Arena v1.11: Man vs. Monster, Pt. 1

Last week we looked at the state-of-the-art Arena gladiator simulator in its Man-to-Man mode, including aging effects over the years, and found that it generated a very reasonable population of fighters of various levels.

Here we'll look at the Man-to-Monster mode, and find that we get a much more problematic set of data. In the current version, there are not one but two updates since we last run the simulator that work punishingly against our simulated gladiators:
  1. Previously we weren't including aging effects (here assessed as both ability-score penalties and level-loss after middle age, as per Gygax's Conan writeup in Dragon #36). And:

  2. Previously we weren't including any of the monster special abilities in the monster roster; that only got implemented at the time of the MonsterMetrics program in the last few months. Now our population of fighters will have to deal with poison, paralysis, petrification, level loss, dragon breath, swallowing, etc., for the first time.
Before we even get into the results of that, we should look back again at the very intimidating Level of Monster Matrix which is present in OD&D Vol-3:

Even at the 1st dungeon level, a character can easily encounter a 3rd or 4th level monster (on the 6-point scale), including such types as giant scorpions, wraiths, lycanthropes, gargoyles (with no expectation of having any magic weapons), etc. By the 3rd dungeon level, these types are the most commonly encountered, and any monster in the game is likely to be in play: dragons, vampires, hydra, balrogs, purple worms, etc. By my estimate, if a stock level of a megadungeon has around 30 encounters, then we would expect to find around 5 such top-level terrors on the 3rd level of your dungeon. Think about that for a second, a guess what you think the results might be, before reading further.

Okay: Let's combine the original by-the-book encounter table, the implementation of monster special abilities, and the effects of aging for the first time and see what the end result is. We consider a gladiatorial system that allows recruits so it has 10,000 fighters at any time, running for 100 years, with 24 fights each year. Monsters are matched against each fighter by rolling on the table above at a "dungeon level" equal to each fighter's level (or for 0-level recruits, the 1st level beneath the surface). The result is this:

The highest level in our population of 10,000 fighters are just a few Heroes of 4th level. Looking at the average age column we see: No one ever survives a single year! Even tracking everyone who ever lived and died for the entire century (what I call SupMaxAge here), no one ever lived past the age of 19. Even this is a bit deceptive: most of the population you see here was likely hired at age 18, one week before the turn of the year, which bumps up the age value to 19 almost immediately. So likely the longest that anyone lives is maybe just a number of weeks. And no one ever, ever reaches a level above 4th regardless of how high their ability scores are, or how lucky the get with magic items or opponent matchups or attack rolls -- even after possibly centuries of throwing 10,000 fighters at a time, on a biweekly basis, at the Monster Level Matrix shown above. It's a nightmare scenario.

I've tried a few tinkerings around that table to develop something more reasonable; for example, I've tried applying a modifier to the d6 roll on that table, perhaps on the order of −2 or −3 or so. Below I'll go whole-hog and apply a −5 modifier; that is to say, assume that every roll on the Monster Matrix is always a 1, and produce the lowest-level monster permitted in every case. Here we get:

That starts to look at little more reasonable, but still: No one ever reaches Superhero or Name level, even after a century of biweekly adventuring. Over the course of the century, we found one Conan-analog who lived to the age of 28, but no one ever survives to their 30's, and almost no one lives for as long as 3 years. For this, we've already really abandoned the Monster Matrix, by effectively ignoring the roll entirely, and looking only at cells that have the "1" result in each row.

I've also tried other more sophisticated techniques in conjunction to this modifier, like assuming that the fighters in question can "choose a lower level of the dungeon" on which to adventure, to some optimal level (like, one that accrues maximal XP before they start losing levels due to age). I'll spare you that series of searches and charts, and just say this: The optimal choice is always to stick to the 1st dungeon level forever. Anything else is just too dangerous.

Let's call a spade a spade: The OD&D Level of Monster Matrix is undeniably, clearly broken. For a playable game it absolutely needs to be fixed. This is highlighted by the fact that no one ever used it again in that format; all later editions of D&D (even Gygax's AD&D) provide some softer take on the monster encounter table. In the next installment we'll look at using one of those to inspire a revised mechanic for our OD&D games.


