Spoilers For Monday August 15th 2016

[James Robinson]

Just 2 shows this week thanks to the horse racing, but 2 is the magic number for Annie Humphries, as if she wins the games today and tomorrow, she will become an octochamp!!!

Charlie Stayt finishes his 2nd stint in DC too.

Join Sean for the recap later.

[Johnny Canuck]

How come horse racing equals cancelling Countdown (rather than moving it to a different time slot), anyway? It (Countdown) apparently costs very little to make. I think someone in Apterous chat should organise impromptu episodes of it instead.

Alternate final numbers: (100 - 7) * 10 + 4 + 5

[Steven M. McCann]

The last letters selection seems to be producing more and more "nines", coincidence or deliberate?

[JimBentley]

The last letters selection seems to be producing more and more "nines", coincidence or deliberate?

I was going to attempt to reply to this with HARD MATHS to show how astronomically unlikely the coincidence would have to be, but on second thoughts, it's just too stupid to bother with. And as you've provided no evidence of this supposed phenomenon, I'd guess It's more like confirmation bias on your part. Unless you're outright accusing the production team of cheating?

[Johnny Canuck]

Two nines in the past two days. No others.

"The plural of anecdote is not evidence." -- don't remember who said this, but it's a smart person

[Steven M. McCann]

These last couple of games have just confirmed what I have believed for well over a year, that the last letters selection is very "nine friendly" for want of a better expression.
It may be a self fulfilling prophecy, but I spot more "nines" in the last letters selection than any other letters selection.
I half recall there was a game last week, had Annie's opponent picked the correct vowel, for their final selection instead of the consonant they did choose, there would have been another "final selection nine".

[Fred Mumford]

It may be a self fulfilling prophecy, but I spot more "nines" in the last letters selection than any other letters selection.

Me too, and after a detailed mathematical and statistical analysis of why this might be, I have concluded that it's because I only get home at quarter to four.

[Graeme Cole]

The last letters selection seems to be producing more and more "nines", coincidence or deliberate?

These last couple of games have just confirmed what I have believed for well over a year, that the last letters selection is very "nine friendly" for want of a better expression.
It may be a self fulfilling prophecy, but I spot more "nines" in the last letters selection than any other letters selection.
I half recall there was a game last week, had Annie's opponent picked the correct vowel, for their final selection instead of the consonant they did choose, there would have been another "final selection nine".

Sigh. Not this again.

What you call a "self fulfilling prophecy" has a well-known name which Jim has already referred to.

Are you really suggesting that the production team fix the letters piles to make nines come out in R13? If you were going to rig the letters to make a nine come out, tinkering with the piles during the game would look a bit suspicious to the audience. So you'd have to do it before the game, before the letters are put into the boxes. How would you do it? Sure, you could put specific letters on the top of the piles to make a nine come out in round 1, but how would you do it for round 13? You can't predict exactly how many vowels and consonants will have been picked in the first nine letters rounds. Where would you plant your nine?

But anyway. Let's suppose, for the sake of argument, that the Countdown team have some secret and nefarious reason to want nines to come out in the last round. Let's further suppose that they've found some magical way to manipulate the letters piles unseen right after R12, or they did it before the game and have somehow got the contestants to pick just the right amount of vowels and consonants throughout the game, possibly involving mind control drugs, chemtrails, and the Illuminati.

Such shenanigans, if done regularly, would show up in the stats.

Let's look at letters games over the past year, like you say. These figures are for series 73 and 74. That's 2,130 letters rounds over 213 games.

