Wilderness Encounter Levels

We've spent some time on this blog in the past measuring the risk/reward level of the OD&D dungeon wandering monster tables (conclusion: as written, in total they're murderously lethal; even Gygax in AD&D massively ramped down the danger level). It recently occurred to me to ask a similar question about the OD&D wilderness encounter tables.

A somewhat theoretical difference is that while the dungeon tables have "levels" which theoretically relate to level of power of the monsters there, and suggested level of PCs adventuring there, the wilderness table don't come with that same packaging. Instead (obviously) it comes distinguished by "terrain types". We might assume that plains are designed to be safer than woods, and woods less dangerous than mountains, etc., but are they really?

What I did was go through all the entries in those tables and compute average Equivalent Hit Dice (EHD) from each type of encounter, using EHDs estimated algorithmically as shown in the OED Monster Database (code on GitHub) For example: Working bottom-up, here's the sub-table for "Typical Men" (i.e., the default for men in most terrain types):

OD&D Wilderness Subtable: Typical Men

The "grand average" of all the encounter averages in the rightmost column is 122, but that masks the bimodal structure of the table. These encounters split neatly into two halves: six are with mass groups of men in the range of hundreds (with a host of leader-types, including a Superhero or Lord for any group 100+; I estimated the EHDs for this by adding 25%), while the other six are with small parties of 2-12 men and a single NPC (like a Superhero or Lord). The average total EHD for the first category is in the 200's, while for the second category it's in the 20's. Clearly there's a big difference between meeting one Lord and his 200 soldiers, versus another Lord and his 10 soldiers.

This was done for all the different sub-tables, and "grand averages" compute for each resulting in the following:

OD&D Wilderness Subtables: Average EHD

Somewhat similarly, there's a big bifurcation in the danger levels of some of these subtables. For the Men and Giants (i.e., humanoids) tables, average EHDs are the range of 100+ -- specifically those tables which can produce bands of men, goblins, etc., grouped in the hundreds (30-300 men/orcs, or 40-400 for kobolds/goblins, etc.). For the other tables, average EHDs are only in the range of 20-50 or so.

Finally, we can turn to the top-level table, which serves as a function from terrain type to the different subtables, and see on average how dangerous each terrain type is on a per-encounter basis. We get this:

OD&D Wilderness Encounters: Terrain Table

What I've done there is compute the average encounter danger across all results for a given terrain type, and then divide by 8 (an assumed large PC party size?) to come up with a rough "suggested PC level" for adventuring in that terrain. Some of those assumptions can be easily debated, but at least it gives us a normalized basis by which to compare different terrain types.

The result is that on average, there isn't that much difference between the various terrain types. Rounded to the nearest integer, the Clear type suggests maybe 9th-level PCs, while Woods, River, Swamp, and Mountains are only one pip up from that, at 10th level. The City is 11th (because technically it generates more bands of 100s of bandits and brigands from the "Typical Men" table, however unreasonable that may seem), and the Desert table is 12th level, somewhat more dangerous (again because the "Desert Men" table skews more towards 100s of nomads and dervishes).

So this series of averages is a somewhat rougher analysis than I've done for dungeons (which have been given a complete simulation in software at the level of individual fighters adventuring and gaining experience in separate encounters). The overall distribution of encounters is not entirely clear, although it's trivial to guess that the tables with fewer Men and Giants encounters (River and Swamp) will have less variation than the other tables. Here are some other factors abstracted out by this rough analysis:
  • No modifiers are made for parties with special equipment (horses, ships, underwater, etc.)
  • No distinction is made for parties that may be parleyed with and turn out to be friendly (likely dependent on alignment).
  • Dragons and lycanthropes do not have family/pack structure simulated (which mandates presence of some immature figures, but also makes adults fight more fiercely).

