Multiplayer:filtering unresponsive players badge IMPLEMENTED

[RiverRaider 1097]

I like the badge icon spot, and the gem count is nice but badge letters needs to be as big as game list letters maybe. Also icon a little bigger )

[benedictarnold]

I noticed my rank was demoted, but it wasn't due to inactivity on my part.

I wonder if the game notification bug is causing inadvertent skips? But even then, I think I would have known since I manually refresh every so often too.

My current badge is the 2nd from the top, but I think it should probably be the top one. Can Daniel check the logs and see if I missed a turn that I'm unaware of?

If not, no worries. 2nd highest badge is good enough I just want to get to the bottom of this.

It's a fun game and I want to help work out the bugs so everyone has something rock solid to play with.

[Stratego (dev)]

i will implement a form where anyone can see how its badge is calculated, and from what data source (skip count, games count)

[Stratego (dev)]

ok, in dev version there is a button in the badge description page to query your badge data.
(multiplayer section/click on badge on the top of the list/hit badge data button)

up on server in 10 minutes.

[RiverRaider 1097]

I am losing badge position because of playing large map games with six players that just don't get done in time, could badge conversion be tweaked to accommodate this style of play?

[Stratego (dev)]

yes, any idea how? (how to modify the current calculation)

dont forget i dont want to give good badge to someone who is creating eg. 10 games and those games are left there unplayed for months.

[RiverRaider 1097]

I just lost another badge that's two in a week? This isn't right the usual 2 player 20x20 map wouldn't even be done in time at this rate?

[COOLguy]

Back down to rookie.
I am pretty good about taking my turns, but I do only have one game going.

[Hardeep]

Back down to rookie.
I am pretty good about taking my turns, but I do only have one game going.

You are what you are

[RiverRaider 1097]

RRrrr...put me sea boots back on and scuttlebutt gems from thirty day wonders!...but first another trip to oak island to unload some booty RRrrr )

[Stratego (dev)]

I just lost another badge that's two in a week? This isn't right the usual 2 player 20x20 map wouldn't even be done in time at this rate?

ok, i am open to suggestions:
(btw: if you had chance to lose badge that means that earlier you had the required finished game per week ratio - so you play longer ones now then earlier)

[RiverRaider 1097]

I respectively disagree, 90% of my games have always been 6 plyr big map games I believe badge system started about the time some of these games was rapping up. Keep in mind I have 20 to 30 such games being played daily...I work when I have time lol )

[thedidey]

Moved this here at Dan's request. I will look into getting dev version and checking out the formula to suggest specific changes.

Happy to start a new thread on this if this isn't suitably inline with the thread topic. I would also prefer a change to the way badges are calculated. I feel like we are being penalized when the rating drops. It's telling people that i'm not reliable and potentially limiting the games I can play. I have been talking with the Dev and you are demoted if you don't finish a certain number of games a week. I have 13 games going and they've all been going longer than a week. I generally take my turns within an hour or two. I primarily play 4 to 6 person games. Seems like a bummer that i'm being demoted for not finishing games when i'm pretty reliable.

That said. I might understand if finishing games was a requirement for the top few tiers. But I can't get past rookie because I'm not finishing games. and I'm not finishing games because I play big long games.

[thedidey]

So I think that finished games is a bad measure of an active player.

A player could play lots of short games and get skipped a lot while still achieving a decent rating.
For Example a player who finishes 20 games in 2 months with three skips has a rating of:
( 20 – 3*5)/15 = 0.3333 or one above rookie.
A guy who has finished 5 games with 0 skips has the same rating.
(5-0*5)15.

So I was playing around with this idea. I don't want to take it too far unless people think it's a good idea and Dan thinks it could be implemented. I went with this formula

[(games finished X multiplier)+(value/Average time to take turn) - (skips X multiplier)]/ (days/4)

This basically takes the original equation, adjusts the multipliers and adds in a variable for average time to take turn (IF that can be calculated by the game). Here are some results.

As you can see. A player who takes 2 hours average to take a turn is decent. Players who get skipped once may be ok if they play fast other wise but are duds after getting skipped 2 or 3 times. Player who finish 5 games in 2 month are decent and 20 games are super studs. The key is to balance the multipliers based on what people think is acceptable. Maybe 20 games isn't that many. Maybe 2 hours isn't that fast. Love to hear your thoughts. and if it's even implementable. If it is, I'd be happy to help tweak it.


Thanks.

