c4countdown

To paraphrase 1080 Snowboard, "Ohh, it's a new records page!"

Finally got round to acting one of Ben's good suggestions: the "Perfect Games" section for each format now shows the nearest-to-perfect games if there are no perfect games. See for example here (unlucky Kirk.)

I've revamped the Daily Duel pages to tidy them up, and added a few new things: a Winners page and a Statistics page.

A bunch of new stuff in Statland, including busiest players, a promiscuity league and, as requested by Gevin, more breakdowns by length (e.g. most missed 9s.)

Most requested feature ever: you can finally disable the goddamn sound on your Account page.

At last, a better metric for Top Players than ratings:

http://www.apterous.org/statland.php?se ... op_players

And further breakdowns by discipline:
http://www.apterous.org/statland.php?se ... rs_letters
http://www.apterous.org/statland.php?se ... rs_numbers
http://www.apterous.org/statland.php?se ... conundrums

I've beefed up the way this is calculated to include all suitable rounds, so we have a new Top 25 Players. At least 3 of those 25 are in the current series

Oh, I also added Top Goatdown players.

Awesome stats Charlie *jism

Extra stats are always cool. I'm curious about how they're calculated though [EDIT - now I'm not, since I actually read the pages properly], since objectively there are multiple ways of measuring competence.

My thoughts:

• In letters rounds you can simply take points scored, but that creates an enormous imbalance between players that play difficult opponents like Prime and Rex, and those that just hammer Prune all day.
• A solution to this would be to consider flat scoring, but then you reward conservative play because disallowed words strongly hurt your score. It's also likely that a player offering a disallowed word had a valid word of shorter length.
• A solution to that would be to only consider rounds in which the player has offered valid words, but then there is no penalty for shoddy dictionary knowledge and it is also open to abuse ("only offer word if likely to be a max or is a 9, otherwise declare GANDISEEG").
• A possible solution to this is to automatically give a player points equal to the floor of half the max (so 1-4 points) if they offer a disallowed word, since it's highly implausible any player would fail to spot at least a 4 letter word if there was a 9 in the selection. Now we're leaving the realm of hard data, though.
• I see the current solution is to consider only maxes, which while not an exact measure of competence, probably has a strong enough correlation to suit our needs [citation needed]. I do wonder about goatdown though, because if you're playing bots you seem to get different kinds of maxes than when playing human opponents (cf. Paul Howe's games against Prime and Rex).
• With conundrums, your score is even more highly dependent on your opponent. You might think to consider only conundrums in which the player had the full 30s (i.e. player got the conundrum, or neither player got the conundrum), but that probably messes up the stats for those players that tend to play against difficult human opponents.
• I can't think of a good solution to this, apart from considering the time it takes to solve the conundrums and leaving it as an average-time-to-solve rating. So if a player is beaten to a conundrum spotted by an opponent in 2.3s, then that player is only penalised by 2.3s added onto their average solving time.
• The numbers game has its own problem, namely that those players who favour harder formats (6 small, 4 large) probably score less on average than those that choose 1 large. I suppose it's possible to add a constant factor to each numbers game depending on type, relative to the statistical average number of maxes for that format in all games. Since Apterous doesn't ask for a solution when you're beaten by an opponent, it's hard to judge based on flat max too. I haven't thought of a way to counter this other than the current solution to count maxes (and I'm getting sick of typing).

In summary, I'm really only dubious about the current methods of showing conundrum competence.

Simon Myers wrote: [*]With conundrums, your score is even more highly dependent on your opponent. You might think to consider only conundrums in which the player had the full 30s (i.e. player got the conundrum, or neither player got the conundrum), but that probably messes up the stats for those players that tend to play against difficult human opponents.

[*]I can't think of a good solution to this, apart from considering the time it takes to solve the conundrums and leaving it as an average-time-to-solve rating. So if a player is beaten to a conundrum spotted by an opponent in 2.3s, then that player is only penalised by 2.3s added onto their average solving time.

In summary, I'm really only dubious about the current methods of showing conundrum competence.

Yeah, I think the conundrum stats are pretty dodgy. But if you play Rex a lot then he'll get every conundrum in about 1 second, so if we just take your suggested optimistic approach then that player ends up with a ridiculously good average even though they never got a single conundrum right. Can't really think of a good solution to this right now, I will talk to my official statistical raccoon.

Had a bit of a dies mirabilis yesterday, fixing three very long-standing bugs in one marathon session at the keyboard:

• Completely re-implemented chat/spectate system, so that it now works all the time for everyone. Seriously.
• Duel sometimes not showing up when you log in.
• Duel being squashed in the challenges window so you can't see what it is until you start it (amusing though this was.)