Another thing that the numbers above overlook is that while the average encounter is roughly equivalent across different terrain types, the rate of those encounters is not. E.g.: Compared to Clear terrain, in Woods the party takes twice as long to cover a distance and has double chance for encounter each day; and in Mountains both time and encounter chances are tripled, etc. That is, for every 1 expected encounter for a given distance in the Clear, in Woods you'll expect 4 encounters, and in Mountains 9 encounters, for the same distance traveled. Ultimately that's where the real difference in danger levels comes from in this system. (On the other hand, with only one encounter per day, casters can unload their entire firepower capacity on each one, giving some buffer against that added danger.)

Finally, this project suggests a significant limitation to the overall attempt at using our EHD values in sum to balance against total PC levels. Here we've come up with a rough suggestion that OD&D wilderness encounters are, on average, a fair fight for a party of eight 9th- or 10th-level PCs. However, we can look back to our experiences in Outdoor Spoliation games using this system, which we've run with fairly large parties of around the 8th level; at least four times we've documented battles with groups of men and goblins in the size of 200+, and not had a single PC fatality in those encounters. (By the numbers a group of 200 bandits should be ~250 EHD, so for a eight-man party we would have suggested they be 250/8 ~ 30th level? That's clearly not right.) This points to a likely breakdown in simply summing EHDs, especially for very large groups of low-level monsters, versus PCs with high-level magic (not currently simulated in our program), very low armor classes for fighters, etc. It may be interesting to reflect on the exact magic used by players in those mass battles in Outdoor Spoliation sessions One, Two, and Three.

Full spreadsheet available here for the tables and calculations shown above.

Edit: Consider Arneson's rule in First Fantasy Campaign that (as I read it) wilderness encounter numbers are really for full lairs only, and encounters outside will only be 10-60% of those numbers (average 35%). If we take the charts above and multiply everything by 0.35 for expected outsiders, then the equated PC level (parties of 8) becomes 3 or 4 in each terrain. Which is kind of interesting, because reportedly at the start of Arneson's games everyone got Heroes from Chainmail -- fight as 4 men, D&D 4th level -- or else Wizards (I presume low-level, likely 4th-level equivalent?).


Sunday Survey: Wizard Armor

A while back on the Facebook 1E AD&D group, a discussion occurred that had me quite surprised by the direction it was going. Intrigued, I asked the following poll question:

This was surprising to me, because given the context, the top result ("elven chain only") is clearly counter to the 1E AD&D rules text. Of course, when we say multiclass fighter/magic-users in 1E, we're just talking about elves and half-elves (the only races allowed for that multiclass). Under Elves on PHB p. 16, it says that they can "operate freely with the benefits of armor, weapons, and magical items available to the classes the character is operating in", with the exception being if thief activities are occurring (so: plate mail and anything else is clearly on the menu for fighter/magic-users). Note that this contrasts with gnomes on the same page who are restricted to leather for any multiclass combination. Furthermore, as of the 1E AD&D PHB, "elven chain" wasn't even a thing yet named or defined; it didn't appear until the later DMG (p. 27, as "Chain, Elfin") which says merely that it's thin and light, with no special notes about spellcasting.

I think partly, the result of this poll can be explained by later edition's rules "bleeding" into the memory banks of the many gamers who played mix-and-match a lot with different edition products. It was the 2E AD&D PHB that established elven chain as a sufficient and necessary requirement for multiclass wizards to cast in armor: "A multi-classed wizard can freely combine the powers of the wizard with any other class allowed, although the wearing of armor is restricted. Elves wearing elven chain can cast spells in armor, as magic is part of the nature of elves." (Ch. 3).

Moreover, we can look at 1E adventure products by Gary Gygax and possibly detect an "implied ruling" in the same direction on this issue. Looking at the many drow fighter/magic-users throughout the D1-3 series, all of them are equipped with fine chain mail (not a single one in plate, to my knowledge). The 1983 World of Greyhawk boxed set's Glossography has wandering encounter listings for that world, including "Elves, Patrol"; these are led by high-level elven fighter/magic-users with base AC 4 or 5 (chain, with or without shield). On the same page, "Elves, Knights" (p. 4) are principally fighter/clerics with better AC, but they have fighter/magic-user assistants again with AC 4 (chain & shield). So the consistency of this pattern may be another telling point.

