Game Duration Filter

[knucracker]

I have enabled a "Game Duration" filter on the custom maps page.  It is above the map list and not very noticeable right now.  If it works properly, I will make it a little more prominent over the weekend.

Here is how it works:
Trivial Duration:  Can be completed in < 3 minutes
Short Duration:  Can be completed in  < 15 minutes but greater than 3 minutes
Medium Duration: Can be completed in < 30 minutes but greater than 15 minutes
All: Every map (ones with no scores and that take more than 30 minutes).

I look at the best score posted on a map to determine the duration.

Note that a I may tweak these minute values and add more granularity and/or change their names to be something like:
Duration: < 3Min   < 15Min  < 30Min   All

[Karsten75]

Nooo!  Not best score!  That means UpperKees, siccles and other madmen! I htink you should take like the 5th best score or something lower, like perhaps the average score of all who submitted times.

[UpperKEES]

It will take quite a while before all 750+ maps will have 5 scores posted. No worries, it's only an indicator how fast is possible. We all know you can do 100 meters by foot in about 9.8 seconds, but I'm very happy with the 13 seconds it takes me. It just makes me aware that I don't have to aim for improving myself by more than 3 seconds.... (although you should always try! ;)).

I like the filter, but I still miss the sort options as discussed in another subject. With soon more than 1,000 maps I would like to be able to 'search' in a few more different ways.

With regard to this filter: I would change the name 'All' to 'Long/unranked', or even better: make two categories out of them. 'All' is misleading, as it does not include the maps from the first three categories.

[SPIFFEN]

I like this , but i agree with Karsten75 .
It should be an average of the top 10 scores or so .

But it's on an way to the right direction ,
so maps might get the right lvl of how hard they are =)

[hi]

mh... I like the idea... But I hate the times you chose... imo medium should start at a time like 7-10 mins as 3 and 15 mins are such a great difference...
I hate the idea with the average and not the best score... (because the skill level of different players differs soo much)

I'd also like to have a search that sorts the maps on the number of scores that have been submitted.

[Sqaz]

Is it normal that when you press trivial, short or medium and you want to switch to another page on that filter, the filter goes to that page on All :-\

[knucracker]

Nope, that is a bug...
Should be fixed now.

[TheBuilder]

Hey virgil, how about u make it so that it categorizes them by their total average score, that is, every score posted for that map, divided by how many scores there are, that way, it could be reasonable .

And by the way, i used ur "get score from time" formula to figure out the "get time with score" formula, here it is:

          (10,000*3,600)/score-3,600=t          t=time in seconds

[knucracker]

I've debated the average/median/max/min/subset max/subset min thing a good bit.  All of them can make sense depending on what the goal is.  The original goal of the filter is to allow a player to find a map that fits the time frame they are wanting to play.  For instance, sometimes I have 15 minutes that I'll set aside to clear my mind.  In this case, I don't want to play a map that is going to take 30 minutes, etc.

So now come the debate of how to determine how along a map takes to complete. The only real data available are the newly added custom scores.  Some maps have no scores, some a few, and some more than 10.  I anticipate that the number of scores will go up over time so that perhaps eventually every map that can have a score has 1 or 100+.

So average is something I can do.  However, it only takes one young kid to spend 30 minutes on what is otherwise a trivial map and the average time goes way up.  I can mitigate this by throwing out the highest and lowest N scores.  But this can only be done when the total number of scores is more than a few.

If I do something more than the best score it has to work for small numbers of scores and also be tolerant of extremely low scores that could be posted by really inexperienced players (or jokers).

Note that tweaking the time ranges also has an interesting effect.  For instance say I had the following ranges:
0-3 min
3-10 min
10-20 min
20-40 min
40-60 min
60+ min

Really good players might be able to do a map in 7 minutes whereas the average player might take 13 minutes.  With these ranges it isn't the end of the world that a map would show up in the 3-10 range but when an average joe plays it takes them just over 10 minutes.  In other words, it might help to just provide a few more ranges and tighten them up a bit.

[bobandirus]

why not bundle together the mean and the standerd deviation equations to get the maps 'playability' rating. Or something along that sort of line.

I.E:

Rating= ((score 1 + score 2 + ..... + score N)/N)/(sqroot((((score 1)^2)+...+((score N)^2))/N)-(((score 1 + score 2 + ..... + score N)/N)^2)))

That should give you a value for the average time taken divided by a value for how varied the results are. The closer the value is to the mean, the closer the mean is to the correct rating.

With a bit of jiggery-pokery, you could end up with a good estimate at how long most people will take to do the map. Eg -  ((rating/mean)*mean)=Map estimate

... I think :P



You might want to check on the accuracy of the equations. I apologize for my love of brackets.

[Karsten75]

I think you can do this (perhaps without too much difficulty). For 1 score, that is the map time. For 2 scores, avg the scores. For 3 or more scores, check how many of the scores fall into which bracket. That is the bracket for the game.

It would be really interesting to use something like the ELO rating system for chess, where a player's capability is factored into the score. But frankly, I'm not sure how to suggest this except as a vague idea.