Ways to improve Skirmish matching time through game modes.

from Demigod Forums

Despite everything it still takes a while to get into a match.

Technical difficulties aside, it simply is a problem with the lower userbase when compared to, say, Warcraft III.

So, when few users are online, it will take longer to get into a game, simply because the few users online ARE already in a game.

With more users online, there will not be a problem like that ... the more people are playing, the more games will end soon, freeing players for the next matchup.

 

This is a fact and a key to the solution: Shorten the length of a match when few people are online.

Stardock should have plenty of statistical data by now. Analyze them for the match conditions that produce short matches and emphasize those when creating new skirmish or pantheon games.

I don't have that data but I have my guesses:

* Fortress games are shorter than conquest games

* Domination games can be time controlled especially easily by tweaking the domination counter.

* Slaughter games with many people probably don't last long either.

 

And then there's always the option of making new game modes, like:

* Time Attack: Human Team must defeat AI team within a certain time limit. Double Favor for best player, half for the rest.

* Undefended: Conquest game, but no base defenses. Reinforcements spawn twice as fast, though.

 

Adds variety and decreases wait time by lowering match lenght.

 

 

 

833 views 5 replies
Reply #1 from Demigod Forums Top

People come and go into Pantheon/Skirmish games all the time.  Right now there might be 5 people online and in the queue but a minute from now that could jump to 50.  We make you wait so that we have the largest pool of people to work with when matching you up.

Bara

Reply #2 from Demigod Forums Top

the wait still is agonizing.

And I'm right when I'm saying that my proposal will fill your pool faster and more frequently, no ?

Offworld Trading Company
Reply #3 from Demigod Forums Top

When a match is being created, if the system notices a small pool of people and sets up a faster match, that means those people will get out faster and back into the queue.  Fine.  But as I said above, the amount of people online now will differ from who will be online a minute from now.  So in the above example, when those people finish their games they may end up in a pool with a lot of people, which doesn't really solve the problem.  The problem is that at specific times there aren't enough people.  Shortening the length of games will only help later on, not at those specific times.

Bara

Reply #4 from Demigod Forums Top

And that's the point where the statistics you gathered can come in.

Let's assume the time window which a player is willing to endure before giving up trying to connect is 5 minutes.

From the current state of population and the statistical data you have, you are able to predict with a certain chance how the pool will change within the, say, next 30 minutes. There will be high and low potential pool sizes. 

Then use the current pool to slot the players in a game mode that will fill the underpopulated future 5-minute-pools with a high probability, allowing for gaussian spill-over.

And you are right when you say that this strategy will only help later on and not at that specific time ... but once you get that running, "later on" will be "that specific time". 

This method strongly relies on the statistical evaluation of match conditions versus length, of course. So if that data is not conclusive, then it's a wasted effort. If they do, the game will feel like it has more players than it really does.

 

 

 

Reply #5 from Demigod Forums Top

That's a good idea, and may work better after the stat reset.  However, it may be a bit too much work for the gain.  I'll throw this by T-Man and see what he thinks later.  Thanks for the suggestion :)

Bara