Sunday, May 19, 2013

Variable Group Size

In my opinion, the next big challenge for PvE MMOs is creating interesting content for a group that varies in size. Right now, good PvE content is built for groups of a very specific size, be it 5, 10, or 25 people. As we all know from the past few years, this leads to all sorts of complications where groups have to maintain a bench. Where people have to sit out in order to guarantee that future attempts will have enough people.

I believe this is the one true strength of Eve Online. There is always room for one more in an Eve Online fleet. This makes participating much more fun. You always get to go with the group. You never have to sit on the bench.

The few attempts in the themepark MMOs to allow a variable group tend to fail for a few reasons. Very often the content just becomes a zerg, where you just throw bodies at it. If the content difficulty is fixed, the maximum group size becomes the "accepted" size. We saw this in Vanilla, where dungeons were capable of being done in a 5-man, but everyone insisted on taking 10 people.

Ideally, the size of the group would scale from n to 2n -1.  So if you wanted content for about 10 people, you could go with 7 to 13 people.  This allows the group to divide into two smaller groups that can still do the content.

There needs to be some difficulty scaling involved. If there is no scaling, everyone will insist on taking 13 people always.

But scaling things appropriately will be a significant challenge. I am not sure that a game with variable group sizes can offer such a tuned experience as WoW does.  However, variable group size would make organized grouping much easier, and smooth out a lot of the current social issues with grouping.


  1. Definitely agree with the above, though perhaps an easier thing for developers to do is to scale the loot instead in reverse proportion to the party size.

    This means you can get the best items if you finish an instance solo (haha good luck). Bring a full team and it will be easier, but the rewards will be muuuuch less rewarding.

    I'd think that would be easier than figuring out how to alter mob behaviour, stats, and having strong enough machines to handle varying mob numbers.

    Fortune favours the brave.

  2. It is really a strength of any open world pvp, group size just isn't as important. Even for large scale PvE encounters, if it is in the open world then group size is rather meaningless

    It ll only become a problem with the predominance of end game content being locked behind instances. Player creation tools like those of a couple SOE games could be a decent improvement as most can be tailored towards a specific group size. There is still a lot if restrictions placed on it but it certainly has potential for accommodating larger, and more mixed groups.

  3. How about filling the group with NPCs? It's not hard to program a DPS or healing bot (tank bot would be problematic)?

  4. You also have to remember that the reason it works in EVE is that it's all PvP. The "content" scales because it's directly affected by the number of people who attend. Though this also means that sometimes the content DOESN'T scale quite so well, because if one fleet can deploy a thousand ships, and the other can only possibly deploy a hundred, then it's certainly not fun for the smaller team. The same thing is true of non-organized/matchmade PvP in other games.

    Conversely, every other "normal" MMO focuses on content against AI opponents. On one hand, that actually makes scaling the direct numbers pretty mathematically easy, but you can't just scale the mechanics in the same way. That's why 10s can sometimes be a lot harder then 25s in WoW: Even if the mechanics are changed, or the numbers tuned, the margins are so slim that any mistake or death can be a failure.

    And that's not even counting the wide difference in fight mechanics in general; Patchwerk could easily scale infinitely to the number of people involved, but the Gunship Battle could be incredibly awkward. Vashj would be super difficult with low numbers, but trivial at large numbers (well, at least if she still worked like she did when she was active content).

  5. Gevlon's suggestion of NPCs is fine. I don't think a tank bot would be problematic at all. Simply have the tank charge at the nearest enemy, and cast AoE threat. Then randomly use abilities (including interrupts) until out of combat. If healer is attacked, taunt whoever is attacking it. If tank's health drops to 25% or less, he should yell "Healer, ffs". This is especially great if the healer is also a bot (healer bots should always say "taunt, ffs" when attacked by anything).

    This fight algorithm works fine for all dungeon bosses as well as trash.

    When no enemy is nearby, wander randomly through the instance until back in combat.

    Should a wipe occur, the tank bot should say "noobs" and then drop group.

    A whole party of bots could be more entertaining than the current "Heroic" dungeons in Pandaria (they were fun in 5.0, when we were appropriately geared, or even undergeared; but now we outgear them).

  6. Here's something that might have an impact on future instances: outfitting for twinks.

    Apparently, one of the big things in Mists is creating a guild group, dropping a person, and adding an L83 toon to the group. The four people --presumably geared up in the latest tier gear-- will steamroll through the instance, while the L83 toon gets all of the drops.

    If Blizz creates content for 5 players but allows up to 10 to join, I can easily see instances turning into merely "twink runs".

  7. Amen. Throw in variable (and selectable) difficulty modes plus downscaling/sikdekicking and maybe fill missing roles with NPCs (in line what Gevlon suggested) and we have a foundation for some casual group oriented gameplay.

    I can very well imagine WoW scenarios to be scalable in that matter. Modestly sized content (20 - 45 minutes to complete), paired with LFG system to fill missing spots if so desired.

    As for encounter scaling: there's a simple method to achieve it for groups of (let's say) four to 40 players without increasing number of enemies or even thier health, as long as developers don't implement specific enounter mechanic similar to Elegon. Tight tuning would probably be hard to achieve--especially in games with trinity, and I expect very challenging content (eg. raids) to be created for fixed group sizes.

  8. Given how much some players talk to each other in a heroic/dungeon/scenario these days, I'm not sure I'd notice a difference if I was playing with Bots.

    It would only be sad to see how many bots would top the DPS meters.

  9. Exactly, scaling is a *hard* problem, which is why no game has yet succeeded at it.

    But solving hard problems is the path to greater success.

  10. This is something where Firefall's doing quite well. The PvE content is openworld and sandboxy, and their single (so far) world boss is tuned for 25; if more people show up, he simply gets X% harder for each additional body above 25. So each additional person has to be able to pull their weight - which rules out twink runs and the like. It's a fairly simple boss, but the principle is sound for future, more complex encounters.

  11. Agreed, however I would add that actually even a minimum of 5 can be an issue. If we want to play with friends why should a set group size be a barrier regardless of how many people are online?

  12. I see another problem with that. If my raid doesn't need me, because they can easily run the content with 1 player less, I might as well do something else today because I don't feel like raiding today. The social pressure that you have to raid every week or your 9 friends can't raid might go away leaving us with less involvement in raids and their eventual death.