieca99249.50
Toby Baldwin296649.33
Natasha Podesser47147.00
Shaun Ellis280646.67
Brett Edwards139346.33
Sebastian Ham231546.20
Michael Nichols90245.00
Karla Treves90245.00
Sandy Clarke45145.00
James Godfrey45145.00
Megan Marks133344.33
Nick Compton44144.00
Mark Arnold87243.50
John Day42142.00
Nathan Dixon41141.00
David Bradley77238.50
Alex van der Kooij153438.25
Adrian Lonigro38138.00
Colin Jones111337.00
Cherie Brody37137.00
Ann Vasconcelos73236.50
Ilona Coote36136.00
Kane Gross34134.00
Duncan Butler33133.00
Mitchell Fly33133.00
Hannah Marshall33133.00
Hiep Do98332.67
Susan Cumming32132.00
Sushma Garudadwajan62231.00
David Armstrong31131.00


Mark said...

What do you think is the best way of ranking the players in order of ability?

Geoff Bailey said...

That's a very tricky question, Mark; mostly I'm going to dodge it. At the moment, all I can be sure of is that a single game is not enough to tell much, even if a contestant is eliminated.

(As a quick illustrative side remark: Sam and myself have been comparing scores as though we were playing against each other. Over the last nine games -- including tonight's episode -- I have won four, Sam has won four, and the other is effectively a dead heat as we cannot tell who would have solved the conundrum quicker. Our aggregates are exactly the same, with an average of over sixty points a game. Trying to meaningfully distinguish between us is obviously going to take a lot longer.)

As a purely practical measure, games won is clearly a useful metric; the problem with it is that two different games are hard to meaningfully compare -- both the performance of the opposition and the options available have a huge impact. The numbers rounds can range from trivial to impossible, and either can be fatal for a contestant whose strength is the numbers.

All that said, I do feel that average score is a metric that tells us something, and perhaps more than games won does. Of course, the more games that a contestant plays the better idea we have of their overall strength; average score for single-game contestants is hard to rely on.

A system that might work better -- if we are looking to rank contestants as opposed to simply enjoying the contest -- is to look at solo scores for contestants, while adjusting the scoring to take into account what was possible, as well as how easy it is to find that answer. A very crude approximation to this is to consider a contestant's solo score as compared to the combined total of David and Lily. It's possible I may put up statistics for that some day.

But that's still a flawed metric, and luck will continue to play a significant role as long as the show is a knockout competition (as it should be in order to be entertaining).

Mark said...

Thanks for your comments Geoff. The question was motivated by your second table, which I gather is how the players are ranked to qualify for the finals. In my opinion there's little doubt that you're better at the game than Sebastian (no disrespect to him intended).

Geoff Bailey said...

Thank you for your kind assessment, Mark. I should draw the distinction between my performance now and my performance then -- I am much better now, in my opinion. Still, we can compare statistics based on the games of record.

Here's a flawed but possibly interesting metric: For letters rounds, take the average number of letters the contestant is away from David's word. (Invalid answer counted as length zero.) For numbers rounds, consider their average distance from the target, or the nearest achievable if it is unsolvable. (Invalid or out of range answer arbitrarily assigned a distance of 15.)

Then Sebastian was an average of 1.96 letters away from David, and 4.07 away from the numbers target. He solved two out of five conundrums.

In contrast, I was an average of 0.85 letters away from David (that invalid selection really cost!), and 0.33 away from the numbers target. I solved 1 out of four conundrums.

To be even more arbitrary, I'm going to make the scoring more discrete: The numbers scores as usual, but letters scores 10 for 0 away, 7 for 1 away, 5 for 2 away, and 3 for 3 away (0 for 4 or more off the target). The conundrum scores 10, 7, or 5 depending on whether it was solved in the first, second, or third grouping of ten seconds. There is no significant thought behind this decision, and one should be wary of reading anything into it.

On that basis, a game is worth at most 90 points and Sebastian scored 269 in five games, for an average of 53.8 points per game. I scored 285 points in four games, for an average of 71.25 points per game.

Just for fun, let's look at Sam Gaffney's figures. Sam averaged 0.93 letters away from David, and 1.72 away from the numbers target. He scored 415 points in 6 games, for an average of 69.17 points per game.

Can we read anything into this? Some, but not a lot. In particular, the above metrics are very flawed because they over-weight invalid results. But I can still use them to tease Sam. *grins*

If you can come up with something that you think could be used to rank performances in different games, I'd be very interested.