c4countdown

SONDAGE as a DC beater in R2.

Dogvanes for R2

Should Susie have allowed GILTER? It is a one-syllable adjective: https://en.oxforddictionaries.com/definition/gilt

Most likely a dictionary error. Gilter doesn't make sense does it?

No it doesn't make sense, but there are plenty of one-syllable adjectives that don't make sense to stick -ER/-EST onto in this dictionary.

Also, however you're generating random numbers needs to be changed, because they aren't random. These are the same numbers from Sam Prouse's game: http://wiki.apterous.org/Episode_6750

Great... looks like CECIL has a virus again! Good spot, Ray.

4th Numbers Alt.: 6 / 3 * 5 * 75 + 25 - 1

Well, it's Sam who spotted it, I can't take credit for that.

So if it doesn't make sense, does that mean that -er and -est is decided on a whim by Susie?

Ray Wilding wrote: ↑Wed Aug 08, 2018 1:56 pm No it doesn't make sense, but there are plenty of one-syllable adjectives that don't make sense to stick -ER/-EST onto in this dictionary.

Also, however you're generating random numbers needs to be changed, because they aren't random. These are the same numbers from Sam Prouse's game: http://wiki.apterous.org/Episode_6750

It's been rectified.

Countdown Team wrote: ↑Wed Aug 08, 2018 2:21 pm

Ray Wilding wrote: ↑Wed Aug 08, 2018 1:56 pm No it doesn't make sense, but there are plenty of one-syllable adjectives that don't make sense to stick -ER/-EST onto in this dictionary.

Also, however you're generating random numbers needs to be changed, because they aren't random. These are the same numbers from Sam Prouse's game: http://wiki.apterous.org/Episode_6750

It's been rectified.

Excellent news, well done team!

Countdown Team wrote: ↑Wed Aug 08, 2018 2:21 pm

Ray Wilding wrote: ↑Wed Aug 08, 2018 1:56 pm No it doesn't make sense, but there are plenty of one-syllable adjectives that don't make sense to stick -ER/-EST onto in this dictionary.

Also, however you're generating random numbers needs to be changed, because they aren't random. These are the same numbers from Sam Prouse's game: http://wiki.apterous.org/Episode_6750

It's been rectified.

Glad to hear -- what turned out to be the issue? Was the algorithm broken?

No idea what the issue was, suffice to say someone noticed it but the target repetitions will continue for a while yet, as we've recorded quite far ahead. It's of little consequence in reality, the 6 numbers picked out vary so much, but yeah, a strange one.

Ray Wilding wrote: ↑Wed Aug 08, 2018 2:08 pm Well, it's Sam who spotted it, I can't take credit for that.

This is an amazing spot.

I've run some queries on the database to see if I can find any other strange patterns.

To start with, I've only looked at "new CECIL", that is, starting from this game in 2013 when the introduced the new displays, up to the end of series 78. This covers 4,897 numbers rounds in total.

The numbers targets in episode 6620 were the same as the targets in episode 6622 two days later, but someone already seems to have spotted that and noted it on the wiki, and the repetition didn't last longer than those four targets.

Episodes 6707, 6745, 6837 all had the same numbers targets in the same order. In the case of episodes 6707 and 6745, the repetition followed the next day as well.

What if we just look at targets, rather than games? If CECIL is following a predictable sequence, then we might see the same long sequence of numbers more than once. And we do.

The 20-number sequence (686, 767, 896, 192, 603, 372, 503, 648, 842, 307, 537, 922, 666, 634, 392, 750, 127, 761, 535, 325) happened twice in six months: in the five episodes starting from episode 6716, and the five episodes starting from episode 6802. Both of those occasions spanned exactly one filming day (presumably, since in both cases the same DC guest was present throughout the sequence, but not for the episode following). It's possible the repeated sequence would have been longer - perhaps it gets interrupted when the random number generator is switched off at the end of the day.

That is by far the longest repeated sequence. There's a shorter, 8-number repeated sequence (986, 910, 301, 596, 399, 330, 109, 344) in the two episodes starting from episode 6680 and the eight numbers rounds starting from R9 in episode 6807.