Not initially knowing about the 2E AD&D rule or the apparent AD&D player consensus, I've done a similar thing in my OED house rules for OD&D for about a decade now; without reference to any special elven manufacture, multiclass fighter-wizards can cast spells in chain but not plate (also must have one hand free, no shield). Actually for quite some time I thought that was a semi-unique ruling; my surprise is that I've unintentionally matched how a lot of people elsewhere also play things.


Friday Figures: Swords & Spells Stand Sizes

I got started in the hobby at the point where the only ruleset for mass D&D combat that I could find in stores was the last supplement for original D&D: Gygax's Swords & Spells (1976). Looking at page 2, I saw numbers for scales and recommend size for miniature figure bases. To wit:

Figure Mounting sizes from Swords & Spells

Those fractional values looked odd, but I never questioned them, assuming that the author had some deeper reason for them. But looking back more critically today: Why 5/8" (instead of say, 1/2")? Why 1-3/8", or 1-5/8"? Why so complex?

Now, consider the following. We'll take a few key measurements in millimeters, in multiples of 5, and convert them to inches -- in each case rounding to the nearest eighth of an inch. We get:

Conversions rounded to eighth of an inch

That is, (noting 6/8" = 3/4") we get exactly the figures on display in the Swords & Spells table.

It occurred to me to check this when looking at Chainmail (1971), which gives the option of using either 30mm or 40mm scale figures for man-size (or in the fantasy section for others races, corresponding options such as 10mm, 20mm, 25mm, etc.). It should also be noted that the base sizes shown in Swords & Spells, and even the Chainmail suggestions for 30mm-scale goblins, orcs, ogres, etc., closely matched those provided by the Warhammer Fantasy product all the way up to 2015!

Theory: Gygax had miniatures that were originally based in metric measurements, which he mechanically converted to imperial figures (to the nearest eighth-of-an-inch) for the Swords & Spells publication.


Damage Scales in LBBs and Supplements: There and Back Again

One of the things I really like, DM'ing games from the Original D&D LBBs, is that all hit dice and damage are d6-based. So I can set up with a big batch of d6's (wargame-style) and use that for all monster hits and damage without poking around for sufficient d8's for hit dice, or 5d4 damage or something. In addition, it's very rare for monsters to be noted with multiple attacks, so combat goes quite rapidly.

This got massively reworked in Supplement-I (Greyhawk), and personally I think it's one of the off-the-rails mistakes in the history of D&D. In this work, you get the establishment of different hit dice by character class, variation of damage by weapon type, and also variation in attacks and damage by monsters (each listed as "Addition/Amendment" and "highly recommended"). Actually, the first two -- giving increased granularity on the player side -- I have no problem with, but simultaneously complicating all the monsters is the part I prefer not to use.

As that was done, the damage output from monsters increased, approximately on the order of being doubled. Let's take a closer look: Below I've compared all the monsters in OD&D Vol-2 that are given explicit damage specifiers in their text blocks, with the "new" damage specifiers given in Sup-I. Note that by default all the other monsters should have 1d6 damage in Vol-2, but for brevity I haven't listed those (and also confidence: did Gygax really make a deliberate choice that dragon bites, purple worms, etc, should do 1d6?).

You can see above that a comparison of the average damage output for these types shows a linear relation from Vol-2 to Sup-I, being a bit less than doubling between those works. We should be a bit careful, because the correlation isn't perfect; for example, ogres have the same average damage in both volumes. There are also a number of monsters not shown here who effectively have reduced damage, by being given less than 1d6 damage in Sup-I (kobolds, goblins, giant rats, etc.)

One thing that complicates my desire to stick with the LBB all-d6 (low damage) method is that while in Sup-I the amendments were quasi-optional, everything that came later on was designed only in those inflated, non-d6 terms. For example, there's a lot of interesting and memorable D&D monsters that only appear in later supplements: like lizard men, harpies, liches, ogre magi, hell hounds, owl bears, golems, giant frogs/toads/beetles, sahuagin, demons, and many more. Stat blocks for these types are only available with the inflated numbers.