  1. One thing that comes to mind is that in the suggested dungeon generation, one in three rooms have monsters while one-half have treasures. So on average, one-sixth of rooms have monsters but no treasure and are to be avoided, one-sixth have monsters with treasure, and one-third have unguarded treasure.

    With this in mind, a lone treasure hunter would be best served to avoid fights entirely and seek the unguarded treasures. The solution to the deadliness of even one-on-one fights is meant to be solved by assembling an organized expedition - and adventuring party. Even then, there is the impetus to avoid overly dangerous fights, or those without obvious rewards, as a single hit can lay low a neophyte adventurer of any class, and at the end of the day every member of the party's number one goal is to return to the surface alive, riches in tow.

    1. Correction - I must have been thinking of one of the basic sets, I just looked at my pdf and it appears that fewer rooms have treasure without monsters - only 1 in 6 out of the two-thirds of monsterless rooms, or 1 in 9 overall. Much less common, but still tempting to a sneaky looter, considering the expected value of a treasure on Dungeon Level 1 is just shy of 300 gold pieces and a 5% chance of magic - with a chance for much, much more if RNGesus is with you.

    2. There's a lot a gray area around what the expected number of monsters should be in an OD&D dungeon, but I do think of Vol-3, p. 11 where it the only note in that regard says, "the number of monsters will be... modified by type... and the number of adventurers in the party". So that at least suggests something like PCs = encountered monsters; or at least an expectation that if you bring a bigger party, then the DM will commensurably increase the number of monsters, to keep the threat ratio balanced.

  2. Great stuff Dan!

    >>>"This is highlighted by the fact that no one ever used it again in that format;"

    Yes, although we do have an example in print of a 6 level dungeon stocked using this table, from the 1976 GenCon Castle Blackmoor dungeon. Discussed here: http://boggswood.blogspot.com/2015/11/appearing-in-blackmoor-dungeon.html

    Now I don't know what other tables you are going to look at, and I don't want to steal your thunder, but there are a couple other more or less "official" OD&D methods. The first is average monster HD (round up for +l) = Dungeon level. This is exemplified by the xp award discussion on page 18 of Men and Magic. Further, this Monster HD "level" equals dungeon level is actually what was finally adopted in the last version of classid D&D (blackbox) "The difference between a monsters’ level and a dungeon level should usually be no more than 2. (p68)"

    The other sort of "official" stocking method for OD&D would be Arneson's point buy method, which actually works really well and is probably the best way to do it. Discussed here http://boggswood.blogspot.com/2014/08/point-buy-systems-for-stocking-dungeons.html

    Last thing to mention is that it would be interesting if you had a look at Gygax's revised Monster Level Determination table in Holmes (p10) which weights the average monster level much closer to the actual dungeon level.

    1. Thanks for the kind words! That's a good insight on the late basic D&D black box set, since I never saw in that. I have noted the Holmes table before. And that's a fascinating writeup of the Arneson points-based method!

  3. Now that the sim is being used for more complex things, there are some things that start mattering more:

    1A: While monster group composition is typically homogeneous, delver groups contain multiple types of abilities that can synergize well. In some situations this might make it so a powerful monster ability that can be negated by a typical PC ability (like gargoyles needing +1 weapons to hit) is also possible in the other direction (Elves virtually immune to Sleep and Charm while their buddies suffer).

    1B: In a dungeon setting, you see a front rank of two or three figures. So a party of 5 zombies for example can't bring all of their numbers to bear. An adventuring party typically works out how to let every member participate in the fight, whether with spells, missiles hurled over short PCs in the front rank, spears wielded from the second rank, casting healing spells on the front-rank fighters, etc. Whatever total power the PCs can offer, a monster party of more than 2 must be able to offer proportionately less. Some exceptions include very small monsters that can slip among the party (like Giant Rats), tool-using monsters who can use spears and missiles (humanoids ... and things like manticores?), or monsters with ranged special abilities (Vampire charm, medusa gaze). Perhaps for missile users, this sustained ability to contribute to the fight from the back rank - and take advantage of arrow slits and high places - is why they are worth more XP, rather than just the free attack before closing to melee and possibly a free attack after the enemy has fled.