See also a 7-number repeated sequence (670, 717, 339, 364, 338, 406, 252) starting from R3 of episode 6625 and R6 of episode 6827.

All these examples seem to be from mid-2017 onwards. In fact, all instances of the same sequence of 3 or more targets appearing more than once come from episode 6620 (July 2017) or later - before then, no sequence of three or more targets had ever been repeated, even if you also include all the games for which we have details, going back to 1982.

So I don't think it's anything to do with the new monitors that replaced the old seven-segment displays in 2013, because it worked fine between then and 2017.

Something seems to have happened in mid-2017 to whatever generates the random number. It's as if it's now got a great big list of random numbers in a random order and it's working through the list in sequence, rather than generating a new random number each time.

Alternatively, perhaps the random number generator is being initialised with a seed value that isn't as unpredictable as it should be. The Wikipedia article on pseudorandom number generators (PRNGs) discusses the topic in detail, but in summary: a PRNG is initialised with a "seed value" to start off with. Ideally this should be something reasonably unpredictable and non-constant (the current date and time in milliseconds is a popular choice). Each time the PRNG is asked to produce a number, its answer depends on the PRNG's internal state, which is only ever affected by the initial seed value and any numbers it's already generated. This means if a PRNG is initialised today with the same seed value it was initialised with yesterday, it's going to come out with the same sequence as it did yesterday.

As far as Countdown's random number generator goes, the chance of a contestant being able to exploit this in their favour is nonzero but small. The main exploit scenario is when a contestant, Cheaty McCheatface, plays a couple of games, then over lunch realises that the eight targets they've been given so far also occurred in the same sequence a few months ago, so now he's pretty certain what the next target is going to be. Even that doesn't help him too much, because, as Countdown Team says, the six starting numbers are as good as random. Mr McCheatface's best bet is to pick four large, but there are still 55 possible four large selections to go away and learn the solutions for over lunch.

So yeah, nonzero but small. Deal or No Deal's "sequencegate" was much more exploitable, but they caught it pretty early on. There doesn't seem to be a single article summarising it, but the "Glenn Hugill writes" section here, and this Digital Spy post sum it up. If I'm reading it right, for a while the amounts of money in the boxes always followed one of a very small number of sequences, shifted to start on any one of the 22 boxes. This meant that after opening a few boxes, you could identify which sequence was being used, then from the known boxes you could work out what was in your box.

See also Michael Larson on Press Your Luck, a more well-known example of weak randomness on a game show, which was actually exploited (Wikipedia, YouTube).

Congratulations on reading to the end of this long post. Have a cookie.

Rhys Benjamin wrote: ↑Wed Aug 08, 2018 2:09 pm So if it doesn't make sense, does that mean that -er and -est is decided on a whim by Susie?

It seems so. It's also invalid on Apterous, which surprises me, since you'd expect Apterous to more rigidly follow the rules.

Graeme Cole wrote: ↑Wed Aug 08, 2018 10:28 pm

Ray Wilding wrote: ↑Wed Aug 08, 2018 2:08 pm Well, it's Sam who spotted it, I can't take credit for that.

This is an amazing spot.

Nice post generally, but it's not *that* amazing to notice the targets were the same as the one show when you were on!

Gavin Chipper wrote: ↑Thu Aug 09, 2018 10:04 am

Rhys Benjamin wrote: ↑Wed Aug 08, 2018 2:09 pm So if it doesn't make sense, does that mean that -er and -est is decided on a whim by Susie?

It seems so. It's also invalid on Apterous, which surprises me, since you'd expect Apterous to more rigidly follow the rules

For the avoidance of doubt in future shows, we've decided from now on that ALL 1-syllable adjectives can be extended with -er and -est, even though it'll be difficult to understand the logic behind some of them. So, GILTER gets the green light, despite it not really making much sense.

Countdown Team wrote: ↑Thu Aug 09, 2018 11:19 am

Gavin Chipper wrote: ↑Thu Aug 09, 2018 10:04 am

Rhys Benjamin wrote: ↑Wed Aug 08, 2018 2:09 pm So if it doesn't make sense, does that mean that -er and -est is decided on a whim by Susie?