Number of rounds in which a nine was available according to the wiki, by round number:

Code: Select all

  ROUND NINES
      1    21
      2    18
      4    12
      5    25
      7    19
      8    17
     10    27
     11    17
     12    16
     13    23

Average length of the longest available word according to the wiki (counting 9 as 9, not 18), by round number:

Code: Select all

  ROUND  AVG MAX LENGTH
      1            7.56
      2            7.47
      4            7.35
      5            7.41
      7            7.40
      8            7.42
     10            7.51
     11            7.39
     12            7.39
     13            7.48

If you'd said "far fewer nines occur in round 2 than other rounds" then it'd be an interesting point. It still wouldn't be evidence of shenanigans, of course, unless you could suggest a compelling reason why they'd go to all the trouble of rigging the letters piles and risking the reputation and integrity of the programme if caught, just to reduce the number of nines available in round 2, but your observational data would at least be correct. (edited, see comment below)

There are ten letters rounds, so you would expect about 10% of the nines to fall in R13. In series 73 and 74, 11.8% of the letters rounds with a nine available were round 13.

Incidentally, in one of my games at CoNot recently, there was an unusual letters selection, which I've been saving for a post like this. Coincidence, or deliberate...?

[Philip Wilson]

Nice to have the spoiler thread back anyway!

[JimBentley]

Cheers to the estimable Mr. Cole for demonstrating what I couldn't be bothered to! Now, Steven:

I have believed...I half recall...had x done y...would have been...

Do you understand why you are going wrong with this?

[Steven M. McCann]

I know it's only based on two series worth of data, but a near 25% better than average chance of getting a nine in the last letters selection, is still significant.

[Johnny Canuck]

When Susie reviewed Annie's offering of PLUMINESS in round 13, she seemed to imply that it could be a variant of the word Annie pronounced, PLUMMINESS (presumably having this variant meaning in addition to its usual formation from PLUME). Odd that she wouldn't mention the more common pronunciation/meaning.

[Jon O'Neill]

I know it's only based on two series worth of data, but a near 25% better than average chance of getting a nine in the last letters selection, is still significant.

I for one would like to see your workings..

[Countdown Team]

The last letters selection seems to be producing more and more "nines", coincidence or deliberate?

Or just preposterous?

[Graeme Cole]

I know it's only based on two series worth of data, but a near 25% better than average chance of getting a nine in the last letters selection, is still significant.

I for one would like to see your workings..

I assume he means that if you look at the numbers in the first table (number of rounds with nines by round number) and take the mean, you get 18.5, and the number of R13s with nines was 23, which is nearly 25% higher than 18.5.

As it happens, I've made a small mistake above. Remember that unusually low figure for R2? That should be 18, not 8. I must have mispasted it. So the average is 19.5, and 23 is only 18% higher than that. Some might say that it still looks significant, but it's within one standard deviation (4.30) of the mean.

[Jon O'Neill]

I know it's only based on two series worth of data, but a near 25% better than average chance of getting a nine in the last letters selection, is still significant.

I for one would like to see your workings..

I assume he means that if you look at the numbers in the first table (number of rounds with nines by round number) and take the mean, you get 18.5, and the number of R13s with nines was 23, which is nearly 25% higher than 18.5.

As it happens, I've made a small mistake above. Remember that unusually low figure for R2? That should be 18, not 8. I must have mispasted it. So the average is 19.5, and 23 is only 18% higher than that. Some might say that it still looks significant, but it's within one standard deviation (4.30) of the mean.

I mean I want to see the workings in significance. How confident can we be that they are fixing it based on the observations?

[David Williams]

Years ago I remember there was a run of nine letter words available in the first round of the first show of the week, which would at least be feasible to fix. But even then it was probably a coincidence. Particularly as the letters were (allegedly) manually sorted a bit. If you were doing that you'd probably not have Q or X or Z too near the top of the pile, and you might leave U out of the top few, so there's a pretty good chance of RELATIONS or TRIANGLES or something similar.

[Gavin Chipper]

I mean I want to see the workings in significance. How confident can we be that they are fixing it based on the observations?

I think they're also fixing rounds 5 and 10, and also round 4 in the opposite direction. I imagine it's hard work, so there must be a reason for it! The stats don't lie!

(Remember to always read the spoilers threads for games you miss when you're on holiday, even if you don't bother to watch the games.)