(Note there is one unique exception here: In Sup-I, the text entry for the new Storm Giant type is the last place to give LBB-scale damage, "unless the alternate damage system is used". So the text says 3d6+3 damage, while the revised table in the same book gives 7d8 damage; a big difference.)

As a possible solution, consider taking the regression formula above and reverse-engineering all the supplement damage scores, so we get something back in scale of the LBBs. For simplicity, I'm only listing the maximum-damage dealing attack for any monster given multiple attacks in Sup-I. I've also made an executive decision that anything up to 1d6 in Sup-I is unchanged (so the kobold/goblin/rat 1d3 or whatever isn't further reduced, and neither is an orc's 1d6, etc.), but everything else is inverted by the formula. Having back-adjusted the average value, I use another spreadsheet function to suggested the best possible all-d6 damage dice. Here's a snippet from the first few results:

The fifth column over has our formulaic suggestion for damage dice in LBB-scale. The sixth and seventh columns are my manual choices for what I'll use in my own OED house rule games. Orange boxes are entries explicitly noted in LBB Vol-2 text, and I'll leave those fixed in each case (note they're generally quite close to our calculated suggestions, e.g., for giants).

That entire spreadsheet is available here, including suggested conversions of everything in the Sup-I and Sup-II tables. Note that the Sup-I damage table has three distinct parts: (1) revisions for monsters in Vol-2, (2) some damage specs for giant animal types possibly in LBB encounter charts but otherwise without stats, and (3) new monsters appearing in Sup-I itself; these are set off in white, yellow, and green sections of the spreadsheet. Meanwhile, looking at the Sup-II table, it's possible that Arneson was even more unhinged on the issue, e.g. damage of up to 24 points for a sub-1 HD fire beetle, 80 points for a plesiosaur bite, or 150 points for a whale fluke! (A lot of those figures were later reined in by Gygax in the AD&D Monster Manual.)

Finally, I've done a recent revision to the OED Monster Database which (a) edited some damage figures to be consistent with this analysis, (b) added a number of giant and aquatic creatures from Sup-II, and (c) expanded the sourcing/reference information in the last column. All of the damage values can now be rolled on d6 (previously I kept some d8 values in there, as per the supplements). There are currently a number of damage values like 1d6+1 or 1d6+2 (as the LBB Ogre), which shades towards fiddly for me, but I think I'm okay with it for now. Some of the EHD values moved up or down by one or two pips in some cases, as well. We now have 174 monsters in the database. :-)


Sunday Survey: Blind Spellcasting

Recently on the Facebook 1E AD&D I group I asked this question:

Before I asked the question, I scoured the OD&D and AD&D books for a ruling on the subject, and was surprised when I couldn't find any whatsoever. To my knowledge, there's not even any statement that a caster needs to see their target in general! Consider the idiom from Chainmail Fantasy, referenced in text for OD&D spell like fireball and lightning bolt, that attack spells must have "range being called before the hit pattern is placed" (that is, casters specify a distance, not a target). AD&D DMG p. 65 has an example of a caster of fireball needing sight to the area of effect, but no general rule to that effect.

However, all later editions do dictate that casters must have sight of their target. This first appears in the 2E AD&D PHB (Ch. 7): "If the spell is targeted on a person, place, or thing, the caster must be able to see the target. It is not enough to cast a fireball 150 feet ahead into the darkness; the caster must be able to see the point of explosion and the intervening distance." (Note the distinct change from Chainmail/OD&D, with 1E being silent/ambiguous on the issue.) The 3E D&D PHB (Ch. 10) says likewise: "The character must be able to see or touch the target, and the character must specifically choose that target." So this is all very consistent in any edition post-1E, and by their wording would seem to definitively shut the door on a blinded spellcaster being able to get their spells off (excepting a target in touch-contact).

