"How Borked Is CR?" Test Thread. Arcanist, Cosi invited. Beheld too.

[johnbragg]

The challenge was suggested to pick a selection of monsters and evaluate if their CR is right. I'm picking CR 5 and CR 10, because that's around the end of low level play where there is some mundane-caster balance (except for Druids) and around the end of mid-level play where Angel Summoner and BMX bandit becomes really obvious.

I'm using the Pathfinder table because it's easily available. (I've never seen an argument that Paizo was much better, or worse, than WOTC at CR, and I found it first on google.)

EDIT: Replaced with SRD CR 5 and CR 10 list

CR 5 air elemental (large), army ant swarm, basidirond, basilisk, bearded devil, cloaker, cyclops, dire lion, djinni, earth elemental (large), fire elemental (large), giant frilled lizard, giant moray eel, gibbering mouther, green hag, half-celestial unicorn, ice golem, manticore, mummy, nightmare, ochre jelly, orca, phase spider, troll, water elemental (large), winter wolf, wraith. 27 CR 5 monsters.

CR 10 bebilith, brachiosaurus, clay golem, couatl, fire giant, giant flytrap, guardian naga, rakshasa, red dragon (young), silver dragon (young), white dragon (adult). 11 CR 10 monsters.

A neutral party has selected, fairly randomly, a monster list.

EDIT: I missed the edit where Cosi said he was working on a 5-random-monster evaluation of the CR system.

EDIT: I declare, decree and stipulate that, despite using a Pathfinder list of monsters, only 3.5 rules shall apply. I'm pretty sure that all of the listed monsters appear in WOTC Monster Manuals. If not, I will remove that creature from the discussion.

[Beheld]

I'm using the Pathfinder table because it's easily available. (I've never seen an argument that Paizo was much better, or worse, than WOTC at CR, and I found it first on google.)

I object on the grounds that I was a primary party in the argument and that I can't have any part of the conversation if it's Pathfinder so we should use 3.5. But we can still just use CR 5 and CR 10 from 3.5.

There's also the issue of testing mechanism, my go to would be to make a party of Rogue, Wizard, Cleric, and Big Dumb Fighter.

Not sure how much actual optimization would be necessary or wanted, but certainly Rogue, Wizard, Cleric can just take 5 and 10 levels in those classes and be fine, Big Dumb Fighter, if Core, should probably be Barbarian 1/Fighter 2/Paladin 3/Moar Barbarian Levels and/or Ranger 2. Truthfully, we could just replace him with a Druid especially at level 10, since we know regular fighting types are just vehicles for Wizard spells at that point.

Casters should probably not use minion tricks like Dread Warrior, Animate Dead, Druid Wilderness Force, or Shadow Army.

[Florian]

Do me a favor and define what "beating a challenge" means in this context.

To explain that: Sometimes it´s enough to hit something hard enough to make it retread, so your way is clear, other times, you have to beat something to a bloody pulp as it doesn´t know the concept of defeat.

[Beheld]

Do me a favor and define what "beating a challenge" means in this context.

To explain that: Sometimes it´s enough to hit something hard enough to make it retread, so your way is clear, other times, you have to beat something to a bloody pulp as it doesn´t know the concept of defeat.

In this case, beating challenges too easily is also a fail for CR system, though harder to manage. But since we are talking about the CR system, beating most of the monsters is going to result in their deaths, or them running away mostly dead.

[johnbragg]

IS it feasible to look at the list, and declare some number of those monsters, "Yeah, that's about right at that CR?" Subject to contradiction followed by argument and/or testing. But I think we've played the game enough to say, yes, that's a solid encounter for a "typical" 5th or 10th level party.

(I assumed at the beginning that the discussion would start with a list of "these look right, not those." I didn't think I was asking people to simulate almost 40 encounters for this project).

[Florian]

In this case, beating challenges too easily is also a fail for CR system, though harder to manage. But since we are talking about the CR system, beating most of the monsters is going to result in their deaths, or them running away mostly dead.

Just interested. Most PF AP monsters have a "morale" entry listed that specifies when that exact monster is considered "beaten". Remember the talk we had a while ago about the difference between CR and EL.

IS it feasible to look at the list, and declare some number of those monsters, "Yeah, that's about right at that CR?" Subject to contradiction followed by argument and/or testing. But I think we've played the game enough to say, yes, that's a solid encounter for a "typical" 5th or 10th level party.

