DPS Queue Times

[Tridus]

You know if you ever wanna go DPS main I will keep you healer-queued up so you aren't waiting :P

And lose our regular game of "who is going to get stuck DPSing this expert?"

That aside... no? I'll elaborate. I don't think I spent enough time on my prior post. When I say 'variety' it isn't just that there are more dps but rather the actual variety offered by dps gameplay.

The variety of dps roles isn't just that there are nine of them instead of 3 tanks and 3 healers it's that they all offer different gameplay. They've got definition to them that you just don't find with tanks and healers. All tanks have a threat combo, an AoE enmity generator on GCD, a ST ranged enmity generator, an immunity, tank stance, dps stance... there is serious homoginization here. All healers have a basic ST heal, stronger ST heal, basic AoE heal, stronger AoE heal, Raise not to mention almost always taking the same role actions with little exception. Let's not look too hard at AST/WHM in particular because we all know what a cookie cutter that is.

That wraps around to the "can they even add another healer?" thread. And yes, that's a problem. Fundamentally, a DPS job has to be able to put out X DPS in personal DPS and group utility to be viable. How it does that isn't relevant. So you can come up with distinct playstyles more easily.

Healing is not that. Putting out X HPS is meaningless, because healing requirements in this game are bursty. You need a big shield and lots of burst healing at one moment, and then little to nothing for a while. That means any healer they add MUST have that burst potential in order to be viable, which immediately kills concepts like a healer that does steady healing via damage dealing (because most of it would be overheal and you'd need to have a way to burst heal). They only really have three cast speeds on spells: instant, GCD length, and "Raise". So without changing that, concepts that play with spell speed for fast & small/slow & large heals don't work.

Tanks are similar in that they all need a certain amount of on demand mitigation due to how encounters are designed, they all need threat builders, etc. There might be a bit more room for variety there, but in general these two roles have requirements far more specific in nature than "do damage". That greatly narrows the options for making them unique. AST is the shining example of the problem, but I don't think it's easily fixable without changing how encounters are designed and opening up more spell cast time options, among other things.

I understand why these similaraties in support roles exist but it doesn't change the fact that it contributes to them feeling very similar overall and especially when compared to their DPS brethren who rarely share anything apart from role actions and for melee the presence of combos (except MNK).

In my opinion (as someome who still consideres themselves a healer main) DPS roles in general were designed with distinct identities in mind and then had potency tweaks to bring their dps more or less in line with each other where as tanks and especially healers were designed around their utilities and core skillset and then separated at some points by little more than aesthetics.

Yep.

The reason I enjoy dpsing now is that each time I do it the experience still feels fresh, probably because I've banged healer reflex memory into my skull through leveling them each to 70. There is a reason Cure/Physick/Benefic all share a button for me and for most healers. For someone who doesn't enjoy the actual healing gameplay I'd wager (much like me with tanking) the thought of leveling another job through basically the same routine seems at best daunting and at worst boring. I am quite sure this contributes in a meaningful way to the dps queue times.

That'd be why PLD is the only tank I have past 30. I have no incentive to go do those ARR dungeons again on the other two.

But if you look at other games, the same imbalances exist even where the tanks & healers are more distinct, like WoW was. Fundamentally, the people who actually want to do those roles are a clear minority. Making another job doesn't change that. You cannot take someone who doesn't want to be responsible for keeping the party alive and get them to play a healing class, no matter how many of them you add. The roles are not balanced on responsibility, how publicly obvious your failures are (ESPECIALLY in a game that bans parsers), and how much you have to pay attention to what the rest of the group is doing. No amount of classes will change that imbalance, and that's a major factor for a lot of people.

In a trinity game, it's a problem that is very hard to solve.

Also I'm very much in favor of 5 man dungeons and I don't think it would take as much tweaking as some people think either. Obviously we would need to increase enemy HP proportionally but there are plenty of dungeons where no actual mechs would need altering.

Tuning every encounter in the game to account for the DPS and more people on mechanics is the obvious cost. While it's substantial, it's not the biggest hurdle. There are two other major things:

1. Anywhere in the code where someone made the assumption that a party size is X (4/8) is suspect, becauase if they used a hardcoded number instead of a constant for the maximum, increasing it will break and probably cause a crash bug. That includes many places in the UI, network code, and so on. Much of that code will be very old, likely from 2.0 and possibly from 1.0 if it was carried forward. Going through all of that is a big deal.

2. Network load. While it looks like you're simply adding one person to a group of 4, or a 25% increase, what you're actually doing is far more impactful on the network. When the server is talking to clients in a party, it has to tell every client what every other client is doing, in addition to various system statuses and what the NPCs are doing. System status/NPCs are fixed cost, they don't change based on party size. The cost to tell a client about the other clients is exponential with party size, and for simplicity purpose we're leaving out the server having to tell your client about itself:

in a party of 2, the server tells 2 clients what 1 client is doing, for a total cost of 2*1 = 2
in a party of 3, the server tells 3 clients what 2 clients are doing for a total cost of 3*2=6
in a party of 4, the server tells 4 clients what 3 clients are doing for a total cost of 4*3=12
in a party of 5, the server tells 5 clients what 4 clients are doing for a total cost of 5*4=20

in a party of 8, the server tells 8 clients what 7 clients are doing for a total cost of 8*7=56
in a party of 10, the server tells 10 clients what 9 clients are doing for a total cost of 3*2=90

in a party of 24, the server tells 24 clients what 23 clients are doing for a total cost of 24*23=552 (these numbers being so large is why Alliance raids tend to be janky with things like raid wide damage taking so long to hit everyone, as the server works its way through each person and tells every client. It's also why world zones are limited to a certain size, because 1000 people showing up at the same FATE would create overwhelming load, although there's various techniques to mitigate this)
in a party of 30, the server tells 30 clients what 29 clients are doing for a total cost of 30*29=870


As you can see, boosting party sizes significantly increases the cost of keeping clients in sync on what the other clients are doing. That's a manageable thing, of course, but it's a much bigger deal than simply having the encounter design team boost the HP on everything and call it a day. You also have to multiply those increases by all the groups playing, which won't particularly change because the number of tanks & healers likely won't change a ton (it may go down a bit if queue times decrease). Total load goes up as a result of the change.

As a programmer, I'd argue that the encounter design is the easiest part. While it's certainly a major undertaking, it's fundamentally data manipulation primarily and is a known workload quantity that doesn't impact anything else the same way the network load changes do.