Frank Mentzer actually chimed in on this discussion, saying, "btb if you can't detect/sense/see a target, AND a target is Required, then you can only hit it accidentally. (If you insist, you roll to hit, basically.)" Now, I don't think his recollection of "btb" (by-the-book) is correct, because I can't find anything in 1E materials requiring sight or detection; I can't even find it in his Red Box rules after a brief search. But, again, if it was a common house-ruling and a constant throughout all later editions, then we should be too surprised at some of it bleeding back into our earlier memory banks.

That said, the consensus in the poll that most DMs would give some kind of probabilistic chance of a successful spell seems eminently reasonable as a ruling. It hasn't come up when I've been running a game, but if it did, I think I'd probably lean in the same direction.

Related, today on WanderingDMs live chat (1 PM ET): How do you like your Infravision to work?


Marvel Money

For a couple reasons, we've been playing a few games of TSR's Marvel Super Heroes (FASERIP) recently. It's an enjoyable system but a bit wonky if you scratch the surface on it -- the numerical values for ranks and FEATs (for a variety of real-world assessments) advance in unpredictable jumps and increments. If it had been me, I would have wanted to establish some kind of consistent math at the outset, and then be able to easily slot in outside assessments to the system. On the other hand, I think the DC Heroes game did exactly that, and I don't see as much legacy of love for that system as FASERIP, so what do I know.

But the most obviously broken part of the system was the Resources (money): it was wildly, insanely broken on a rarefied level for gaming systems in my experience -- on par with man-to-man missile fire in classic D&D. Resources was the sub-system that was entirely torn out and replaced with something brand-new in the switch from MSH Basic to Advanced rules. Here it is in the Basic game (Campaign Book, p. 8):

So: An individual of a given rank gets the indicated "resource points" to spend weekly as they see fit, and all items in the game are price in terms of these resource points. Further up the same page there are some sample costs: a knife costs 1r, a plane ticket 10r, an acre of empty land 100r.

Campaign book p. 9 says, "One resource point equals anywhere from 50 to 75 dollars". Let's take $60 as a rounded average. Then we see that the "Typical" salaried employee is making about $360/week, or $18,000/year -- in the same ballpark as the 1984 U.S. median income of $22,415. But on the upper end, a large nation like the U.S. at "Monstrous" rank is indicated as only getting 75r = $4,500/week, or $225,000/year. E.g.: The U.S. government can only pay the salaries for a staff of 12 federal workers total, and absolutely nothing else. In reality, the 1984 U.S. revenue collected was approximately $666 billion, so this figure is over 6 orders of magnitude in error. Lesson: Income advancement isn't linear, it's exponential. 'Nuff said about that.

In the Advanced game released two years later (all editions are by Jeff Grubb), you get the following alteration (Judges' Book, p. 6):

Note that the whole idea of "resource points" is simply gone. Instead the system now uses the standard MSH mechanic of rolling on its Universal Table for success, comparing one's Resource rank versus a Cost rank of similar description. (If the cost is lower, then it's a very easy "green" roll; if equal, a difficult "yellow" roll; if more, then a nigh-impossible "red" roll.) One roll is allowed per game-week. The justification for this is as follows (Player's Book, p. 18):

Resources are modified in the Advanced Set to cut down on the paperwork. As things stood previously in the Original Set, characters gained Resources like money. They had a physical amount of Resource points, and everything cost a certain amount of RPs. This may work for Peter Parker, who has to make the rent every month, but for millionaire Tony Stark who can buy roadsters out of petty cash, this is a bit harder to handle.

While the stated reason is to reduce record-keeping, I'd say the true benefit of this switch is to possibly correct -- or at least obscure -- the prior set's obvious lunacy on the issue. Costs for all items in the game (mostly weapons, vehicles, and headquarters furnishings) are in descriptive ranks, so it's possible that the underlying dollar costs are in a geometric progression. Or not.

