Caveat 1: I have never played any of the Dark Souls games, and it seems unlikely that I will any time soon. Everything I know about Dark Souls has come second-hand from Anderson. This is both good and bad.
Caveat 2: Given that it is hardly new, other OSR bloggers have probably already beaten it to death. It sounds like something The Delvers might've talked about. But, I do not care.
Here are some things that stuck out to me from Anderson's videos:
- Jayquaying - The sections of Dark Souls 1 that Anderson praises heavily are mostly thoroughly-jayquayed environments. He rags on DS2 for switching to largely linear levels, and then has mixed praise for DS3 in this regard. He notes that DS1 was jayquayed on two scales - both in local environments (eg, multiple paths and routes through a town), and in connections between environments (with links between, say, town and a valley full of dragons by way of both an aboveground road and a magma-filled cave below the mountains). DS3 did a good job of jayquaying on the small scale, but not at the large scale. Varying the level of jayquaying in megadungeon zones is an interesting possibility of looking at things this way - the megadungeon can have a nonlinear, loopful superstructure, with zones varying in their degree of nonlinearity. It also expresses one of the problems I had with Rathell - the zone itself had an extremely nonlinear / loop-heavy microstructure, but the overstructure of the dungeon it was supposed to be a part of didn't (at all).
- Resource model - DS1 provided the player with pretty limited healing resources that could be restored by resting at selected "bonfire" locations, which also respawned all of the enemies in the level. This leads to an expeditionary playstyle that sounds very similar to what we expect to see in the OSR, but with a focus on making it to a boss - you explore a level and get to know its monsters, gradually getting good enough at traversing the area that you can make it to the boss with enough healing remaining to win that fight. DS2 added inexpensive healing-over-time items to allow players to fight larger groups of enemies, while DS3 just provided much larger pools of healing than DS1. Similar changes to resource management have taken place over the editions in D&D.
- Investment - Anderson praises Dark Souls 1's system of investing limited resources in particular bonfires, which allows them to restore additional healing resources. This is useful when there's a boss you're having trouble with; you invest in a nearby bonfire and can bring more healing resources to bear in traversing that zone and defeating that boss. This is something that it seems like low/mid-level ACKS should be able to do a lot better than it does, with PCs investing in towns as bases of operation and receiving tangible, dungeon-relevant benefits for doing so. Currently town-buyable, dungeon-relevant resources in ACKS fall into three categories: gear, personnel, and reserve XP. None of these are really tied to the town itself; they're all pretty movable. A decent approach might be to take commissioning equipment one step further and allow a lump sum to be spent to increase availability of certain types of goods permanently in that market (eg, acquire a little land, build a building, hire a guy, and establish Doctor Comfrey's Nursery and Garden Center to increase available quantities of healing herbs forever). Have half the money spent count towards urban investment, and you add another link between the mid- and late-games.
- Predictable, Preventable, Decisive Damage - Anderson claims that damage in Dark Souls is largely avoidable, because enemies telegraph their moves, which allows the player to dodge / block / parry, but that when hits land, they hurt a lot (~3 hits to a player kill, usually). This ties into the healing resource management game, where you only have to spend resources when you've made a mistake, and part of mastering a level to make it to the boss is learning the attack patterns of the enemies on that level. Decisive damage is one of the things I like about OSR D&D (on a good day), but it does less well at predictable (ergo preventable). Bad Trap Syndrome describes this issue in the context of traps, but combat damage is sort of unpredictable too - if you engage in combat without a win-button like sleep, turning, or surprise, damage is a predictable outcome, but the details are left to chance. In a sense damage predictability is more nuanced in D&D than in Dark Souls - rather than making a mistake, you're taking a risk, and instead of a binary outcome you get a distribution.
- Training Wheels - Part of the reason that Anderson claims that Dark Souls is a mostly-fair game is that things are almost always introduced in a relatively safe way before being introduced in a dangerous way. When you meet a new type of enemy, you probably only meet one of them; deeper in the area where they appear, you start meeting multiple. When you enter the trap zone, you're alerted that it's a trap zone by a low-damage arrow trap triggered by a pressure plate, and then the traps escalate from there. This process of gradual escalation that helps make damage predictable. It is also something that I, as a table-driven DM, have not been doing well.
- Shortcuts - Another part of mastering a level in Dark Souls is finding and opening shortcuts - changing the environment open shorter routes to the boss. Keys to locked doors, lowerable drawbridges, levers that move terrain, that sort of thing (one neat example was destroying structures that were shielding monsters along a route, thereby making it effectively shorter for resource conservation purposes). This is something that makes sense in jayquayed dungeons, but usually rather than opening new routes from the other side they're used to gate entirely new areas. In practice the closest my players ever came to developing a shortcut was clearing (or befriending) monsters on preliminary expeditions in order to open a route that was safe to move quickly on. The trouble with building shortcut opportunities into an OSR-style megadungeon is that what counts as a shortcut depends on where you're trying to go, and player objectives usually vary per session, so what is a critical shortcut one day is irrelevant the next. Dark Souls overcomes this with the focus on getting to the boss. But...
- Boss Monsters - Bosses are something I always hate in videogames, but in tabletop games they can actually be kind of fun. They're pretty well-supported by ACKS' worldbuilding guidelines (where every tribe has a chieftain and every warband of barbarians is led by a 9th-level fighter), and would work pretty well with megadungeon factions - kill the boss and the faction disintegrates, opening up space for others or allowing players recruit the survivors. We saw the beginnings of this emerge in Rathell, where the Marrowgnawer (5HD nonmagic-weapon-immune giant rat) served as a "boss" of sorts of a ratman tribe. As usual for D&D "bosses", Marrowgnawer died like a chump to a 3rd-level party, due in large part to...
- Action Economy - Curiously, this has been a persistent issue for the Dark Souls series too. Anderson notes that in DS1, the best fights are one-vs-one duels, while any fight of multiple nontrivial enemies versus the solo player was usually quite difficult, and led to players using dirty tricks to isolate enemies, while in DS2 additional healing was made available to make these fights workable, and in DS3 healing was mostly-reverted but other changes to the combat system were made for this reason. In D&D the same problem rears its head on the DM-side. No bosses without bodyguards (and not chump 1HP 4e minions, either...), and also no bosses that don't one-shot henchman or two-shot PC frontliners.
- Gauntlets - In addition to nonlinear exploration zones and straight-line combat slogs, Anderson notes another sort of zone / level in the Dark Souls series, characterized by testing the player's ability to deal with some sort of complication that forces the player to reconsider and adapt their tactics. Examples that he cites include a level that is heavy on harassment by ranged attackers, a level with darkness (which requires the player to use a torch instead of his shield), and a level with environmental damage-over-time. Designing megadungeon zones based not merely around cosmetics/theme but also with a particular kind of tactical challenge in mind seems like a really good idea to me, especially because the shield-phalanx has come to dominate our games (of course, precisely because the shield phalanx has come to dominate our games, players are now reluctant to enter areas that require a change in tactics).
- "Explore cautiously, fight bravely" - I'm not going to go watch all the videos again to find the section where Anderson talks about this, but he claims that Dark Souls rewards players for exploring cautiously, taking it slow and not biting off bigger encounters than they can chew, but also for playing aggressively once combat is engaged, getting inside the reach of larger enemies, rolling behind them, and backstabbing. I like this philosophy, even if I'm not sure how to produce combats that encourage it in ACKS.