Charlie Reams wrote:Completely re-implemented chat/spectate system, so that it now works all the time for everyone. Seriously.

This might be relevant - just logged in and apparently there are 125 other people in the room according to the chatbox.
Only twenty or so in the player list, I'm guessing if game/chat systems are separate then people are logging out of the game but somehow still logged in to chat?
Which might well be how you have fixed it, just thought it was worth mentioning.

Charlie Reams wrote:Had a bit of a dies mirabilis yesterday, fixing three very long-standing bugs in one marathon session at the keyboard:

• Completely re-implemented chat/spectate system, so that it now works all the time for everyone. Seriously.
• Duel sometimes not showing up when you log in.
• Duel being squashed in the challenges window so you can't see what it is until you start it (amusing though this was.)

Just had my first session since these changes and have to say that so far so good with the chat, Charlie. Great stuff mate!

Some new statland crap:

Most darrenical 15 rounder (unique winner, woo!)
Most darrenical Speedgoats
Most darrenical Goatblitzen (I assume this is the plural of Goatblitz.)

The Duel is now resumable. Please, no more emails asking me to reset it for you.

Charlie Reams wrote:The Duel is now resumable. Please, no more emails asking me to reset it for you.

Phew, read this just in time.

Charlie Reams wrote:The Duel is now resumable.

Great news, yet another useful refinement!

Just a thought.
What would happen if your connection fooked up in the middle, but you didn't manage to get back on before midnight?
i.e. would it record your score so far, and zero rounds after, or as if you hadn't played at all?

Julie T wrote:What would happen if your connection fooked up in the middle, but you didn't manage to get back on before midnight?
i.e. would it record your score so far, and zero rounds after, or as if you hadn't played at all?

as if you hadn't played at all.
same thing that happens now if you start the duel but don't finish it before the midnight deadline.

New statland page: so that's his secret...

I think I see an opportunity to get myself onto a table.

Rob Thomas wrote:I think I see an opportunity to get myself onto a table.

Considering my wobbly ODE/CSW knowledge, I doubt I'll be gracing that table for a while.

I was thinking more about playing 200 games where the only words I suggest are things like "ME" and "THICK". That would get me 200 thrashings but top of the table right?

Can I just say to Charlie that I love the new tab in the game which pops up items [new] when you get a new item and then at the end you can check the tab and it itemises (excuse the pun) which round you got which items. Very neat mate, well done.

Kirk Bevins wrote:Can I just say to Charlie that I love the new tab in the game which pops up items [new] when you get a new item and then at the end you can check the tab and it itemises (excuse the pun) which round you got which items. Very neat mate, well done.

Yes, you can

I realise this is only vaguely on-topic, since it's not really a new feature, more the disappearance of an unwanted one, but I've been meaning to say for a few days how pleased I am that that annoying game sound that was triggered whenever anyone entered or left the room, or scratched themselves or farted (I can only assume that's the reason why sometimes late at night it was sounding more or less continuously) seems to have been dropped.

I like the sensory feedback of having sounds associated with events in the game I'm playing, and obviously it's useful to be alerted if someone challenges you; but it could get really distracting having that extraneous squeaking noise going on all the time in the background (I've no idea what it was meant to be, but it sounded like a chunk of expanded polystyrene being rubbed on a wet window, or possibly a raccoon being surprised sexually).

Good riddance say I!

It was my impression of a dog, or possibly a monkey, I'm not sure which. It was supposed to play whenever someone logged in, but 1) it was broken 2) it's not really worth the notification, given how many players there are now.

OK, so the "Triple Speed Remix" of the clock timer? Odd as hell

I'm loving the new colour coding and also loving the 5-second clock music (awesome) and loving the interactive nines in aptochat. Brilliant updates Charlie. Well done.

Kirk Bevins wrote:I'm loving the new colour coding and also loving the 5-second clock music (awesome) and loving the interactive nines in aptochat. Brilliant updates Charlie. Well done.

Thanks The colour coding was Matt's idea, it works very nicely.

Interactive nines really liven up the chat, it's great.

Charlie Reams wrote:Duel all-time points table

Good work. Not that I really looked at it. I just saw Bevan (first time I've been on for days) and I'm still laughing. Brilliant.

Matt Morrison wrote:

Charlie Reams wrote:Duel all-time points table

Good work. Not that I really looked at it. I just saw Bevan (first time I've been on for days) and I'm still laughing. Brilliant.

What joke did I miss?

Charlie Reams wrote:Duel all-time points table

Nice stats, thanks Charlie. Surprised I'm on that list though, I didn't think I'd entered enough duels to stand a chance.
Is there any chance you could give us a column of average points per duel entered, please? At least then someone has a chance of beating Kirk!

