Toribash
Original Post
matchmaking brainstorming
okay here goes matchmaking system i came up with just because why the fuck not

matchmaking table, which is used to find apropriate matches

id | playerStatus | nick | ELO_player | time
1 | LookingFor | rapist | 1605 | 13:58
2 | LookingFor | child | 1600 | 14:00
3 | Ranking | scrub | 1650 | 13:58
4 | Finished | noob | 1500 | 14:00


"there could just be an option to matchmake and an option to go to server browser. This way, the tourneys can still exist and be ranked. All matchmaking will be executed in official competitive mods that can be chosen as a filter."(c) Bercat

how? by adding new column with mod player selected and adding one more filter for mod selected


anyway table explanation:

playerStatus is a value that dictates the state of player, if he is looking for match or not
can be 3 values:
LookingFor - means player is looking for match
Ranking - means player is playing ranked match
Finished - means player finished his match or his search for match

this thing is called fl0w chart



okay, so whats going on here?

1) player click multiplayer, he is given option to go ranked or not. if he goes unranked he goes to server browser and just plays as usual
2) player select to play ranked, matchmaking is initialized
3) checking if the player have an entry in matchmaking table ( played ranked match before ) if he didn't, new entry created and neutral ELO is set, if entry exist, continue
4) playerStatus in player's matchmaking entry is changed to LookingFor, ELO_player is retrieved from database and added to entry, current timestamp added
5) checking entries in matchmaking table if there are other players LookingFor, if not found, ask to cancel search, change playerStatus to Finished to exclude his entry from search, if not, wait and check again
6) if players are found, compare ELO, by making and interval of ELO_player +/- 10, filter entries in matchmaking table that fit the interval, if not found, return to cycle and add +/-10 more, each time untill opponents are found
7) from players found select the one with oldest timestamp why? the key priority is speed of matchmaking, therefor it's a deciding criteria
8.) set both players playerStatus to Ranking to exclude their entries from search
9) actual gameplay, can be first to 3 session or just one game, after match is resolved, update each players ELO_player and set their playerStatus to Finished
10) start looking for another opponent
Last edited by snake; Aug 6, 2015 at 01:25 AM.
tell me about aikido
~referencing Dark Souls in suicidal threads since 13/01/15
Originally Posted by snake View Post
okay i will explain why it's bad and why i agree on hax statement

hax, my despise towards you is positively changing from "get ran over by truck" to "get ran over by car"

so instead get ran over by this little fella



back to the point:

we have players pool of 100 who are currently online ingame

we have fraction of players who want to play ranked, let's say 20

of these fraction we have a fraction of people who want to play grappling 14

and 4 to play striking

and 2 to play kicking


LoL have a huge player pool so thats a terrible example.

if toribash will ever grow to something remotely large and will stop resembling an university side project, you will be able to afford fragmentation of player pool by their specific choices, but you can't do it now, no matter how you summarize 12+6 12+2 6+2 you are still pretty low on numbers to think of that shit.

10 mods is a bit too much, but the short list of official cumpetative mods should be pretty decent, i would even say 3 mods from each category to begin with

again, i have to agree with hax as he surprisingly do understand why taking fraction from fraction of fraction of player pool is horrible idea for such small game as tb



i will bold the text so you will notice

THE MAIN PRIORITY OF MATCHMAKING IS TO FIND A MATCH UP

next priority is to match you up with player who have relative skill to you

and only after that your prefferences


don't be mistaken, we are still talking about cumpetative gameplay and cumpetative matchmaking, but these are the priorities which you need to recognise first, before "i want this, i want that", as they are fundamental


if you click "matchmaking" and wait for a hour to find an opponent - your matchmaking is a complete fail

when numbers grow - you can allow fragmentation, but when it's slim as fuck it's plain stupid.

