Raid Boss Dissection: FFXIV, Thordan Extreme

Because I am obviously much closer to the development process on WoW, I don't really get the opportunity to dissect our fights as a player in the same way that I do for FFXIV. So I thought it'd be a fun exercise, and might be interesting for people to read, to have an MMO Raid Boss designer's PoV into the design of another game's raid boss.

Note this is pretty casual and I didn't spend a ton of time going into depth and preparing information for this, so if you do find this interesting let me know and I may spend more time on things like this in the future. Also please comment if you have any questions/arguments/corrections/etc!

Fight Info

First, here's a link to a video of the encounter:

https://www.youtube.com/watch?v=HuDTxp2iedg

Here's a link to a video guide, which explains many of the mechanics in detail (with lots of bad puns):

https://www.youtube.com/watch?v=-pZ5MfWpOTw

Next, some things, if you haven't played FFXIV but have played WoW, that you should know:

• All "real" raids in FFXIV are 8 players, and groups are typically 2 Tanks, 2 Healers, 4 DPS. There is also a type of content where three raids join together in an instance, but this is generally for more casual content, sort of parallel to WoW's LFR, but with only one difficulty.
• Personal defensive cooldowns, and especially external healer cooldowns, are pretty minimal in FFXIV. Overall, players are substantially weaker in FFXIV than they are in WoW - or, maybe more accurately, players have much less opportunity to use abilities to counteract mechanics.
• FFXIV has no dungeon journal, very few boss emotes, no support for UI addons, and even damage meters require a 3rd party program that is technically against the ToS.
• Players can be freely resurrected in the fight when they die, but suffer a stacking -max health and damage done debuff for one minute after being resurrected. Additionally, the resurrection spells have exceedingly long cast times (8s, interrupted by moving) and while you can use Swiftcast to raise someone instantly(which makes it instant), Swiftcast is on a 60s cooldown.

As a quick breakdown, here's the general fight structure. Note the timing is based on the first video linked, DPS can change it somewhat:

• 0:00 - 1:40 : Introduction to Fight Mechanics
• 1:50 - 2:05: Dance 1 - Chains, Area Denial, Soak Towers, 
• 2:05 - 2:28: Sir Zephirin: DPS Check/Mercy Rule 1
• 2:28 - 2:38: Dodging "divebombs", moving in to position to fight the brothers
• 2:38 - 4:08: The brothers + dragoon fall off damage
• 4:08 - 4:28: Dance 2 - Ice Bombs, Charges, Growing Area Denial + Knockback
• 4:28 - 5:14: Dance 3 (World in Flames + Player targeted missiles) + Burndown
• 5:14 - 5:51: ULTIMATE ATTACCCCKKKKK!!!! (plus little breather)
• 5:51 - 6:52: Dance 4 - Healer Stun + Line Meteor + Tank Buster + Look Away + Raid Damage
• 6:52 - 7:41: Dance 5 - Soak Towers + Look Away + World In Flames, Growing Area Denial, Knockback + Meteor
• 7:41 - 8:30: Dance 6 - Lightning + Divebomb + Charge + Dragoon Fall Off + Meteor
• 8:30 - 9:27: Dance 7 - Chains + Large Player Targeted Area Denial + Player Targeted Missiles + Ice Bombs + Pie Wedges
• Rest of Fight: Burndown w/ Ramping AoE Damage

Wow! That looks like a whole lot of junk! At first, when I was looking at guides to this fight, it looked a lot more like "mechanics vomit" than "one of the most elegantly designed and well tuned boss fights I've seen". But, first appearances are deceiving - it really is an incredible boss fight.

How the hell does it work?

Consistent Visual Language

First, and this is crucial - every mechanic on the fight has a distinct visual tell - no mechanics lack visual tells, and no mechanics' visual tells are similar.

For instance, this ring is the generic "you're being targeted with something that's going to go off soon" visual:

This visual is only used for two mechanics on the fight, and these are very simple avoidance mechanics - this attack, and the pie wedge slash that he uses in the beginning of the fight and in Dance 7.

The other attacks in the fight are quite distinct visually. For example, this means you are being targeted by a meteor:

This means you are being targeted by a dragoon for Falloff damage:

And so on. It would be easy for a game to try to enforce a single "you are being targeted by an effect" mechanic, and in so doing create something that is very confusing and hard to parse. Even though the huge variety of targeting visuals in the fight may seem at first overwhelming, as you progress and learn the mechanics, you appreciate being able to instantly associate a visual with a very specific game mechanic. Which leads to...

Memory, Re-use, and Clarity of Purpose

Probably the single greatest aspect of the fight: while there are a large number of phases, there are not as many mechanics as it looks at first, especially considering you've seen many of these in the game before.

Looking through the fight I counted 7 new mechanics, 2 of which are very simple avoidance mechanics, 6 mechanics that you've encountered before with different visuals, and 8 mechanics you've seen 100% identically before.

Now, some players will make the argument that this is lazy re-use, but on the contrary - the mechanics exist to elicit specific reactions from the players, and the quicker you can get players to understand what the intended response is, the more easily you can ask them to do complex things in reaction to combinations of mechanics.