In the past I spent a lot of time trying to rationalize this system (I won't recreate all of that here). But it's still going to be very awkward when one puts normal-people and the U.S. federal government on the same list. If we note on the table above that Typical people ($30,000/year) and U.S. Unearthly revenues ($666 billion/year) are 7 ranks apart, then the simplest geometric model would be to have each rank represent a multiplier of the 7th-root of (666 billion/30 thousand) = 7th root of (22 million) = about 11. Let's say it's times-10 per step to make it as simple as possible.

Now, among the problems here is the attempt at equating personal revenues to large companies and countries. Looking at relative values today, the largest company is indeed about one order of magnitude below U.S. revenues. But the wealthiest person should be two orders of magnitude below. A "standard" millionaire should only be one step above a Typical middle-class person (not 4 steps higher, as shown above).

Then if we look at the many copious price charts, a lot of the prices seem to be out-of-sorts with this suggested times-10 model. A simple Axe is Good cost: say the weekly income of a Good-resourced person, so $300,000/50 = $6,000.  A standard Sedan is Remarkable cost, suggesting the weekly revenue of a "large business", i.e., $30 million/50 = $600,000. A large Office Building (30+ floors) is weirdly set at a cost of Shift-Z, that is, 3 steps beyond what any Earthly entity can actually afford (around $6 trillion?). Maybe it's unfair for me to pick on cases like these; I'll stop for now. But you can sort of imagine trying to massage this system and just never getting rid of the many short corners.

Now, one thing I noticed recently is that the 1991 Revised rules, which mostly just edits and repackages the prior Advanced Rules under a different name, has yet another go at this. It gives a fairly brief table of about 50 example Resource ranks (Revised Basic Book p. 41), including salaries and costs of many common comic-book items, and it has the distinct advantage of leaving out the attempt at including national governments. I took that table and did some research to fill in current real-world estimated dollar values, and then a regression on the logarithms of those values, expecting broadly for the standard MSH Resource lunacy appear. But what I found was actually not the most crazy thing I've ever seen:

You can draw a simple straight regression line through that data, including the origin (0, 0), and have it be a 97% correlated match. The indicated model of f(x) = 0.80x means that the cost-multiplier for x ranks should be about 10^(0.8x); since 10^0.8 ~ 6.3, we could say roughly that each rank here represents about ×6 value over the preceding one (perhaps not what I'd have picked tabula rasa, but a more gentle advancement than the previously considered ×10 one). If we pick the 0-rank to be $1 cost, then the ranks represent costs with perhaps lower-bounds of $6, $40, $250, $1500, $10,000, etc. for Feeble, Poor, Typical, Good, Excellent, and so forth (and annual salaries of about 50 times those numbers). The other costs in this version of the rules are -- surprisingly -- kind of consistent with that model. I could find a half-dozen items in the given list off from the real-world estimated value by 2 ranks, but nothing any more than that.

Disclosure: I did put my thumb on the scale here a tiny bit by re-interpreting a few of the items on the list from my first estimates. For example: Low-rank hotel costs I interpreted as per-night, whereas higher-ranked apartments I took as monthly rentals (none are defined one way or another in the published list). For "Private Plane" I used the cost of a multi-engine Piper instead of, say, a corporate jet. I used entry-level "Old Masters" artwork at around $10 million, instead of the world-record $450 million for a da Vinci painting in 2017 (and likewise for examples of "Archaic Texts").

At the top end of this scale, the Mega-corporation does get promoted 2 ranks from Unearthly to Shift-Y (judging from the example of Saudi Aramco's $356B/year revenue; identified as the one real-world example in the Wikipedia Megacorporation article). If we were to include the U.S. federal government, then that would come in at the Shift Z level (based on revenues of $3.5T/year).

In summary: This is now a system that I think I could use for Marvel RPG purchasing power, and be able to estimate and convert real-world prices into in-game mechanics pretty easily, and not think I'm going to stumble over things that are obviously insane and broken on a regular basis. I did massage a small number of the given ranks in those rules and printed a copy for my MSH house rules. Data and analysis in the spreadsheet below if you want to see it. Excelsior!