We can always ignore their preferences on the back end if we can't find a match.
Originally Posted by hampa View Post
We can always ignore their preferences on the back end if we can't find a match.

a diagram on how you will choose to ignore or not please
tell me about aikido
~referencing Dark Souls in suicidal threads since 13/01/15
Originally Posted by snake View Post
a diagram on how you will choose to ignore or not please

A flow diagram, get it, because the flow clan. huehuehue


TEXT VERSION

Finding a match with preferences
...........I
...........I
...........I
Found a match with same elo within x minutes? ---------------------- yes -------------- ok good
...........I
...........I
...........I
..........NO
...........I
...........I
...........I
..Ignore preferences
Last edited by cowmeat; Aug 13, 2015 at 06:23 PM.
you forget to notify the client that matches in specific mod/category are not found and ask him to continue search or not, which he can continue to wait for waiting +x cycle time more again and again

if you don't ask him if he wants to continue to wait for specific prefference, meaning if not found it becomes irrelevant, you are basically begging for this block to be removed.




you can argue and debate what can be added, but i would like to remind that instead of adding useless stuff matchmaking should be focused on streamlining everything in order to minimize waiting for matchup.

if implemented streamlined system will prove to be reliable - it can be expanded later, but right now we are talking about adding lots on unnecessary stuff on top of something non-existant yet.
Last edited by snake; Aug 13, 2015 at 07:21 PM.
tell me about aikido
~referencing Dark Souls in suicidal threads since 13/01/15
I do agree with that last statement.

As I previously said:
"Functionality > preferences"

But yeah, that's just something id really like to see if the core matchmaking works well enough.
I don't have the dev knowledge to really help with the core system tho at this point anymore.
What is so hard to understand about, 5 mods which you don't get a say in what one you play, first to 3 duel and it automatically pairs you with the closest elo player online.

The logic behind this has been explained about 1000x times and surely you can work out this yourself Hampa. Nothing to even brain storm here, this is how every ranking system works so just do the fkin same.

In SC you DON'T pick what enemy race you want to face.

In hearthstone you DON'T pick what class you want to play against.

Why would it be different in toribash
Originally Posted by FistofLife View Post

In SC you DON'T pick what enemy race you want to face.

In hearthstone you DON'T pick what class you want to play against.

Why would it be different in toribash

In SC you dcide what race YOU play

in hearthstone you pick YOUR own deck

In the current idea the player has no say over his OWN mod. bad example.

The 5 mods idea is still fine, i have no problem with that.
Originally Posted by snake View Post
a diagram on how you will choose to ignore or not please

In the Wii matchmaking we scored each player. +1 for belt match etc. -1 if you just played the player.

Higher score the bigger change of getting match.
Originally Posted by cowmeat View Post
In SC you dcide what race YOU play

in hearthstone you pick YOUR own deck

In the current idea the player has no say over his OWN mod. bad example.

The 5 mods idea is still fine, i have no problem with that.


It's not a perfect example, but in all fairness you do get to pick what mods you're good at. So while you pick your own deck in Hearthstone, you also pick which mods you're good at in Toribash so that when it comes up in Matchmaking you can better guarantee victory. Or when you're called upon in clan league.

Just wanted to touch on practicality with a point you said earlier: "if Killer was the best lenshu player in Toribash I think he deserves praise". This goes back to the cons of banning mods. We wouldn't actually know who's the best in plenty of mods because there wouldn't be competition. The amount of people who take Judofrac seriously is less than 10. For TK? Less than 25. By forcing players to play mods they aren't comfortable at, it creates a larger talent pool which also creates more competition. It'll be easier for you to strut your stuff in TK if people are forced to actually play you in the mod.
Need help?
Creati0n says: still my favorite. <3
I sacrificed my firstborn for this great human being to join (M) ~R
Just Use Thunder!
Originally Posted by DrHax View Post
It's not a perfect example, but in all fairness you do get to pick what mods you're good at. So while you pick your own deck in Hearthstone, you also pick which mods you're good at in Toribash so that when it comes up in Matchmaking you can better guarantee victory. Or when you're called upon in clan league.

Just wanted to touch on practicality with a point you said earlier: "if Killer was the best lenshu player in Toribash I think he deserves praise". This goes back to the cons of banning mods. We wouldn't actually know who's the best in plenty of mods because there wouldn't be competition. The amount of people who take Judofrac seriously is less than 10. For TK? Less than 25. By forcing players to play mods they aren't comfortable at, it creates a larger talent pool which also creates more competition. It'll be easier for you to strut your stuff in TK if people are forced to actually play you in the mod.

To me its the matter of the modpool or to be exact, the size of it.

If the modpool is

Wushu, lenshu, erthtkv2, abd. Then im fine with having all of those. since 4-5 mods isnt too much to learn tbh. But with over 5 mods I feel banning would benefit the system.
Last edited by cowmeat; Aug 13, 2015 at 08:41 PM.