Looking at those lists, there´s nothing on them that makes me excited in any way. That mainly means that the are "on spot" for their CR and no real challenge at all besides being road-bump that should waste some resources and that´s all.

[Cosi]

Honestly, none of that looks particularly bad. The only thing that look potentially worrying on the CR 5 list are the Orche Jelly (oozes are all puzzle monsters, OTOH they move slower than you have have no ranged attacks) and the Wraith if you don't have a magic weapon yet somehow. There's nothing particularly lethal on the CR 10 list, but there is some stuff that's probably under-CRed (golem, melee bruisers in general). Possibly the dragons, but problems with those are less issues with CR in general and more issues with the designers using a different formula for dragon CRs (most stuff is CRed to be equal to one PC of level X, but dragons were supposed to be equal to four).

[eggynack]

I don't think you should use PF monsters. My understanding is that their CR is a bit closer to good. Here's a source for 3.5 monsters, one that can be sorted by CR. Should do the job just fine, and isn't assailable on the grounds of completely different sources. If you want to use only some monsters out of the full set, I'd just use this random number generator, with the total number of monsters listed at the CR as the top end. For example, CR 5 has 107 monsters listed, and I got a 48, so that'd point to an adult kython. The monster finder doesn't have all sources, but it has a lot of the big ones, so it does the job well enough. No reason to toy around with crossing the edition streams, and allowing in all that inclusion bias and general oddity.

[johnbragg]

I don't think you should use PF monsters. My understanding is that their CR is a bit closer to good. Here's a source for 3.5 monsters, one that can be sorted by CR. Should do the job just fine, and isn't assailable on the grounds of completely different sources. If you want to use only some monsters out of the full set, I'd just use this random number generator, with the total number of monsters listed at the CR as the top end. For example, CR 5 has 107 monsters listed, and I got a 48, so that'd point to an adult kython. The monster finder doesn't have all sources, but it has a lot of the big ones, so it does the job well enough. No reason to toy around with crossing the edition streams, and allowing in all that inclusion bias and general oddity.

OK, later tonight I'll revise the list to the CR 5's and 10's from the SRD. I think the best test of the system is on the most common monsters. (I could be wrong--it's the uncommon monsters where you maybe need it more. But I'll start with the CR 5 and 10 SRD.)

[Florian]

OK, later tonight I'll revise the list to the CR 5's and 10's from the SRD. I think the best test of the system is on the most common monsters. (I could be wrong--it's the uncommon monsters where you maybe need it more. But I'll start with the CR 5 and 10 SRD.)

Consider two things from PF first and then work from there:
- There´s an actual chart on how to allocate stats and such on monsters. You can find it in Pathfinder Unchained.
- There´s a marked difference between "solo" and "support" type of monsters and you notice that difference.

Note that I don´t say that they managed it better, but at least they learned some lessons.

Look at older 3,5/3,0 monsters and notice that the both over and undervalue some things:
- Monsters that could do longer-lasting or permanent damage, like undead, are generally overvalued whilst being pretty easy to destroy
- Simpler monsters are generally undervalued because of that simplicity. That Damn Crab? I chose you!

[eggynack]

OK, later tonight I'll revise the list to the CR 5's and 10's from the SRD. I think the best test of the system is on the most common monsters. (I could be wrong--it's the uncommon monsters where you maybe need it more. But I'll start with the CR 5 and 10 SRD.)

You may want to broaden things a bit more. Monsters from the MM II, III, and the fiend folio are some of the ones most often cited as objectionable, I think. Ultimately, what you search depends on your hypothesis. If your hypothesis is that CR isn't borked, you're going to want to sample more from stuff like the MM II. If your hypothesis is that it is borked, then I dunno what book is considered least borked. Might need a pretty broad sample in that situation.

[Florian]

In addition, CR does not really function without EL. The fun y thing is, CR is not the important part here, EL is. As that is tailored towards your actual player and the gm should know those, that tends to be mostly accurate unless you, as a gm, where pretty much drunk or stoned when building an encounter.

[Troacctid]

MM2 and Fiend Folio are 3.0 material, though.

[eggynack]

MM2 and Fiend Folio are 3.0 material, though.

Still a part of the game though, and an oft criticized one. I dunno. I guess I fall a bit more in the, "CR probably isn't so bad, so the sample should pull more from bad stuff," camp. Really though, is there any reason to cut off what 3.5 legal books we're sampling from at all? As long as we look at randomly selected creatures, and not all creatures, it shouldn't take much longer than using just a few books at all, and won't have all that much bias.

