Continuing my conversion of the Age of Worms Adventure Path to 4th edition D&D, and received additional insights as to the structure of D&D's 'resource management' and why Mike Mearls's "Three Faces of Evil" is such a PC lawnmower.
If you ask the question, 'which of the twelve adventures in the Age of Worms Adventure Patch is the most deadly?', you might naively think that the answer would be the last -- it's the final adventure of the campaign, the party might well be breaking into epic levels, and the goal is to stop the apotheosis of a mad god of undead. Why wouldn't the designers pull out all the stops? If you go to the Paizo.com Age of Worms message boards looking to answer that question, though, you get a surprising and different answer -- the most lethal of the Age of Worms adventures is the second, Mike Mearls's "Three Faces of Evil", designed for 3rd-4th level characters.
The reason this is true has more to do with the concept of resource management than any specific flaw in the design of the adventure -- after all, each individual encounter's EL is appropriate for a party of 3rd-4th level characters, and until the very last encounter, no single monster is of overwhelming power. The reason the adventure as a whole, and the Temple of Hextor specifically (where our party had its TPK the one time I ran the 3.5 version of the Adventure Path) are so lethal has to do with design expecations about player resources, and the sometimes unconscious ways in which players have internalized those expectations. These expectations haven't changed from 3.5 to 4th edition D&D; if anything, the underlying principles were brought even closer to the surface in 4th edition, which is why modifying "The Three Faces of Evil" for a 4th edition setting is such a challenge.
Begin with the premise that D&D and similar RPGs are, in one perspective, exercises in resource management. The fundamental resource for characters in D&D are their hit points -- lose all your hit points, and your character perishes. 3.5 D&D sets up a multi-layered approach to hit point management, though:
- Hit points (HP) are reduced via attacks (usually versus armor class) or enemy spells (usually mitigated by saving throws), and can be regained via rest, spell, or other magic.
- The ways in which HP are regained are themselves resources, subject to different 'recharge cycles'. A cleric can cast a cure spell either in combat or after combat to immediately restore some HP; the cleric can only recover the spell after a certain amount of time has passed, however. Likewise, a healing potion will restore some HP, but the party must then either purchase or create a new potion to replace the one used.
The same basic structure exists in 4th edition, with 3.5 saving throws being replaced with 4.0 Non-Armor Defenses (so that all methods of reducing HP are now, in effect, attacks), and with the healing powers of the cleric in 3.5 being generally distributed to all the 'leader' role classes in 4.0 (ignoring for the moment the impact of the 4.0 'healing surge', which makes the process of HP management much more sophisticated).
Ultimately, every resource recovery mechanic eventually defaults to time -- you need the time in combat or out of combat to cast spells or use items, then the time to rest to regain spells or recharge or repurchase items. The biggest challenge to a party, then, is finding that time in a place where enemies aren't able to continue attacking their HP resources.
The wrinkle that 4th edition makes more obvious, though, is that the time component is divided into two different cycles -- a 'short rest' cycle whereby characters can regain some of their HP recovery resources, and an 'extended rest' cycle whereby characters can regain all of their recovery resources. Players in 4th edition internalize these cycles by checking their HP to see if they need to take short rests (i.e.: use their top-tier HP recovery resources), and also checking their healing surges to see if they need to take extended rests (i.e.: recover their HP recovery resources). A party at full HP but almost no healing surges won't want to continue adventuring, because they know that their HP resources aren't able to be recovered as efficiently, while a party at low HP but full on surges will take one (or more) short rests to use their surges to recover their HP and continue the adventure.
The problem with the Hextorian battle temple in "Three Faces of Evil", and to a lesser degree, every major adventure section in the adventure, is that the adventure prevents the players from being able to exercise that judgment. The Hextorian battle temple encounter begins, in effect, as soon as the players first enter the dungeon -- tiefling guards stand at the bottom of the elevator shaft where the party will appear, and one of them is directed by tactics to run to the door of the temple and deliberately trigger the temple's defenses. This creates a running series of battles where the party has almost no opportunity to stop and assess their resources between battles to determine when the optimal time to withdraw would be. The ideal execution of the strategy behind the Hextorian battle temple would be for the party to be drawn into the gladiatorial arena room (and locked into it) after having expended significant resources battling through cultists, guards, and a guardian beast on the way there; by the time the party has an opportunity to assess their need to rest, they no longer have the option to rest or withdraw from the encounter.
To a lesser extent, each of the three temples is organized this way, though the tactics given to the DM are not as overt as in the Hextorian battle temple. In addition, if the party slays each of the cult leaders (which most adventuring parties will do, given player tendencies to kill and loot their enemies), the final encounter with the Ebon Aspect occurs within the entrance to the complex, so that the party has no opporunity to withdraw to the safety of town prior to the climactic final battle. As such, the greatest challenge in converting "Three Faces" to 4th edition is not in finding appropriate monster stats, but in trying to find places within the adventure where the party is able to take short rests to recover their encounter powers despite not having 'cleared' the temple they're in.
(Keep in mind as well that, although we're only discussing the HP resource 'tree' here, a party's ability to deliver damage to and thus defeat opponents follows a similar resource management structure; thus, the longer a party goes between short rests, the less effective they become in defeating the opponents they face. Since the toughest opponents are usually at the end of these serial encounters, again, it's easy to see where the party is effectively stuck with few options for victory if they've used their larger powers too soon. So it's really a double-whammy -- less ability to withstand the enemy's attack, combined with a reduced ability to eliminate the enemy to stop the attacks.)
One thing I find ironic is that, as far as I can tell, Mike Mearls wrote this adventure as a freelancer for Paizo, then parlayed this (among other) adventure writing gigs into a full-time job at Wizards of the Coast. How is that ironic? Because the current edition Dungeon Master's Guide specifically warns you against designing adventures in the style of "Three Faces of Evil", so in effect Mearls was hired based on his experience in writing flawed (yet flavorful) adventures.