Matchmaking. Does this game need it? I think so.

Public games way too often end in a matter of minutes due to the teams being a total mismatch. This isn't fun for anyone, even the winners.

If I were on the dev/update team, I would propose an algorithm that takes into account:

• Level
• Play time in last 2 weeks
• Stats in recent games
• Number of players on teams

Heck, I could write this myself if I had the chance to. Opinions?

Are you gonna donate the necessary servers?

I'm level 40 and shockingly bad.
Would this system put me on a team of people who are awesome?

The processing power of matchmaking for a single match is probably on par with shooting a single bullet, so yes, I would gladly donate a single server from the 90's.

As a programmer, I believe the decision to implement or not implement is more about developmental resources (people, time) than the resource overhead.

Me too, lol. I would think this can be accounted for.

Getting away from the peer-hosted matchmaking of the 360 version is one of the things I like about the PC version

or he could donate the algorithm, and devs could make it an option that can be turned on or off on existing servers.

I will 2nd this.
Peer hosting on the 360 was horrendous. It's the major factor of why I quite playing the game completely. It would take 10m to get into a decent game only to have the host drop with 2m left because his team was getting stuffed.

He's not talking about Peer2Peer matchmaking. He means that when the server shuffles the players into two teams it can be VERY uneven, resulting in steamrolling or being steamrolled. Neither of which is very fun.

I think it simply needs to balance teams by money earned. Each player will have a lifetime money earned. If you total each players lifetime earnings on each team they should equal roughly the same.

E.g.

BLUE:
Level 60 player = $600,000 lifetime earnings
Level 15 Player = $50,000 lifetime earnings
Level 23 Player = $200,000 lifetime earnings
TOTAL: $850,000

ORANGE:
Level 51 Player = $300,000 lifetime earnings
Level 21 Player = $250,000 lifetime earnings
Level 67 Player = $300,000 lifetime earnings
TOTAL = $850,000

In this case the teams would be even. Even though blue team appears to have more experienced players, the Level 67 player doesn't have much lifetime earnings so you can conclude they are a bad player or non-team player.

The only problem is those who like to play together. If 2 or 3 high level players pick the same team then the system would have to do its best to even it out. Perhaps even giving one team more players if its that uneven.

Wait, I thought lifetime earnings were tied to your level. Because if I'm not having a sudden amnesia moment here, it's impossible to be level 67 and have $300,000 while someone else is level 60 and has twice your earnings.

And lifetime earnings carry over from Blitz and vice-versa. So a seemingly "skilled" player can just play Blitz more than Crossfire.

You guys haven't thought about one thing: What if a player decides to play a class he is not good at? I'm a decent gunner and assault but I doubt I could pull my weight as assassin atm. What if I try to get accustomed to her? Now all my stats indicate that I know what I'm doing when I actually don't. And class selection is AFTER team selection. How is the algorithm supposed to know if I chose one of my strong classes or not?

Server factors come next: Server with or without juice and overtime. A player who is not used to defending against juiced opponents will have a hard time if he suddenly is thrown into a server with a juice chaining assassin.

Unless you know all the players in a match well and can make an educated guess at their performance it is very hard to perfectly balance teams. However, when you look at the large picture it evens out somewhat. Without teamstacking the auto-matchmaking is actually pretty decent all variables considered. I rarely get into games where the match is decided in the first minutes.

Matchmaking will also get better over time. Right now there are a lot of new players and not much information on their actual skill level.

Are you trying to make a case for why random matchups would necessarily be better than algorithmic matchups?

I understand what you are saying, but nothing of what you said disproves the concept of better algorithmic matchmaking, it simply states that it couldn't be perfect -- which is generally accepted and acceptable.

My life-time earnings were completely made up so don't expect them to be even close to reality. I'm fairly sure lifetime earnings are not tied to your level, but are the cumulative sum of all the money you have made every match.