[Beheld]

MM2 and Fiend Folio are 3.0 material, though.

I think if you are really looking at "DM pulls a monster out of his hat to play against the party" then MM 1, MM II, MM III should all be included.

I can't remember MM IV off the top of my head, but one of them I recall having very few actual monsters, because most everything was just "X with class levels" we should use the other one.

Certainly if I want a Demon or Devil, I check the Fiendish Codexes and Fiend Folio, and if I want undead, I check Libris Mortis, and if I want a Dragon, I check Draconomicon, but if I just want "a monster" I check the MMs.

I'm fine with making this a collective project that takes months if we get something actually useful out of it by doing all the monsters in all those books of some specific CRs. But I can certainly see why others may not want to do that.

In addition, CR does not really function without EL. The fun y thing is, CR is not the important part here, EL is. As that is tailored towards your actual player and the gm should know those, that tends to be mostly accurate unless you, as a gm, where pretty much drunk or stoned when building an encounter.

Ugh... Not this nonsense again. It was terrible last time too, but it's literally infinity times more terrible this time. We are specifically talking about how well CR allows you to pull specific monsters of a CR up, and say "this monster is good enough for this party, and I don't need to do any other work" Saying "CR doesn't matter because all CR 10 monsters are EL 46 encounters because my party is crappy" is meaningless on a grand scale.

We are specifically testing the ability to pull out CR monsters and run them against parties without work, the CR system is the only thing that matters for that at all. Yeah, if the monsters says it comes in packs or whatever, you do the calculations for EL based on the table in the DMG and that's it, because the entire purpose is to test the ability to pull monsters out and use them without having to rebalance things.

[JNAProductions]

The Ekolid isn't very well-CRed. Not sure what book it's from, but it's much tougher than CR 3 should be.

Edit: Or is it CR 4? Either way, much too powerful for it's CR, 3 or 4.

[johnbragg]

Still a part of the game though, and an oft criticized one. I dunno. I guess I fall a bit more in the, "CR probably isn't so bad, so the sample should pull more from bad stuff," camp. Really though, is there any reason to cut off what 3.5 legal books we're sampling from at all? As long as we look at randomly selected creatures, and not all creatures, it shouldn't take much longer than using just a few books at all, and won't have all that much bias.

I'd argue that looking at the entire CR 5 and CR 10 sample from the SRD, and sorting by "everyone agrees this CR is okay" "everyone agrees this CR is wrong" and "this CR is disputed" is an important piece of information.

Name CR Size Type (subtype) HD Al. Environment Home plane Book Achaierai 5 Large outsider(evil, extraplanar, lawful) 6 LE Any Any Lawful SRD Animated Object, Huge 5 Huge construct 8 N Any Material SRD Arrowhawk, Adult 5 Medium outsider(air, extraplanar) 7 N Any Elemental Air SRD Barghest, Greater 5 Large outsider(evil, extraplanar, lawful, shapechanger) 9 LE Any Any Evil SRD Basilisk 5 Medium magical beast 6 N Warm Deserts Material SRD Cloaker 5 Large aberration 6 CN Any Underground Material SRD Devil: Bearded Devil (barbazu) 5 Medium outsider(evil, extraplanar, lawful) 6 LE Any Baator (Nine Hells) SRD Dire Lion 5 Large animal 8 N Warm Plains Material SRD Elemental, Air, Large 5 Large elemental(air, extraplanar) 8 N Any Elemental Air SRD Elemental, Earth, Large 5 Large elemental(earth, extraplanar) 8 N Any Elemental Earth SRD Elemental, Fire, Large 5 Large elemental(extraplanar, fire) 8 N Any Elemental Fire SRD Elemental, Water, Large 5 Large elemental(extraplanar, water) 8 N Any Elemental Water SRD Genie, Djinni 5 Large outsider(air, extraplanar) 7 CG Any Elemental Air SRD Gibbering Mouther 5 Medium aberration 4 N Any Underground Material SRD Hag: Green Hag 5 Medium monstrous humanoid 9 CE Temperate Marshes Material SRD Hieracosphinx 5 Large magical beast 9 CE Warm Deserts Material SRD Hydra, 6-headed 5 Huge magical beast 6 N Temperate Marshes Material SRD Manticore 5 Large magical beast 6 LE Warm Marshes Material SRD Monstrous Spider, Huge 5 Huge vermin 8 N Temperate Forests Material SRD Mummy 5 Medium undead 8 LE Any Material SRD Nightmare 5 Large outsider(evil, extraplanar) 6 NE Any Any Evil SRD Ochre Jelly 5 Large ooze 6 N Temperate Marshes Material SRD Phase Spider 5 Large magical beast 5 N Warm Hills Material SRD Rast 5 Medium outsider(extraplanar, fire) 4 N Any Elemental Fire SRD Ravid 5 Medium outsider(extraplanar) 3 N Any Positive Energy SRD Shadow Mastiff 5 Medium outsider(extraplanar) 4 NE Any Shadow SRD Skeleton, Ettin 5 Large undead 10 NE Cold Hills Material SRD Snake, Constrictor, Giant 5 Huge animal 11 N Warm Forests Material SRD Spider Eater 5 Large magical beast 4 N Temperate Forests Material SRD Sprite: Pixie 5 Small fey 1 NG Temperate Forests Material SRD