[RiverRaider 1097]

Just happened again, level 2 from 5 in just a few days COOLguy was right rookie here I come, Why can't we simply let the game engine do the kicking after the second skip, nobody is blamed or feels guilty this way? Plus keeps a simple Talley of kicked players and demote them accordingly. Isn't this basically what we was after anyway getting rid of the abandoned download gamers?

[Stratego (dev)]

@riverraider: how can u drop back to level 2- i am playing only 3 games mostly with you that lasts for weeks, and nothing more, and i am only down in the second best badge.

what is the number of your finished games and skips? (mine is 9/0 currently)

(you can see it by clicking onto your badge, and then the "badge data" button.)

Average time to take turn

currently i have no such information without more logging implementation.
i have these
- number of games and their statuses
- number of skips/kicks of all games
- last modification on game (mostly: last player turn-taking)

if we could use these for a formula, that would cost very few time on my side.

my idea was modifying the current formula to add to the "finished" games count the ones where
- last modification (turntaking) was within 2 days.
- and the game started 1 weeks ago or more.
and lets call them: "long but running games"

so the current formula is:

Code: Select all

(finished games - skips)/10 = value
>1 : top badge
1 > value > 0.8 : 2nd badge
0.8 > value > 0.6
0.6 > value > 0.4
0.4 > value > 0.2
0.2 > value > 0
0 > value > -0.33
-0.33 > value > -0.66
value smaller than -0.66 gets the worst badge

and i planned to modify to this:

Code: Select all

(long but running games + finished games - skips)/10 = value
(the result is "badged" the same as above)

[Hardeep]

Just happened again, level 2 from 5 in just a few days COOLguy was right rookie here I come, Why can't we simply let the game engine do the kicking after the second skip, nobody is blamed or feels guilty this way? Plus keeps a simple Talley of kicked players and demote them accordingly. Isn't this basically what we was after anyway getting rid of the abandoned download gamers?

I dont think that it is to bad to skip/kick on your own, I've done it plenty of times with no remorse whatsoever.
It is great to sympathize, but it's war, so everything is automatically made right and moral

[thedidey]

If long running games are being counted i think i'd be happy and actually be able to get ranks.

My other thought that I forget to say was that maybe 60 days is too long. But it seems like that interval is less in the new formula so that's cool too.

thanks Dan.

[benedictarnold]

I think 1 week is not a long running game. All of my games are over a week. Hell, some games I only get about one turn per week since everyone is so slow.

I also have some legacy games that I gave up on since everyone left and its just me vs the AI. Those games should be excluded as well.

I think inactivity should go on skips and kicks only. Long running games are just a normal part of the way this game is set up.

[Stratego (dev)]

maybe 60 days is too long. But it seems like that interval is less in the new formula so that's cool too.

why? also in new formula it is still 60 days that is inspected counting the "finished games".

I think 1 week is not a long running game. All of my games are over a week. Hell, some games I only get about one turn per week since everyone is so slow.

yes, lets call it "not a fake game" meaning not a game that someone made to instantly inprove its badge - a 1 week game is mostly a real game

I also have some legacy games that I gave up on since everyone left and its just me vs the AI. Those games should be excluded as well.

you should do not leave those, if you dont like the "turn to AI" mechanism, there is a thread about it, i already suggested a different mechanism.

I think inactivity should go on skips and kicks only. Long running games are just a normal part of the way this game is set up.

[/quote]
disagree: a player with zero skips (rookie) should not join to a master game with high badges.

[benedictarnold]

If you just want to track activity then you could use win or loss as a Boolean flag that excludes the group of people that just start games but never finish them.

The arbitrary 'game duration limit' is not a good idea. First, it is completely valid to have a very long game and users shouldn't have to worry about demotions just because they're playing a long game.

Second, the distributed turn based nature of this game means that games will take a long time to complete.

It is unavoidable and it is valid gameplay. For those two reasons I think it should not be an indicator for inactivity.

Lastly, what was your recommendation for the AI? I don't know what thread you are referring to, but I'm happy to do it.

[TheBluePhoenix]

Can someone list all the problems in short which we are facing with the present formula? I think i can devise a good way for the badge system.as far as i feel
The formula should make way for all types of games
It should just remove ppl who dont play after joining and not those who play seldomly but with full devotion( me for eg)
It should include the skips and kicks not wins or loses
It should have some concession(not very large) for people who play many games .
Anyone has any more problems? If yes pls post
Also before formulating, i wanted to ask dev whether the no. Of turns a game got over or the area of the map can be possibly included in the formula

[benedictarnold]

Heres a problem. I play games every day, but only like 4 or 5. These take a long time to complete. Sometimes multiple months. By using the formula now I'm penalized by finishing less than 10 games a month. This doesn't track my likelihood to be responsive.