Neil Zussman wrote:

Charlie Reams wrote:Duel all-time points table

Nice stats, thanks Charlie. Surprised I'm on that list though, I didn't think I'd entered enough duels to stand a chance.
Is there any chance you could give us a column of average points per duel entered, please? At least then someone has a chance of beating Kirk!

Well average doesn't really bear any relevance since the scoring range of different duels varies widely, unless you meant a % of max score average or something.

Dinos Sfyris wrote:

Neil Zussman wrote:

Charlie Reams wrote:Duel all-time points table

Nice stats, thanks Charlie. Surprised I'm on that list though, I didn't think I'd entered enough duels to stand a chance.
Is there any chance you could give us a column of average points per duel entered, please? At least then someone has a chance of beating Kirk!

Well average doesn't really bear any relevance since the scoring range of different duels varies widely, unless you meant a % of max score average or something.

I didn't mean average points scored on the actual duel, cos obviously that varies a lot. I meant average duel points scored, if you see what I mean. As long as roughly the same number of people enter each duel, this will be a fair statistic (although I guess in ye olden days of apterous, there were less entrants, and hence lower scores. There must be a way to take this into account, but I can't think of one. Average number of people beaten on a duel, perhaps? I guess that would be a bitch to implement though. In fact average number of people who have beaten you on a duel would probably make more sense).

How about (total score on all duels taken)/(sum of maximum scores on all duels taken) ?

NB By "score" I mean the duel points rather than the score in the actual game.

I think Junior Mode is fantastic!

Thanks all involved.

Eoin Monaghan wrote:I think Junior Mode is fantastic!

Thanks all involved.

Kai Laddiman wrote:

Eoin Monaghan wrote:I think Junior Mode is fantastic!

Thanks all involved.

Just incase it wasn't just Charlie.

Junior mode is great, some of the conundrums are a bit obscure for juniors, though it's a good way of teaching kids new, crazy words.

Ben Hunter wrote:Junior mode is great, some of the conundrums are a bit obscure for juniors, though it's a good way of teaching kids new, crazy words.

Yeah, agree with that - clever idea but conundrums probably need some work. Not so much for obscurity but for bot fine-tuning - I played a couple a couple of Junior 9s against Prune and Nude and they were getting the conundrums in under a second.

Charlie Reams wrote:http://www.apterous.org/longstanders.php

LOL, my hyper nine record is pisspoor - someone beat it just to take the bad look off me.

David O'Donnell wrote:someone beat it just to take the bad look off me.

Due to English being a bit shit I can't tell if this is an imperative or a past tense, but in case you didn't pick up on this, the record was broken quite a while ago, but still held the top spot for a ridiculously long time.

Charlie Reams wrote:http://www.apterous.org/statland.php?se ... eatability

Unreal! The amount of stuff in Statland now is just insane. Cheers again for the continued amazingness, mate.

JimBentley wrote:

Charlie Reams wrote:http://www.apterous.org/statland.php?se ... eatability

Unreal! The amount of stuff in Statland now is just insane. Cheers again for the continued amazingness, mate.

Much appreciated, Jimbo. Here's some more:-

http://www.apterous.org/statland.php?se ... requencies

Surprising result in Unlimited, although the data are quite sparse. I'll try to play some 5-vowel rounds in the next few days.

Oooh! Ahhh!

Charlie Reams wrote:Oooh! Ahhh!

Why is AUTOTOXIC in the most improbable 9 list twice?

Charlie Reams wrote:Oooh! Ahhh!

Ian Volante wrote:

Charlie Reams wrote:Oooh! Ahhh!

Why is AUTOTOXIC in the most improbable 9 list twice?

Click the links and you'll see.

Charlie Reams wrote:

Ian Volante wrote:

Charlie Reams wrote:Oooh! Ahhh!

Why is AUTOTOXIC in the most improbable 9 list twice?

Click the links and you'll see.

Ah, I see, but it seems a bit pointless to have it in more than once.

I have to say I clicked the links when Ian first posted that and didn't work it out, or is it simply that it has been missed twice and that I'm trying to overanalyse it? In which case it's kinda crazy none of the others have been missed twice.

Matt Morrison wrote:I have to say I clicked the links when Ian first posted that and didn't work it out, or is it simply that it has been missed twice and that I'm trying to overanalyse it? In which case it's kinda crazy none of the others have been missed twice.

It's not that crazy when the definition is "super low-probability words". In fact I think it's rather remarkable that any of them have come up twice, which is my answer to Ian's point above.

File under "Why hadn't you done this before, you lazy bastard?"

Charlie Reams wrote:File under "Why hadn't you done this before, you lazy bastard?"

Potential add-on: sortable columns. Numbers doesn't change that much but conundrums count varies quite a bit.