Additionally, because they are willing to accept a large number of mechanics, each mechanic can be quite simple - a single problem with a single solution. This increases clarity of purpose, and the likelihood that players will understand what is being asked of them, as well as the likelihood that the designed solution is the solution that players will use.

A common pitfall in attempting to use too few mechanics, given a certain level of complexity you are trying to achieve, is that each mechanic has to shoulder a much larger portion of the complexity burden of the fight. This often ends up with an individual mechanic that has several elements - multiple effects, bizarre fail conditions, etc. or, fails to achieve its intended gameplay because of some gap in the design of the ability.

As a designer, each mechanic in a fight is both a tool to create interesting gameplay, and a type of problem you are asking your player to solve. While it is fun to think up crazy new mechanics, re-use has major advantages in that you can much more quickly get to the gameplay you want without having to re-teach players the mechanic.

Organic Complexity & Player vs. Designer Fun

It's fun to design new mechanics. But something that's more fun to design isn't necessarily more fun to play. Sometimes, clever combinations of existing mechanics can be just as interesting as a new mechanic.

For example, a trick that floored me when I saw it in the fight was very simple: target a meteor at a player, and then knock all the player back in a common direction from a single point shortly before the meteor detonates. Additionally, hitting the wall will kill you. Each of those mechanics is incredibly simple - seen in numerous boss fights in MMOs. But the combination requires a very different reaction - typically, your instinct as a player after having been knocked back is to run as quickly as possible back towards the boss. But if you do that, then the meteor doesn't hit enough people. Your instinct when soaking a meteor is to move as little as possible to try to soak the meteor - but if you do that, you get knocked in a different direction from the clump, making soaking very difficult. You intuit that you need to prestack near the center to get knocked in the same direction, and then stand still to make sure you soak the meteor properly.

This mechanic doesn't necessarily make a designer feel super smart when designing it. You don't get to give your timing of 2 mechanics a special name and unique visual. But overcoming that combination of mechanics makes the player feel smart. Obviously you've designed your solution, but because the complexity originates from organic combinations of simple mechanics, the player feels like they've come up with the solution rather than just finding the bespoke solution that the oh-so-clever designer intended in the first place.

Lawd Have Mercy

A design value we talk about a lot on the WoW encounter design team is minimizing the time between when the player made the mistake that will kill them, and when they see that failure. We often talk about the problem with the player making a mistake at two minutes into a fight that causes them to wipe to the berserk timer, eight minutes later. This is, generally speaking, very frustrating and not fun.

The concept is similar to mercy rules in sports - points where, when one team or player gets ahead by enough, they win instantly without having to play out the rest of the match.

This problem is magnified in FFXIV because, since players can resurrect with a damage down debuff, often times you can limp through a fight, constantly losing people, only to wipe at the very end to the berserk timer.

Thordan uses two "mercy rule" strategies to end fights early when players aren't performing well enough:

• Heavy unavoidable party-wide damage shortly after bursts of avoidable damage, intended to finish off the party when they fail a large number of mechanics.
• Short but fairly forgiving DPS checks through the fight that you are unlikely to pass with dead players.

Often these two are combined - for instance, the two Brothers phase of the Thordan fight deals heavy damage to individual players via the Dragoon Dives - but these also deal falloff damage. The Brothers themselves also have a large unavoidable AoE. Done correctly, healers have just enough time to heal the Dived players with a single target heal, and top off the party with an AoE, before the group damage hits. If the players targeted by Dragoon Dives do not move away from the group, then while the damage they take doesn't change, the fact that the rest of the party also took heavy damage means that the AoE heals are not enough to save the party from the unavoidable damage. And, the brothers' unavoidable AoE ramps in damage with each cast - ensuring that, at some point, you will wipe if you fail the mechanics badly enough. This is preferable to letting the player limp through the phase with constant damage, but wiping minutes later to a berserk timer.

Peaks and Valleys

From 5:17 to 5:50 - a full 33 seconds - there is essentially nothing happening in the encounter.

This is something that is VERY different between WoW and FFXIV, and that I know is very controversial when I talk to people about their boss fights. But personally, I really appreciate that their game will follow bursts of very strict mechanics checks (in this case, the past two minutes of dodging mechanics + dps check) with a rest period - a chance for healers to top everyone off, to raise players who have died, to take a breath. It makes the level of tension of the mechanical bursts acceptable and avoids overly stressing players out. It feels like a reward in itself - you did good, take a little break. It doesn't need to mean you do nothing, either - for instance, they often include burndowns at the end of their fights that are mechanically very simple, but are essentially healing and DPS checks. If you made it through everything else, and you can do a decent amount of DPS, you win. These can be more or less tightly tuned - Thordan's is very loose - but that break, to just sit back and do your rotation - feels hard earned after 8 minutes of dancing around mechanics combinations.

Additionally, they often use those moments to sell how badass the boss is with a crazy animation, which helps. It makes you feel incredibly strong that you can take out this creature that can blow up the whole world around you.

Anyways...

Hopefully this gives some insight into the thought process of designing a raid boss. I'm sure it could be more interesting or better written, but I figured just writing something beats giving up on it halfway. Please leave feedback if you liked/didn't like/would want more!