[COOLguy]

I think we might be making this more complicated than it should be. Here's why:

-What we are measuring is just how active a player is
-Meaning how often he takes his turns
-So a simpler formula could be "turns taken in a 30 day period minus a penalty for skips"

On a scale it would look like
0-30 (one turn a day on average) for rookie
31-60 (two turns a day on average) for level 1 reliability
and so on

The penalty could be -10 for skips and -20 for kicks.
So basically one game kicked is worth an average 1 turn a day.

How does that sound?

[benedictarnold]

I think that could work better.

[COOLguy]

Average time to take turn

currently i have no such information without more logging implementation.
i have these
- number of games and their statuses
- number of skips/kicks of all games
- last modification on game (mostly: last player turn-taking)

if we could use these for a formula, that would cost very few time on my side.

my idea was modifying the current formula to add to the "finished" games count the ones where
- last modification (turntaking) was within 2 days.
- and the game started 1 weeks ago or more.
and lets call them: "long but running games"

so the current formula is:

Code: Select all

(finished games - skips)/10 = value
>1 : top badge
1 > value > 0.8 : 2nd badge
0.8 > value > 0.6
0.6 > value > 0.4
0.4 > value > 0.2
0.2 > value > 0
0 > value > -0.33
-0.33 > value > -0.66
value smaller than -0.66 gets the worst badge

and i planned to modify to this:

Code: Select all

(long but running games + finished games - skips)/10 = value
(the result is "badged" the same as above)

Actually this is best here - it will not require much new implementation. It's almost the same as the turn counting, but won't need any more information log to be implemented.

[benedictarnold]

Average time to take turn

currently i have no such information without more logging implementation.
i have these
- number of games and their statuses
- number of skips/kicks of all games
- last modification on game (mostly: last player turn-taking)

if we could use these for a formula, that would cost very few time on my side.

my idea was modifying the current formula to add to the "finished" games count the ones where
- last modification (turntaking) was within 2 days.
- and the game started 1 weeks ago or more.
and lets call them: "long but running games"

so the current formula is:

Code: Select all

(finished games - skips)/10 = value
>1 : top badge
1 > value > 0.8 : 2nd badge
0.8 > value > 0.6
0.6 > value > 0.4
0.4 > value > 0.2
0.2 > value > 0
0 > value > -0.33
-0.33 > value > -0.66
value smaller than -0.66 gets the worst badge

and i planned to modify to this:

Code: Select all

(long but running games + finished games - skips)/10 = value
(the result is "badged" the same as above)

Actually this is best here - it will not require much new implementation. It's almost the same as the turn counting, but won't need any more information log to be implemented.

It looks like there are two general problems. The first problem: people join multiplayer games and aren't responsive. They have to be skipped and kicked. This damages the experience for the user.

The quoted bit above is how the game works now, and it seems better than it originally was, but still seems off. For example, I would need to play 10 games at once to receive the highest badge. Is the badge supposed to track my likelihood to respond and not need to be skipped/kicked (i.e., solving the main problem)? If so, why require so much activity? I don't think that knowing things like speed of response or number of games user is playing at a given time is really solving the problem. You have a 24-hr clock to respond, some people are going to use 5 min, and some people are asleep for most of their turn, wake up, go to work, and have a 10 min period during lunch where they respond. Both of those players are being responsive, yet the second guy is being penalized.

The second problem: experienced and skillful players are hard to find. Often games are very unevenly matched. Experienced and skillful players will get bored, inexperienced and new players will get frustrated when they lose too fast. This also damages the experience, but to a lesser degree.

The badge system is trying to solve the first problem. It's super close, and I love that you guys are implementing it since it solves a really huge annoyance in the game. My only problem with the current solution is that it has this seemingly arbitrary "time" constraint that seems to be trying to figure out how quickly a user responds within that 24-hr window.

How about giving users a bucket of 100 pts. They start with 80. They can't go above 100 or below 0. Each skip is -10. Each kick is -20. Each complete game is +5. This would quickly separate players that are likely to be skipped form those that are going to respond.

Anyway, just my 2 cents. Thanks!

[RiverRaider 1097]

Best two cents I've seen, Thanks for your time and input Ben . Bucket list sounds good to me as well short and to the point )

[COOLguy]

The quoted bit above is how the game works now,

No actually that is Daniel's proposed fix. Currently, the game doesn't take into effect long games in progress only finished ones.

