Underworld Overhaul, Pt. 2: Monster Level Matrix

You can’t really use the by-the-book OD&D Monster Level Matrix (Vol-3, p. 10) to generate the level of monsters in standard encounters; it’s almost comically, ultraviolently, overkill. We’ve noted this a number of times (links one, two, three, four). Every later edition of D&D agreed that it needed to massively dial down the danger level; including Gygax in AD&D (who might have actually over-corrected in that case). Here's a look:

Original Monster Level Matrix

Observation 1: According to this table, on the 1st dungeon level, any of monster levels 1-4 can appear regularly. Based on our initial Equivalent Hit Dice (EHD) analysis (link), the average monster on this level is not 1 EHD in power; it’s actually a little over 3 EHD that our 1st-level PCs can expect for a “standard” encounter there. Any one such encounter is a deadly threat, even solo against 4 PCs, say.

Observation 2: By the 3rd level of the dungeon, monsters across the entire spectrum can be expected regularly. For example, within a half-dozen encounters, our presumably 3rd-level PCs can expect to confront one 6th-level monster, such as: an adult dragon, 10-headed hydra, vampire, balrog, or purple worm (various EHDs between 12 and 39 for these types). And there’s simply no way that a 3rd-level party can hope to regularly run into encounters with even a solo creatures of EHD in the 20’s or 30’s and expect anything other than a TPK.

You get the basic idea regarding that original Monster Level Matrix. Just to nail the issue home, here’s the demographics resulting from stocking dungeons blindly via that table. Using a data file that matches the original Monster Level Matrix, this is produced with the current Arena program using the switches -n=10000 -v -z=4 -rs (that is: a population of 10,000 men, versus monsters, in parties of 4 at a time, reporting summary statistics; with default dungeon treasure, 24 fights/year, and 50 years total time):

Note that even after the 12 million separate combats that this represents (the 10K population is constantly refreshed as adventuring men get killed off), it's almost impossible for anyone to have graduated past 6th level. Perhaps more troubling is the fact that their isn't any evidence of higher average ability scores (esp., Str, Dex, Con) at the higher levels; this is a signal that character ability is effectively a non-issue, and affords no benefit in survivability. Basically everyone is just being thrown into a big random meat-grinder.

Revised Monster Level Matrix

So what do we recommend for a replacement to that table today? Here’s a simple solution concept: Let’s say that an Nth-level party, operating on the Nth-level of the dungeon, should expect to run into monsters of Nth-EHD, on average. (Obviously, that is far from what the original matrix produces). To the extent that the Monster Level Tables partition various EHDs into coarser chunks (see last post), we’ll simply reflect the same in our new matrix: six monster level tables are synchronized with six rows in the new matrix. Each row of the matrix covers the same dungeon levels that the equivalent monster table covers in terms of EHD. In each case, the average die-roll of “3-4” should sit where the row matches the column number (i.e., where party level matches monster EHD; symmetric down the diagonal). The result is the following (as in the data file on GitHub):

If we run our Arena simulation using this new matrix (same parameters as above), then we get the following demographics for the adventurers in question:

Now, that's still a hard game; while there are more 7th- and 8th-level characters present, still no one succeeded at cracking Name level (after 50 years of adventures by the 10K population). No one can accuse us of giving away the store with this modification; arguably we could dial down the generating matrix even more than this. Note, however, that character abilities do now seem to correlate with survival, especially in the average Dexterity column here (for man-vs-man fights, we've previously seen that Strength is more fundamental). So this might represent a simple and reasonable upper bound for Monster Level Matrix difficulty. Also, if the DM regularly places "special" treasure at a steep multiplier over table-generated treasure (as sayeth the book), then advancement would be more swift than this; at least the PCs would have some chance of surviving to benefit from such treasure caches.

Still to come: Monster numbers and treasure value.


  1. This is great, but all this use of the word level can get dizzying. http://www.giantitp.com/comics/oots0012.html

    1. Very true. See also: 1E PHB p. 8, "An Explanation of the Usages of the Term 'Level".

  2. Can I just say that this was a series of posts I've been excited to see you tackle for a while now? I can't wait to see the completed series! As a sometimes OD&D GM, I think these adjusted tables will make a valuable tool.

    Also, I think if PC parties are surviving to the higher levels, it suggests that either the other non-fighter classes add a significant power boost to any party, or at least that the extra diversity is important. That, or it confirms that avoiding combat is a must for anyone hoping to reach high levels.

    1. I'm so glad to hear that! This sort of came out of me a few weeks back like unavoidable birth-giving. I think you're very right about magic powers changing stuff greatly if cunningly utilized. Would love to have that assessed someday by super-powerful AI. :-)

    2. (Which argues somewhat for a PhD in that area by yours truly.)

  3. Likely I'm misunderstanding some part of the metrics being used, but can "literally no one made it to name level" part of what could be considered a workable solution? Haven't you just raised the lethality bar two levels? That's an improvement, but it still seems to have a rather large flaw.

    1. You're right -- it's not a tremendous change, nor did I want one. But it's a step the direction of improvement, and it gives a coherent system for insertions of other new monsters. I do think that Gygax in AD&D over-corrected, and I didn't want to be accused of too soft a game.

  4. I suspect that a major reason behind the dearth of high-level characters is foolhardiness -- they fight to the death against overwhelming odds and, well, die.

    An actual PC would have some more options for survival, even beyond magic. Avoiding Monsters (U&WA p.12) is very generous.

    Consider the three biggest killers in Monster Level 6: the Purple Worm, Vampire, and Basilisk.
    Purple Worms and Basilisks move at 6", and can be outrun by literally anyone who's willing to throw away their bags to survive. (And isn't surprised and immediately eaten/stoned, that is.)
    Vampires move at 12/18 and thus are much more difficult to escape, but they're also intelligent so dropping a bag of loot has a 90% chance of slowing them down.

    Let's consider the case of The Strategic Review 1's Solo Dungeon Adventures, for a lack of better options. Monsters are either encountered as Wandering Monsters on the Periodic Checks (hallways only) or in rooms/chambers.
    If they're a Wandering Monster, the player must have either come from a room (and thus through a door and 4-in-6 chance to evade), a passage (something along the lines of a 60% chance of there being a turn or door in the last 60'), or a chamber (and thus likely a turn or door beyond that).
    If they're in a room, there's a door that they just came through and a 4-in-6 chance of avoiding the encounter entirely by just closing the door and leaving immediately.
    If they're in a chamber, refer to the bit on passages above.

    In any case, you'll note that there's a good chance of being able to evade a monster if you're at least as fast as it is (even before taking into account dropping food/treasure). Faster monsters are much trickier, however.

    I do understand why this is not modelled, however: the presence of doors/stairs/turns is hard to determine, it's tricky to figure out what the optimal threshold for caution is, and the entire section on dropping treasure would require bringing in an entire inventory management simulation.
    It would require making an actual dungeoncrawl-lite simulator rather than merely a combat simulator, and that's a lot of feature creep.

    1. That's a nice point -- and thanks for the observation of the low move rates of some of those top-end monsters. I hadn't observed that before (and MV is not a factor in the simulation at all at this time).

      On the other hand, the simulation does assume every "expedition" is just a single encounter with all the party members at 100% perfect health. (Or in other words: they retire and rest up for a month after every encounter.) So I think that somewhat counterbalances the buffer from retreat options. I might hypothesize that before a party knows to retreat, likely one of their members will get dropped, and then from the perspective of the individual you're right back in the same lethality of the simulation as it stands.