[johnbragg]

And the CR 10 list

Name CR Size Type (subtype) HD Al. Environment Home plane Book Animated Object, Colossal 10 Colossal construct 32 N Any Material SRD Couatl 10 Large outsider(native) 9 LG Warm Forests Material SRD Demon: Bebilith 10 Huge outsider(chaotic, evil, extraplanar) 12 CE Any Abyss SRD Formian Myrmarch 10 Large outsider(extraplanar, lawful) 12 LN Any Mechanus SRD Giant, Fire 10 Large giant(fire) 15 LE Warm Mountains Material SRD Golem, Clay 10 Large construct 11 N Any Material SRD Hydra, 11-headed 10 Huge magical beast 11 N Temperate Marshes Material SRD Hydra, cryo/pyro, 9-headed 10 Huge magical beast 9 N Cold/Warm Marshes Material SRD Monstrous Scorpion, Gargantuan 10 Gargantuan vermin 20 N Warm Deserts Material SRD Naga: Guardian Naga 10 Large aberration 11 LG Temperate Plains Material SRD Rakshasa 10 Medium outsider(native) 7 LE Warm Marshes Material SRD Razor Boar 10 Large magical beast 15 N Temperate/Warm Forests/Mountains Material SRD Salamander, Noble 10 Large outsider(extraplanar, fire) 15 E Any Elemental Fire SRD

[Florian]

@Beheld.

As usual, you´re giving a very "Technocrat" answer to it. EL is a part of the core rules as it actually explain why and how CR functions. Knowing you, you will ignore that as a large part here is "eyeballing" it and that´s a thing you don´t seem to do well. Re-read the rules and get over it.

[johnbragg]

Does anyone object to approaching the CR 10's first? Since there are only 13 of them, we might get somewhere before the entire argument and conversation goes in the toilet.

I'll try to use this post to track agreement and disagreement.

All
Florian

1. Colossal Animated Object
Florian Aye, Beheld "Depends"

2. Couatl
Florian Aye, Beheld Aye

3. Bebilith
Florian Aye, Beheld Aye

4. Formian Myrmarch
Florian Aye, Beheld Aye

5. Fire Giant
Florian Aye, Beheld Aye

6. Clay Golem
Florian Aye, Beheld Nay

7. Hydra, 11 headed
Florian Aye, Beheld Aye

8. Hydra, Pyro/cyro
Florian Aye, Beheld Maybe

9. Gargantuan Scorpion
Florian Aye, Beheld Nay

10. Guardian Naga
Florian Aye, Beheld Nay

11. Rakshasha
Florian Aye, Beheld Nay

12. Razor Boar
Florian Aye, Beheld pass

13. Noble Salamander
Florian aye, Beheld Aye

[Florian]

Looking at those lists, there´re some critters that make Sneak Attack harder but nothing that can´t be ultimately countered. There´s none that can do any serious drain or other lasting effects and none where you´d need special attack modes to put them down.

Only edge cases I can see so far are the Pixie and Hydras as they have very good defensive modes that you´d have to beat/disable first. The Gibbering Mouther might actually be hard for its CR, though.

[johnbragg]

Looking at those lists, there´re some critters that make Sneak Attack harder but nothing that can´t be ultimately countered. There´s none that can do any serious drain or other lasting effects and none where you´d need special attack modes to put them down.