For example, I would need to play 10 games at once to receive the highest badge. Is the badge supposed to track my likelihood to respond and not need to be skipped/kicked (i.e., solving the main problem)? If so, why require so much activity? I don't think that knowing things like speed of response or number of games user is playing at a given time is really solving the problem.

The badge primarily tracks how active a player is - not if he is likely to be skipped. However, an active player IS unlikely to be skipped.

I am most definitely one of the 10 minute lunch break people. However I don't get skipped that often. I generally have 1 or 2 games going at all times and I take my turns within 24 hours (discounting special circumstances). Therefore I should not get the highest activity badge, but I should be in the mid-pack badges somewhere.

My only problem with the current solution is that it has this seemingly arbitrary "time" constraint that seems to be trying to figure out how quickly a user responds within that 24-hr window.

There is no time constraint to figure out how quickly a person takes his turn. To do that would require Daniel to implement extra things.

How about giving users a bucket of 100 pts. They start with 80. They can't go above 100 or below 0. Each skip is -10. Each kick is -20. Each complete game is +5. This would quickly separate players that are likely to be skipped form those that are going to respond.

That was discussed earlier in the thread. The reason that was not the best way was because a player could be 'active' for a week or so, and then go for years without playing a game. By that score he would be considered "active" although he hasn't been active recently.

Activity is measured by how many turns on average a player takes. Measuring that would require Daniel to implement new things that he does not have time right now currently to do. So the next best thing in my opinion is Daniel's suggestion that I reposted above.

[benedictarnold]

The quoted bit above is how the game works now,

No actually that is Daniel's proposed fix. Currently, the game doesn't take into effect long games in progress only finished ones.

Actually, when I look at the information on how badges are calculated, the formula you quote is the one I see in the game. So I think that the proposed fix is in production right now.

For example, I would need to play 10 games at once to receive the highest badge. Is the badge supposed to track my likelihood to respond and not need to be skipped/kicked (i.e., solving the main problem)? If so, why require so much activity? I don't think that knowing things like speed of response or number of games user is playing at a given time is really solving the problem.

The badge primarily tracks how active a player is - not if he is likely to be skipped. However, an active player IS unlikely to be skipped.

I am most definitely one of the 10 minute lunch break people. However I don't get skipped that often. I generally have 1 or 2 games going at all times and I take my turns within 24 hours (discounting special circumstances). Therefore I should not get the highest activity badge, but I should be in the mid-pack badges somewhere.

Would you agree that this is no longer trying to solve ONLY this one problem (the problem of if someone is going to be skipped or kicked)? It's trying to determine which users are likely to respond AND determine which users have a bunch of games going on at once, or play a huge number of games that end quickly. My point is that second part is unnecessary. I think the main problem to be solved is reliability not frequency of game-play.

My only problem with the current solution is that it has this seemingly arbitrary "time" constraint that seems to be trying to figure out how quickly a user responds within that 24-hr window.

There is no time constraint to figure out how quickly a person takes his turn. To do that would require Daniel to implement extra things.

I think all I meant by this was when he divides (finished-skipped) by 10, that is saying that every 2 months you should be finishing at least 10 games. That's a time constraint that I don't think adds much.

How about giving users a bucket of 100 pts. They start with 80. They can't go above 100 or below 0. Each skip is -10. Each kick is -20. Each complete game is +5. This would quickly separate players that are likely to be skipped form those that are going to respond.

That was discussed earlier in the thread. The reason that was not the best way was because a player could be 'active' for a week or so, and then go for years without playing a game. By that score he would be considered "active" although he hasn't been active recently.

What's wrong with that? Games are only started after everyone joins. The filter would exclude people who are unreliable from ever joining. If someone is inactive for 2 years, but then decides to start playing again, I don't see why you should throw away the information you've learned about that user from the past. He should be able to join.

I'll get on a soapbox for a sec. One of the things I love about this game is that you can pick it up and put it down without worrying about it breaking your day apart. It's a mobile app game. I play it on my phone whenever I have a few minutes of downtime. I like being able to turn it off, save it,do something else, and come back without it interrupting my day.

By attaching a 'frequency of game-play' requirement to badges, the game is no longer structured to revolve around my schedule. It's going to require you to immediately drop what you're doing and play the game as soon as it's your turn. This means I have to schedule the rest of my life around this game. Not something I want to do.

Thanks!