1 Golem, 3 Ways, 7 Challenge Ratings -- HELP!

[Garnfellow]

I've designed and advanced a lot of monsters. And I fully understand that assigning challenge ratings is a bit of a black art. But I've never had such a wide range between projected CR and actual monster power. I'd greatly appreciate any help with assigning an accurate CR here.

Let's start with the brass golem. This is a Large Construct from the Monster Manual II. It is AC 31, 16 HD, has a Str of 20 and an Intelligence score. It comes equipped with a Large-sized +3 wounding greataxe, which does 3d6+10/x3. It has standard golem magic immunity, DR 10/adamanitine, and a 1/day maze spell-like ability. The MMII gives it a CR of 10. Originally designed for 3.0, the 3.5 update gives the golem +30 hp for size and six new feats, mostly from the Power Attack chain.

Since I needed a CR 15 version of the golem, I advanced the creature from 16 HD and Large to 33 HD and Huge size: http://www.enworld.org/forums/showthread.php?p=1916129#post1916129 . Using the Monster Manual 3.5 guidelines, the CR should increase by +1 for every 4 additional HD, with an additional +1 for the size increase. So 10 + 17/4 + 1 = 15. But then I began looking at the results -- 33 HD creates one bad muthah, indeed. A single attack from this advanced construct will easily hit even the best armored PC in my little band, and even average damage will be lethal. And with a 15 ft reach and its magic immunity, this thing has DANGER! written all over it.

Since the golem has a 3 Int and thus could take class levels, I then tried a different tack. Rather than advancing by HD, I added Fighter levels: http://www.enworld.org/forums/showthread.php?p=1917659#post1917659 . Assuming fighter would be an associated class, +5 levels = +5 CR. Although by the book this also was a CR 15 monster, it was clearly much less powerful than the 33 HD version. But it also looked a little tougher than CR 15, too.

So I used a couple of other techniques to estimate the CR for all three versions.

1. The Savage Species method. Take the monster's hp, divide by 4.5, divide again by 3.0, then add +1 for each special ability. The preliminary result is then tested against average NPC ACs, attack rolls, and saving throws.

2. The Grim Tales Creature Creation guidelines, designed by EnWorld's own Upper_Krust. This is a much more detailed system that adds up all the monster abilities and traits and outputs a CR.


As call be seen, the results of these different calculations are all over the board.

                  | Original |          |          |
                  |   Brass  | Advanced | 5 levels |
Method            |   Golem  | to 33 HD | fighter  |
===================================================
Monster Manual    |    10    |   15     |    15    |

Savage Species    |    13    |   22     |    16    |

Creature Creation |    13    |   21     |    18    |

Average           |    12    |   19     |    16    |

So, what should the real CR rating be for these monsters? Bonus question: why is there such a huge spread between the results calculated be different methods, which normally are somewhere within 1 or 2 CRs of each other.

 

[drnuncheon]

One of the reasons the MM system tends to 'lowball' creatures advanced by HD is because frequently, they have fewer and less effective special abilities than other monsters. For example, poison is pretty ineffective against most high-level parties thanks to high Fort saves and various spells.

In the case of the brass golem the special abilities (especially the magic immunity) are generally still worthwhile - but any 15th level party that can't easily bypass it to do whatever they've got to do has serious issues. (That's assuming that the goal isn't to kill the golem outright.) Heck, fifth-level spells used creatively ought to keep it out of your way - the various wall of spells, transmute rock to mud, or even teleport...

 

[Garnfellow]

In the case of the brass golem the special abilities (especially the magic immunity) are generally still worthwhile - but any 15th level party that can't easily bypass it to do whatever they've got to do has serious issues. (That's assuming that the goal isn't to kill the golem outright.)

You may have just explained away a lot of the puzzlement I've had over certain WotC Challenge Ratings. I've always assumed the CR testing was done under duel-to-the-death scenarios. But if you widen the definition of victory conditions beyond kill or be killed, that /does/ help make sense of many problematic CRs.

Do you know any quotes from the designers or online articles that help explain this piece of their methodology?

 

[drnuncheon]

The section of the DMG on 'Experience Awards' (36-37 in the 3.5 version) talks about this - there's an example about sneaking past a sleeping minotaur. Since most of the section focuses around defeating the monsters in combat, though, it's easy to miss.

Edit: That said, even in combat the brass golem has serious flaws, the big one being mobility. With it's sole method of movement being a 30' land speed with no running ability, any PCs on the ground should easily be able to outdistance it, and those that can take to the air can attack with relative impunity.

Also, IIRC 'magic immunity' is basically treated as 'infinite SR' (at least in 3.5), which means that various sorts of conjuration spells ought to work on it just fine (although incendiary cloud is not recommended).