It seems so. It's also invalid on Apterous, which surprises me, since you'd expect Apterous to more rigidly follow the rules

For the avoidance of doubt in future shows, we've decided from now on that ALL 1-syllable adjectives can be extended with -er and -est, even though it'll be difficult to understand the logic behind some of them. So, GILTER gets the green light, despite it not really making much sense.

Please no... I don't wanna go back to PAINEDER/PAINEDEST, CANEDER/CANEDEST, CAGEDER/CAGEDEST and so on...

Ray Wilding wrote: ↑Thu Aug 09, 2018 12:15 pm

Countdown Team wrote: ↑Thu Aug 09, 2018 11:19 am

Gavin Chipper wrote: ↑Thu Aug 09, 2018 10:04 am

It seems so. It's also invalid on Apterous, which surprises me, since you'd expect Apterous to more rigidly follow the rules

For the avoidance of doubt in future shows, we've decided from now on that ALL 1-syllable adjectives can be extended with -er and -est, even though it'll be difficult to understand the logic behind some of them. So, GILTER gets the green light, despite it not really making much sense.

Please no... I don't wanna go back to PAINEDER/PAINEDEST, CANEDER/CANEDEST, CAGEDER/CAGEDEST and so on...

All of those are 2 syllables, though, if you take “-ed” as a second syllable as I do.

I should maybe put on record that I don’t really have a problem with Susie deciding these as long as she is consistent. Especially now that she is the only lexicographer and it’s no longer as whimsical as it was in the old days (SCREES being an example: initially disallowed by Damian as lex but overruled by upstairs (Ballheimer, 2003)).

Rhys Benjamin wrote: ↑Thu Aug 09, 2018 11:33 pm

Ray Wilding wrote: ↑Thu Aug 09, 2018 12:15 pm Please no... I don't wanna go back to PAINEDER/PAINEDEST, CANEDER/CANEDEST, CAGEDER/CAGEDEST and so on...

All of those are 2 syllables, though, if you take “-ed” as a second syllable as I do.

Are you one of those infuriating people who say "Horn-ed Viper"?

Well it's really one-and-a-half syllables, isn't it? I mean, a word like "milk" is clearly not one syllable, but neither is it a full two either.

How many syllables in BANNED? Or BAND? (Actually a special Countdown rule that words like PAINED are considered to have two syllables is not a bad idea.)

David Williams wrote: ↑Fri Aug 10, 2018 11:32 pm How many syllables in BANNED? Or BAND? (Actually a special Countdown rule that words like PAINED are considered to have two syllables is not a bad idea.)

I don't think you can define the number of syllables in a word based on whether you want it to have a comparative and superlative form!

Something along these lines might work, however.

That's what needs to be achieved. I hadn't got into the precise wording. Your link provides it. Mr Team?

Gavin Chipper wrote: ↑Sat Aug 11, 2018 7:25 pmSomething along these lines might work, however.

Crikey, I'd forgotten I cared so much.

Henry W. Fowler's Dictionary of Modern English Usage has an informative section on comparatives and superlatives. The first sentence seems especially apt:

Neglect or violation of established usage with comparatives and superlatives sometimes betrays ignorance, but more often reveals the repellent assumption that the writer is superior to conventions binding on the common herd.

Ray Wilding wrote: ↑Thu Aug 09, 2018 12:15 pm

Countdown Team wrote: ↑Thu Aug 09, 2018 11:19 am

Gavin Chipper wrote: ↑Thu Aug 09, 2018 10:04 am

It seems so. It's also invalid on Apterous, which surprises me, since you'd expect Apterous to more rigidly follow the rules

For the avoidance of doubt in future shows, we've decided from now on that ALL 1-syllable adjectives can be extended with -er and -est, even though it'll be difficult to understand the logic behind some of them. So, GILTER gets the green light, despite it not really making much sense.

Please no... I don't wanna go back to PAINEDER/PAINEDEST, CANEDER/CANEDEST, CAGEDER/CAGEDEST and so on...

Uh-oh... and painedest *still* doesn’t have a valid alt. At least paniered stayed when paidreen got destroyed