I've never played Blitz, but if they do carry over then this theory is blown out of the water, or perhaps there is a way that the system can subtract any money they've earned during Blitz.

Yeah I think the whole idea is to make matchmaking possible, not required/automatic.

It would be a lot of fun to get a team of 6 together and do some skirmishes. Not everyone wants to necessarily join a clan, but they may want to get together with a group of friends and play against another organized group.

I think it would be great. The servers still get to do what they're currently doing, but you could group up in various numbers and go against other groups.

This could require some servers on Uber's part, or just enough interest in dedicated servers that are specifically matchmaking servers (maybe a setting where a server only hosts 6v6 matchmaking matches or something).

As for determining skill, that shouldn't be too hard. Aside from money/level, the game also keeps track of your kills, per class and with each type of weapon, and all sorts of other stats that could be used.

Edit: I was mainly getting excited at the idea of group-based matchmaking, but it could work just as well in a "pub" matchmaking sense where it tries to find a server with people around the same general skill level.

Again, there's no perfect way to measure skill, but with all the stats of time played, kills, deaths, money, turret kills, assists, etc, it can surely be done decently.

I'm pretty sure Blitz does carry over and despite being made up numbers, a level 60 player should not have more money then a level 67.

I mean, I know I'm speaking from the console, but it's still Monday Night Combat we're talking about. It's still the same basic game, just rebalanced classes.

I personally side with Col Jessup asking if it's even possible considering variables such as actual player skill and different class choices.

But Decker is right. It will never be perfect, we know that. But surely some sort of algorithm is better than NO algorithm? Even if it simply sorted by level would be better than it is now. 1 in 3 games I find that one team ends up stacked, even when no-one has chosen a team.

I assume currently it is completely random?

I don't thinks it's completely random. It does a pretty good job too if people don't start stacking teams. Could it be improved? Yup, sometimes it creates some mismatches. But it can never be perfect. The longer people play the better it will get.

Meh. Fair enough.

Probably.

Actually, right now, aside from people choosing red or blue...it IS completely random. Why do you think it will get better with time? Because people will play longer and we'll all be experts?

Also, here's food for thought: Programmers use computers to implement logic that humans use. The mere fact that we are able to recognize a mismatch before the match starts (i.e., red team all level 50-60 and blue team all 5-15) implies that there IS some room for improvement.

K, I'll give you the whole story. Sit down, it might take a while.

Back at the beginning of the beta there was a matchmaking algorithm. It didn't work very well. Some players ended up on the Icemen team 90% of the time and the Hotshots always got the additional player when the number of players was not even. Then a lot of players left the match when their team was losing. Icemen were pretty much shafted. Search for a topic called: Icemen Blues.

I talked to Scathis and he said he would try to balance teams a different way. First thing: Hotshots didn't always get the new players. And I think the whole thing was adjusted around the level of the players. There were many complaints and ragequits in the beta because one team might get more high level players. That looked imbalanced. But what if one of those high level players just played a load of Blitz before his first Crossfire match? Still, it was already a much better system than the first one.

I have actually no idea what system is currently used, but I can tell you that it seems to work pretty well once you have players with a higher level. The reason is simple: A level 10 player might have played in the beta and might have a huge advantage over another level 10 player who has played WoW for the last 25 years. Once they both reach level 50 they will usually be on a more even skill level and the matchmaking works better.

An algorithm needs data to work. The more data you have, the better your results. Right now we have a large population of new players and the whole thing is really more of a guessing game. Plenty of players will play with a class they are not good at to gain experience. And there are plenty of other factors as well.

I'll try to find out how matchmaking works, but its 3:30 am in Kirkland and there is nobody awake to ask atm.

Anyway, my advice is: Play more, gain experience and learn to push your bots. If you have enough experience, you can turn matches around that don't look favorable. Sometimes you will get stomped, sometimes you can stomp the other team but most of the time it will be even enough for you to make a difference.