Only edge cases I can see so far are the Pixie and Hydras as they have very good defensive modes that you´d have to beat/disable first.

Shall I put you down as accepting all but the Pixie and Hydras as good CR 10s? Or are you saying that they're a little soft to be CR 10?

[Beheld]

As usual, you´re giving a very "Technocrat" answer to it. EL is a part of the core rules as it actually explain why and how CR functions. Knowing you, you will ignore that as a large part here is "eyeballing" it and that´s a thing you don´t seem to do well. Re-read the rules and get over it.

I eyeball things just fine, thanks for the baseless insults. The point is that if what you are testing is the ability to pull a monster out of a book and use it unchanged, declaring that all CR 10 monsters are an EL between 1 and 468 is not helpful, and does nothing at all for the conversation, and no matter how much you grind your willful obtuseness about this point into the discussion, it will never become even remotely relevant.

If eyeballing a CR 10 monsters results in you saying "this monster is clearly too strong for a 10th level party" 50% of the time, then clearly the CR system is garbage and we can move on with our life. If eyeballing results in deciding that it runs fine against a level 10 party 99% of the time, and then when you play it against those parties, it turns you you were right about that eyeball 95% of the time, then CR system is 94.8 percent accurate, and that's really good and we should all throw a party about how great the CR system is.

[Florian]

Shall I put you down as accepting all but the Pixie and Hydras as good CR 10s? Or are you saying that they're a little soft to be CR 10?

Define soft in that context.
I don´t see anything on that list that you can´t solo with any given class and without any foreknowledge besides the Pixie and Hydra.

Let´s talk about CR a bit here: "Equal CR" is not really a challenge. It´s a quarter of a challenge and that´s how it´s supposed to be. It should be "soft". EL+4 is what a challenge should be and only if it doesn´t turn into a binary encounter.

[Florian]

I eyeball things just fine, thanks for the baseless insults. The point is that if what you are testing is the ability to pull a monster out of a book and use it unchanged, declaring that all CR 10 monsters are an EL between 1 and 468 is not helpful, and does nothing at all for the conversation, and no matter how much you grind your willful obtuseness about this and every other point into the discussion, it will never become even remotely relevant.

If eyeballing a CR 10 monsters results in you saying "this monster is clearly too strong for a 10th level party" 50% of the time, then clearly the CR system is garbage and we can move on with our life. If eyeballing results in deciding that it runs fine against a level 10 party 99% of the time, and then when you play it against those parties, it turns you you were right about that eyeball 95% of the time, then CR system is 94.8 percent accurate, and that's really good and we should all throw a party about how great the CR system is.

Pissed because someone uses an aggressive tone with you?

Or might it actually be because you do not actually want to understand the CR to EL transformation because what it implies?

Make no mistake, I hold the system mastery you show in high regard, but I also think that it exactly showcases what EL is about.

CR works because of "no assumptions". It can´t be a topic if defensive options can be bypassed or offensives options can be defended against. That is what we have EL for.

CR breaks mostly not because of monsters but because of things with character levels using builds. At that point, all rules are off.

[Troacctid]

All we really need to do is compare them against each other, yeah? The most important thing is for them to be consistent.

If we want to throw PCs (or NPCs with class levels) in the mix, I suggest sample PC builds from PH2 as examples of unoptimized characters.

[Florian]

All we really need to do is compare them against each other, yeah? The most important thing is for them to be consistent.

If we want to throw PCs (or NPCs with class levels) in the mix, I suggest sample PC builds from PH2 as examples of unoptimized characters.

Nah, that´s petty much not how it works.
Simply look at what options are available and what is needed to counter them. If both are en par, CR is good. CR breaks when there´s a disparity here.

Take a look at the Pixie, CR5. The main issue here is invisibility. Glitterdust and See Invisibility are readily available options, so the CR stands.

The really interesting cases are when options are not available at the level indicated by the CR.

[Beheld]

If we want to throw PCs (or NPCs with class levels) in the mix, I suggest sample PC builds from PH2 as examples of unoptimized characters.

I recommend not using those garbage builds and just agreeing to really basic characters. I think "Barbarian 1 with extra rage, then levels of Fighter" or whatever is just so much better than "Whirlwind attack Barbarian" a character that two weapon fights with a weapon he isn't proficient with and takes skill focus Tumble. Or "all the rogue builds" zero of whom actually try to get extra attacks, and one of who explicitly tries to give up all his iterative attacks to attack once.