That last brings up an interesting idea for a slight hijack - you could also increase the EL of the encounter by use of the environment. Brass golems heal damage due to fire whereas most PCs do not, and fire breaks the slow effect of an electricity spell. Consider having the golem deliberately set off an incendiary cloud trap, flaming oil, or something similar. Maybe even put it in the center of a maze made of walls of fire. It'll heal the golem, hurt the PCs, and make the fight all the more exciting and memorable.

J

 

[dcollins]

So, what should the real CR rating be for these monsters? Bonus question: why is there such a huge spread between the results calculated be different methods, which normally are somewhere within 1 or 2 CRs of each other.

I think the key to your problem is simply that the original MMII Challenge Rating is incorrectly measured, and should be higher. Your other measurements both think it should be CR 13. If that were the case, then your advancements per MMII would have been approximately CR 18, which is more in line with the others.

As a side note, I would never consider giving class levels to a monster that didn't say "Advancement: By class". Yes, the rules mechanics can statistically handle it, but for most monsters (like automoton golems) the idea is rather cartoonish.

Do you know any quotes from the designers or online articles that help explain this piece of their methodology?

You're not going to find any quotes like that because frankly that's not how the designers do it. Whenever I've read an anecdote about a WOTC playtest it's a flat-out duel to the death.

The quote from the DMG in the Rewards chapter (sneaking past a minotaur) is usually incompletely quoted. What it really says is that a flat-out duel to the death is "simple to do" and clearly gets XP. Sneaking past a minotaur is harder; the answer is "... probably yes. It's up to you to make such judgements."

So in short, the normal case is killing monsters. The abnormal case, getting by them, is maybe-yes, maybe-no according the circumstances and the DM.

 

[Marimmar@Home]

One reason for Constructs to only gain 1 CR / 4 HD is the lack of Skills and Feats. Adding these should place the Construct in the 1 CR / 3 HD, thus we get roughly a CR 17 from adding 17 HD and a Size increase. The Grim Tales method produces way to high CRs. My players would tear apart any monster created by Upper_Krust's rules since they tend to be wusses at their respective CR-levels.

~Marimmar

 

[drnuncheon]

My players would tear apart any monster created by Upper_Krust's rules since they tend to be wusses at their respective CR-levels.

If I remember right, UK's CR rules don't produce a CR that you use in the same way as the WOTC CR. (It really ought to have been called something different to avoid confusion.) I never muddled through the rules well enough to get a sense for the relationship, though.

J

 

[Garnfellow]

Also, IIRC 'magic immunity' is basically treated as 'infinite SR' (at least in 3.5), which means that various sorts of conjuration spells ought to work on it just fine (although incendiary cloud is not recommended).

One of the PCs is a 15th level conjurer. I suspect his earth element elephants will be able to do some decent damage.

That last brings up an interesting idea for a slight hijack - you could also increase the EL of the encounter by use of the environment. Brass golems heal damage due to fire whereas most PCs do not, and fire breaks the slow effect of an electricity spell. Consider having the golem deliberately set off an incendiary cloud trap, flaming oil, or something similar. Maybe even put it in the center of a maze made of walls of fire. It'll heal the golem, hurt the PCs, and make the fight all the more exciting and memorable.

Man, I like the way you think.

 

[Garnfellow]

I think the key to your problem is simply that the original MMII Challenge Rating is incorrectly measured, and should be higher. Your other measurements both think it should be CR 13. If that were the case, then your advancements per MMII would have been approximately CR 18, which is more in line with the others.

I think this is an important piece of the puzzle. The 3.0 version had no feats, but the 3.5 version gets six and some extra hit points. CR 10 may be appropriate for the 3.0 version, but the 3.5 CR should probably be bumped to reflect this significant change.

As a side note, I would never consider giving class levels to a monster that didn't say "Advancement: By class". Yes, the rules mechanics can statistically handle it, but for most monsters (like automoton golems) the idea is rather cartoonish.

Yeah, I had similar reservations, which was why I started out with adding HD. A couple of things suggested using class levels, though. There's a nimblewright in a recent Dungeon adventure with class levels. And I'm toying conceptually with a feat that would allow a construct creator to treat her intelligent constructs something like cohorts, and "feed" them XP, sort of how Monte Cook handles intelligent magic items in the Book of Eldritch Might III.

If you never add class levels to monsters that don't specically say "Advancement: by class," you're missing out on some good fun with undead. There are some nice prestige classes specifically for undead out there, and a ghoul with some rogue levels can be nasty.

So in short, the normal case is killing monsters. The abnormal case, getting by them, is maybe-yes, maybe-no according the circumstances and the DM.

Well, if the designers were considering evasion as an acceptable out, it might explain the seemingly low CR on that new undead in the MMIII with 20HD and the drowning ability.

 

[dcollins]

Well, if the designers were considering evasion as an acceptable out, it might explain the seemingly low CR on that new undead in the MMIII with 20HD and the drowning ability.

But a far more likely hypothesis is simply that the quality of CR estimations (and playtesting) has been going down every year since the release of 3.0 (and reduction of WOTC design & editing staff).