May 22
A Theory on the Quality of Grouping
I would make apologies for my absence, but I hope you understand.
Tobold has put up another thought-provoking post (man’s good at that) concerning the reward you get out of grouping and soloing.
So if people are trying to maximize the reward per session, they would prefer soloing if the solo session reward is bigger than the group session reward. But they would prefer to group if the group session reward is bigger than the solo session reward. As the term of the group finding time is always greater than zero, if the solo reward rate is equal than the group reward rate, the group session reward is diminished by the group finding time term, and people prefer soloing, because it gives greater rewards.
There’s a variable here he’s not considering, though, and I think it’s an important one.
He goes on to talk about improvements that could be made in World of Warcraft’s grouping mechanisms, and I wholeheartedly agree with them. I’ve had the chance to do a lot of LFGing in Ironforge lately, and to say it’s a crapshoot is a vast understatement. I live on an active server, so it’s actually not finding a group that I find frustrating. A variable Tobold doesn’t take into account is the quality of the group. I would actually amend his equation to this:
Group session reward = ((Group session time – Group finding time) x Group reward rate) x N
Here, ‘N’ represents something I’m thinking of as ‘group cohesion and experience factor’. I know this is a very squishy definition, but it’s definitely a factor to consider. A well balanced group that has good play habits will have an ‘N’ approaching 1. Exceptional groups could even exceed that, making the end rewards of time invested far outweigh the associated costs. Inexperienced (or, unfortunately, in many cases young players) will bring the N down towards 0, as will poor class balance within the party. I’ve participated in a group of all mages, and it was fun, but in higher-level situations it is just not a feasible party.
Defining N so low seems very harsh, but as any raider can tell you it only takes one inexperienced Warrior to wipe the whole party. Inexperience in the face of difficult odds more than just chips away at the value of a party, it effectively nullifies it.
The big problem here is that a potential party’s N is so hard to pin down. For example, I and three of my RL cohorts gathered together last Wednesday for a trip to UD Strat. We were okay, class-wise, with the exception of a dedicated healer. Our friend who plays a healer is doing a lot of commuting at the moment, and he was unavailable for our little jaunt. So, we put out the word via the LFG channel for a priest. We caught one, and after the usual bumbling around we headed in.
We started off quite inaspiciously, but after a little bit of tweaking to get our party dynamics working correctly, we were successfully taking on groups of three and even sometimes five critters without a character death. We managed Nerub’Enkan without incident, and were gathering our strength to take on Baroness Anastari when the priest sends a somewhat confusing tell to the party channel. “I forgot my work I need to go AFK for 15 minutes. brb.”
You already know the end of the tale, as he never returned and we misspent about 20 minutes waiting for him. We’re all friends, and have gotten really good at wasting time with each other, but it was still annoying. The N for that group seemed higher than it was than we started, but during the course of play it was revealed to be quite low.
At the end of the day, while I think teleporting group stones would make forming a group somewhat easier, it wouldn’t neccessarily make the group any more successful.
5 Comments so far


You didn’t finish the train of thought, perhaps because you figured it was obvious. But I think it’s worth finishing the thought:
If you’re looking to group with strangers, you can to assume that “N†is going to be lower than you’d like because some percentage of strangers suck. Exactly where you put your estimate of N will vary based on your experience, but you’d be a fool to assume it’s going to be close to 1. Conclusion: You need to set the bonus or group high enough that it counteracts the bad grouping experiences. To pull numbers out of a hat, if 50% of the groups an average player ends up grouping with are pretty good (close to 1) and 50% are bad (close to 0), the bonus for grouping needs to 100% to compensate. Anything less essentially rewards soloing. Yet if you announce a 100% bonus for grouping, you end up rewarding people with more reliable groups (rewarding people who play a lot and get into exclusive guilds); effectively penalizing the casual player who is stuck with random groups And as an added bonus, the solo fans will scream bloody murder because they perceive a 50% penalty on XP (which isn’t present in practice).
I’ve long preferred an eBay style voting system. Each character gets one. The people allowed to vote on them have spent, say, at least 30 minutes grouped with them in an adventure zone (for WoW it’d only be instances) and have some sort of good rating of their own (or some other factor).
Accountability. Without it no formula of sufficient complexity could be used as a prediction of a good session experience. In my opinion ;)
I agree, some sort of reputation system would help is getting higher average levels of N. I’d do something more intelligent than eBay’s system, however. MMOGs are fundamentally social activities. Take that social aspect into account. Ratings are easy to game (eBay at least charges you a little bit if you game the system by auctioning things to yourself). A cumulative rating is useful (it will weed out some people) but far more useful is a web of friends. This is Fred, who you don’t know. But you know Bob (and indeed rated him a 10 because he’s your real life friend and trust him). Bob rated Fred at 8. Since you trust Bob’s judgement, you can take Bob’s rating of 8 at face value. If necessary, go multiple hops. The more hops necessary, or the less you trust the judgement of your immediate hops (”Joe’s a great guy and I love playing with him, but he’s a terrible judge of character and gives everyone a 10″), the less reliable the result, but at least you can gauge the reliability.
Sounds very similiar to the idea I had for DDO if the game would of had real live Dungeon Masters. An eBay style system where you get to rank your DM and the DM ranks the players. That way the chumps get identified.
The only problem is people grouping together and rating each other as good just to make their ranking better. Pretty big hurdle to clear.