Or might it actually be because you do not actually want to understand the CR to EL transformation because what it implies?

I understand perfectly what the "transformation" is about, both what it actually is about, and also what you claim it is about. But it doesn't matter. This is not a conversation where you whine about how CR 10 monsters are all EL 46 encounters so your fighter feels better, this is about how good the results are for DMs to pull a random CR X monster out, and play it against a standard/average/random Party level X.

If you want to declare to high heavens that no one should ever ever ever ever ever be allowed to do that, and that all DMs should burn their books and melt their dice before filling in a random encounter with a random monster of CR X, you can do that somewhere else. But whether or not people should do it is not a question that is even remotely prior to figuring out how it works.

If it turns out the answer to the question is "don't do that, because it's terrible" that will be great evidence for your completely different argument that has no bearing on the outcome of this discussion whatsoever. But when people are explicitly seeking to test the ability of DMs to pull out CR X monsters and use them, declaring that it is impermissible for this to be tested because you have authoritatively decided no DM is allowed to do that is not useful or helpful at all, it is just you demanding that we talk about a completely different topic because you don't want to see the results of this one. Tough luck.

[Beheld]

1. Colossal Animated Object

This is a Puzzle Monster. It's colossal. Either you encounter it somewhere where you don't care, or your express goal is to stop it from flattening [place], or it's in a giant cavern underground where it can still reach all the sides. In case a) you just ignore it and it costs no resources, and is arguably too easy, in case b) and c) it's just a question of resources expended. You can probably get by with a Mirage Arcana if you have that ready in b), and c) can be bypassed with some kind of teleport or burrowing. On the other hand, you might have to explode the poor thing to save the city, or stop and reprepare spells to get past the cavern, both of which might cost a great deal in resources (time or spells or HP or limited use abilities) that you might care about.

2. Couatl
Just harrasses you until you figure out a way to find it and kill it, but all the abilities are appropriate around this CR. Check.

3. Bebilith
It's a good ambush monster, and then once it's ambushed you blow it up. Not bad. Check.

4. Formian Myrmarch
Truthfully, these aren't really for level 10 PCs. Level 13 or so PCs are supposed to face a war party more than just one of these. but as far as it goes, if it teleports in which some charmed muscle it's a viable PC hunter encounter, where they pissed of some Formians in the past, and this is the revenge strike. It's a pretty decent encounter, I mean, nothing the Formian itself does to you is a big deal, but has the ability to stalk you, deliver some enemies, action deny you, harass you a bit, and then run away. PCs aren't going to lose, but they are going to want to plane for future assaults, and that's a good encounter. Check+

5. Fire Giant
It's a CR 10 Bruiser that throws rocks at you, it's about as good as can be expected from that at this level. Check+

6. Clay Golem
It needs a master... Alone this thing is either chump change or it's going to beat up the PCs with unhealable damage, depending on whether or not they know how to deal with it (mostly ignore it and lock it in a box of stone or illusion). With a Master it's actually fine, but on it's own I have to give it: X.

7. Hydra, 11 headed
Fine. A bit subject to plan, don't be within 40ft and be faster than it, but acceptable closet troll.

8. Hydra, Pyro/cyro
If Resist Energy negates the breathe weapon, this is probably an acceptable closet troll/ambush monster, if not, it's probably and X.

9. Gargantuan Scorpion
Puzzle Monster, and too big, I vote X.

10. Guardian Naga
Check.

11. Rakshasha
These guys are just hard as but to kill, but don't really do anything either. I guess it's fine as a weird tank/manipulator, so I give it: Check.

12. Razor Boar
Pass? I didn't see that in my search and it doesn't sound interesting enough to look up right now.

13. Noble Salamander
Check.

8 Definite Checks.
1 Pass
2 Eithers, Pyro/Cryo Hydra and Animated Object.
1 probably X Clay Golem.
1 Definite Mega X, Scorpion.

[Cosi]

There should probably be separate categories for under-CRed and over-CRed monsters. The fact that high level bruisers basically stop mattering once you get flight means that those monsters aren't a meaningful fight, but nothing particularly bad happens to the game if you throw one at the party - they just get some easy XP and loot. On the other hand, the fact that low level parties have a lot of trouble dealing with incorporeal enemies means that an Allip can TPK far too easily.