    2: Players are more likely than DM-run monsters to focus fire on one target until dead and then move on. The effectiveness of this tactic can be reduced by requiring action declaration before resolution, because it will tend to result in either the monster remaining alive at the end of the round because the party underestimated the attacks required, or more damage wasted in overkill. Regardless, I've found players far more frequently redirect their attention to spend an attack to kill off a monster than a DM doing the same. This may fall under the "DM tends to be less than 100% ruthless" which a simulator - or reality - wouldn't include. However it makes sense that low-INT monsters and animals might lack the tactics required to focus fire. PCs fighting animals should also benefit from their much greater ruthlessness in that animals will typically try to avoid the fight or waste time deciding whether to fight, while the humans make that decision immediately and then lay waste.

    3: Surprise can completely change the combat results, especially if using the 1E method where you're surprised 2 in 6 and the surprise duration is the number rolled on the die (and especially terrible is when a stealthy character surprises on 4 in 6!).

    4: What do the adventurers do with their nonmagical treasure? Do they have the choice to consume single-use magic items like potions in a dire situation?

    1. Those are fair points. Admittedly strategy can make a difference (and we hope it does under the "excellent play" rubric!), although simulating that would effectively take strong AI. I'm not simulating any equipment or magic items other than straight bonus-adding weapons and armor.

  4. Wait, if this is going back to one-on-one fights, why do you use it as a basis that this encounter table is "broken"?

    Aren't you generally going to see a relatively large group of PCs with NPC hirelings, and aren't they going to be rolling reaction checks to see whether they start out negotiating rather than straight-up fighting? You've pointed out before that most of the powerful monsters fare less well against groups of soldiers, and presumably even then the fatality rate is going to be cut significantly by some members of a party fleeing after a deadly opening salvo shows that engaging a given enemy would mean a TPK. Beyond this, parties can ally with some monsters and talk their way out of fights with others. I'm not sure how you'd take stuff like this into account in your program, and it's entirely unrelated to XP-for-combat calculations, but it sure should inform our perceptions of what kind of encounters are "appropriate" on a given dungeon level.

    1. Well, this partly gets into a matter of interpretation with what's written in OD&D.

      In Vol-3 p. 11, the note on numbers of monsters is that it's "modified by type... and the number of adventurers in the party". So it seems to imply that PCs = monsters, or at least that the DM is expected to proportionally increase monster numbers when a bigger party shows up. In that sense, an estimate of 1 PC to 1 monster seems like a fair model.

      Likewise, when I read p. Vol-3, p. 12, the first thing I see is "Monsters will automatically attack and/or pursue any characters they 'see', with the exception of those monsters which are intelligent enough to avoid an obviously superior force". The famous Reaction Table there is for the exceptional case of "more intelligent monsters" deciding what to do, whereas the default is simply always-attack. That's one of those text expectations that got lost in AD&D, and from a little of people's minds because of the visual impact of the table. (Link)

    2. I see! Thanks for the specific citations!

  5. I know this is years later, but I'm going through the encounter tables interpretive dance and maybe...

    "If the level beneath the surface roughly corresponds with the level (monster level table) of the monster than the number of monsters will be based on a **single creature**, modified by type **[as in single types vs group types]** ... and the number of adventurers in the party. A party of from 1-3 would draw [1x], a party of 4-6 would bring [2x], and so on [7-9 = 3x]... [blah blah blah about other variables and ref judgment.]"

    If you take that rule, and add in a wide-specturm random element, the base number appearing should be either 1 or 1d2 for 1-3 players, and something like 1d3 or 1d6 for 7-9 players.

    If you narrow the randomization a bit by just doubling the base number appearing from the original random roll, you get something like 3 or 3d2 for 7 to 9 players.

    Then it's just a matter of identifying which monsters are singles and which are group-types.

    1. Thanks for the comment! I agree, and that's pretty much what I'm doing these days myself. Assuming a party in the 4-6 range, I'm rolling a base 1d3 for wandering monsters, or 1d6 for a dungeon lair. My current megadungeon is designed for the 7-9 PC range of a certain level, so I'm rolling 1d6 for wandering, and 2d6 for lairs when I need them. (I did slightly more fiddly stuff like 1d4+1 a short while ago, but got sick of it and just switched back to d6's.)

      You can see that in the last paragraphs of the suggested handout here.

      Admittedly I just ignore the "group types" clause, which seems would be numerically broken to me. Everything follows this same system. However if a monster is _weak_ in terms of EHD then it gets multiplied by that ratio, so maybe the end result is like Gygax was thinking.