24 Hours of D&D

Over the July 4th weekend, our Wandering DMs channel livecast a total of 24 hours, 13 minutes, and 16 seconds of D&D play, with us battling for our lives in the lowest depths of Dyson's Delve. That's all available at our Wandering DMs YouTube channel if you want to check it out.

I'm currently crashing and my throat is pretty torn up from yelling in terror and laughing hysterically over the weekend. I'll point you over to Paul's Gameblog for more specifics and links to the individual episode/sessions. Hope your holiday was half as awesome as ours!

More live D&D play than you can shake a magic sword at.


Birthday Broadcast

A quick reminder that the Wandering DMs channel plans to livestream nonstop D&D play (well: some stops for meals and sleep and pool time) all this holiday weekend from the evening of Thursday July 4 to Sunday July 7th. This will feature birthday-boy Paul DM'ing sessions of Dyson's Delve using our OED house rules for original D&D, and myself playing with a number of our friends and family. Simulcast on these fine sites:
As a preview, here's "the story so far" from our sessions one year ago:

Paul's Birthday Business

Hope you can tune in at some point when you've got downtime from your own weekend festivities. Tell us what you think! :-)


Carrion Crawler Coaching

The Facebook AD&D group had an interesting question posed the other day: "Okay DMs, what was the biggest mistake you ever made trying to homebrew something?" Here's one response that caught my eye:

I kind of really love the honesty here. The major reason I love this is as a case-study that even the biggest-name D&D principals didn't always get things right the first time. First, it serves as a great counterexample to the camp of fundamentalist players who argue that everything in a given edition of D&D is perfect, beyond critique or improvement, and intentional in all ways by the original author (although, am I unwise to spend any time responding to that camp?). Second, it serves to highlight that gauging the danger level of a given monster is not something that even the most experienced DMs can do correctly by sight or instinct. Rather; it needs serious large-scale playtesting -- that I would argue needs some component of computer simulation to get to the right scale.

Consider the Arena/Monster Metrics program (and related blog posts here you can search for) that we've developed to assess Equivalent Hit Dice (EHD) measurements for monsters -- results available in the OED Monster Database. Consider that carrion crawlers and other zero-damage monsters were highlighted as particularly broken in Turnbull's MonsterMarks and related a-la-carte point-buy systems. Recall that the stated "monster level" for carrion crawlers jumped around radically in early versions of D&D -- just 2nd level (of 6; say, 33%) in their first appearance in Sup-I (p. 64); then up to 6th level (of 10; so, 60%) by the time of the AD&D  DMG (p. 178). I think that's the single biggest adjustment for any individual monster between those editions.

Regarding Mentzer's comment above, I asked a follow-up question: "I've seen some people play that a crawler can only attack one PC at a time (w/all 8 atks), others they can attack 8 PCs at once. How'd you play that?" His reply:

So that's clearly "more than one", normally around 3-4 from how I read that. Interestingly, if the designers had a systematic model like our Arena program available, then the danger of carrion crawlers would have been immediately evident. If I run the Monster Metrics assessment with the crawler allowed 8 attacks against different opponents, it estimates an EHD value of 12 (i.e., roughly 50% likely to win a fight against 4 3rd-level fighters, or 3 4th-level fighters), putting it in league with the top 6th-level bracket in OD&D (comparable to a chimera, gorgon, balrog, etc.). That's why in my own game for some time I've actually house-ruled them to halve their attacks, i.e., a total of 4 attacks -- basically the same as what Mentzer suggests for targetable opponents here. At this level carrion crawlers are estimated to be EHD 9, or about 5th level in OD&D terms.

Anyway, big props to Mr. Mentzer for this important peek behind the screen, and the not-too-surprising lesson that we can always continue to make improvements to our game art.

Don't forget about our July 4th game with WanderingDMs on YouTube and Twitch: broadcasting live play all weekend, four days straight! (Starts Thursday night.)