Print
All servers are not created equal. Some leave you feeling all warm-and-squishy after each match, while others insult humanity as a whole nearly as often as they insult your mother. Valve understands this and – with an eagerness to please its fans that’s borderline depressing (Just imagine: you’ll probably never be as devoted to anything as Valve is to you) – has braved the numerical gorilla pen that is mathematics in order to bring you a solution.
"After kicking around some proposals, we came up with a simple system built around the theory that player time on a server is a useful metric for how happy the player is with that server. It's game rules agnostic, and we can measure it on our steam backend entirely from steam client data, so servers can't interfere with it,” said Valve’s Robin Walker.
The finished product, then, operates on a point system -- sending well-behaved servers out for some time in the yard and booting rabble-rousers straight to the chair.
“In short, servers that have lots of players joining & leaving rapidly will score badly. Servers that consistently have players join and stay on for long periods of time will score well,” Walker explained.
“Our first step in improving this part of the player experience has been to delist all the really bad servers. The master server will simply stop giving these to you when you fire up the serverbrowser.”
“After that, we're going to keep improving our ability to measure this kind of problem.”
Comments are closed on this article
BroadSide
March 19, 2009 at 1:41pm
To be truly effective, they need to track 'Whining Adolescents'. I find nothing can effect my enjoyment of TF2 so adversely as a 'Whining adolescent'. Yes, it can be extremely satisfying to utterly dominate said whiners but the novelty wears thin very quickly. I would like a head's up on the level of 'whining' so I know which servers to avoid.
PhelanPKell
March 15, 2009 at 11:20am
My apologies in advance for any drawn-out points I make.
I think this may turn out to be a bigger job than Valve anticipates, but I will start by giving applause for their intent to please their user base.
There are some valid points made here that are really the tip of the iceberg.
Among some of the things Valve will have to factor into their numbers are:-User "ratings." As in, how good are the players on that server, compared to someone who shows up, dies in a hail of uber-death bits, and leaves.
-Ping/Latency/Location. This applies to all the factors of the network side of things. Some users disregard the ping to join a server anyways, only to leave or be kicked when they can't play or todays player base can't adjust for lag shooting. This also applies to circumstances with players who simply have crap connections, and leave a server they have a particularly bad connection with.
-Errors. This is everything from SteamID errors (even when the ID is legit), to some sorta crazy Windows glitch.For me, this last point hits home right now while I'm using my dying computer to play games until my new comp is fully built. I get random windows errors that lock up my game or system, sometimes requiring a restart. How would that effect the servers I enjoy playing on?
These are simply some of the thoughts I've had towards this idea, but I still feel that Valve should continue with implimenting this idea, and learn to tweak it as they go along.
Muerte
March 17, 2009 at 8:23am
None of those things will effect the scoring to make a significant difference. Good servers have people who come on and stay because they know the server is good. Most people trying new servers look at the type of gameplay then the ping so not too many will go to a server they know they will be at a disadvantage. The ones that do are few and far between aka newbs.
Now if they could measure the amount of time a person waits to get on a given server, they really would be able to separate the uber servers from the also rans.
PhelanPKell
March 17, 2009 at 8:55am
I certainly concede that you can't generalize the entire Steam community with the same statement(s).
But, there are still plenty of solid players that don't play professionally, or hold any one Steam game as their primary game to play, so they do look for random servers to play on.
Personally, I haven't done it in a while, but I've often done that in the past. The only reason I haven't done that lately, is because I prefer to play on the MaxPC server for TF2, and the GHP server for CS:S as they provide me with a solid challenge to test and improve my skills (despite getting owned about 55-60% of the time).
And as for ping, it can be a fickle biznitch. Good one sec, and bad the next. Given the any number of hundreds of causes of such a circumstance, it's not impossible to be an issue.
I just think that they're going to end up factoring in more stats than they expect.
Muerte
March 19, 2009 at 12:25pm
Yes, but once they find a server they will stay and play there until they get bored. And that's what they are measuring.
Once you stay for 15 minutes or more your actually recommending the server. Its only when you leave before that time that you have a detrimental effect to their score. You can go to dozens of different servers but once you settle on one you tend to stay.
If you move to a different server every 5 or 10 minutes well, your not really playing then are you?
DrCoolP
March 15, 2009 at 6:56am
There are some TF2 servers Igo on, and the quality of play is even too good for an average player like me to stay on, because no one likes getting pwned every 10 seconds.
Gailim
March 14, 2009 at 8:36pm
since my internet is fairly slow, I ussually just go with one of the servers with lower latency. and which server that is depends on when I decide to play. I dont frequent servers at all. According to Valve's new system I would be "downgrading" any server I join, becasue I'm probably not coming back regardless of my experience.
Muerte
March 17, 2009 at 8:27am
No, as long as you stay on the server for 15 minutes or more you are helping their score. How many different servers you patronize will have no effect unless you jump on and off very quickly which means your not really playing anyhow.
By the way this is a way people will be able tomanipulate the system so admins will be banning people who jump onand off servers more readily.
DOOMHAMMA
March 14, 2009 at 1:21pm
What stops people from the bad servers from just organizing and flooding the good servers with rapid login/disconnects?
shellpc
March 14, 2009 at 1:57pm
Every Steam account has a unique Steamid associated with it. Their system could check to see if its different accounts doing that or if its just the same set of accounts. Theoretically it might be possible to trick the system, but unless its easy to spoof steamids, it wouldn't be very practical, as each account would have to have bought a copy of TF2 on it.
CharlieGill
March 14, 2009 at 12:16pm
It is a good technique and has been for at least 40 years. In the early 70's, AT&T and the Bell operating companies used this technique to "test" their trunk lines. They simply monitored the call duration times. Noisy or distorted lines would be abandoned quickly, "hello..hello..can you hear me..click". Normal lines show a normal duration of call characteristic. You are using human responses to test a complex system.
Charlie
Nastyman
March 14, 2009 at 11:31am
Now this is a good idea as badly run servers will fall by the wayside and you won't waste your time on them.
Nasty
xveganx
March 14, 2009 at 12:52pm
Newbs servers are by far the best. I just wish they had a stock rotation server. Chumly knows whats up! [N] for life! :)
