Max Exceptions Debate Thread

The issue of the "correct" compatibility with regards to max exceptions is pretty complex and full of pitfalls. I could have a field day with this but I'll just state a few points I want to make, without going through a thousand examples.

Sunder (2010 release) is a prime example of a wad that has various issues when played with complevel 2, none of which actually prevent you from exiting a map. Can you guess which source ports are recommended in the text file?... no, it's not Boom-compatible source ports.

I don't agree with the general notion that if you use a compatibility that's different from the one stated by the author(s), you're using the "wrong" compatibility. Instead, I think if the authors state a certain compatibility and there are issues that can be solved by using a different compatibility, they are stating a "wrong compatibility".
That said, it gets complicated when you consider that:
- complevel 9 sucks because of monsters falling off and possibly becoming impossible or a pain to kill
- megawads where it's literally 1 issue in the whole wad that gets resolved by using a different compatibility (D2INO)
- wads that were released before Boom even existed that have issues that could be solved by using Boom compatibility (dot, strain)

Nevertheless, let me ask a few questions anyone who thinks that the compatibility stated by the authors should be preferred (let's assume complevel 2 here):
- say a cyberdemon doesn't teleport in unless you use cl9, do you think cl2 should still be used in that case?
- say most monsters in the map don't teleport in unless you use cl9, do you think cl2 should still be used? (this is actually exactly the issue in sckrpnch MAP09, which is otherwise a perfectly cl2 mapset)
- say a large group of pain elementals that is clearly meant to be a threat to the player is pretty harmless if you use cl2 because of the lost soul limit; do you think it should still be used?
- say a secret exit in MAP31 is not accessible in cl2, meaning if you want to do a D2ALL/episode run you'll need to skip MAP32; do you think it should still be used?

There's a whole world of maps broken spectacularly in the compatibility stated by the authors. Ultimately, I think which compatibility should be used should be judged on a case by case basis and little heed should be paid to the authors' stated compatibility when it's pretty clear that's not the compatibility that was used for testing. I think getting 100% kills/secrets better captures the spirit of "playing as intended" than using the compatibility that authors provided because they are expected to provide one and get yelled at if they don't and not necessarily because they specifically aimed for said compatibility.
 

13 hours ago, Revved said:

The Journey MAP31: This one's interesting because neither exit will allow you to get 100% kills or secrets. The paths to get to either exit cut you off from the rest of the map.

5 hours ago, Andromeda said:

Similar case to sacrment MAP04. I'd say take the path which allows you to get more kills and secrets.

The Journey MAP31 happens to have more monsters on the way to the secret exit which seems to be the ideal case. The other way around you have a pretty bizarre scenario: you'd want to go for the normal exit in an IL but for the secret exit in D2ALL/episode run. There are a lot of possibilites for varied scenarios as well - for example one route can be a lot harder despite having fewer monsters or one route can be a lot slower than the other. I would go as far as to say that it might make sense to have separate runs for these scenarios in some cases.
 

17 hours ago, Grazza said:

Regarding lost souls, I don't think this exception should be generalized to "anything that doesn't count towards the kills percentage". If it looks, walks and takls like a monster, it should be considered one, unless there is a specific reason to exempt it (in the case of lost souls, this is the fact that PEs can easily spawn them into the void). For instance, Batman Maxes don't exist just because of the authors' odd decision to make the monsters not count.

Possibly an unpopular opinion but I fully agree with this duck test for the monsters, I don't think that monsters that clearly are monsters for all practical purposes should be ignored in a UV Max. Regarding Batman Doom, the authors say in the text file that the reason for the decision is that they wanted the wad to be more accessible on Nightmare. It prevents the monsters from respawning and they considered the 0% kills as an unfortunate side effect. A funny situation because it makes Respawn no different from Max and Nightmare barely different from UV with -fast. You would also presumably want to use a dehacked patch that restores the kill percentage for a UV Max but would need to avoid using it for Nightmare because respawning monsters is precisely what the authors intended to avoid (BTW, I did try to make such a patch but it turns out it's a bit problematic which monsters should be included in the kill percentage because of some tricks the authors used.)
 

16 hours ago, Maribo said:

I guess that's my argument for why instances like this should have an exception, the fact that the 2nd secret is completely inside of the first one makes it redundant.

 

I can't say I agree with the phrasing that if a secret could be considered "redundant" should come into consideration but I definitely agree that you shouldn't be required to go through a contrived set of steps and make the whole run about bypassing an obvious bug in pursuit of some misguided sense of completionism *cough cough* lv15 ev15 pl27 *cough cough*.

I don't think falling off into an unreachable area is a good excuse for being able to skip it, unless it was some very particular scenario where the monster fell off before you could do anything about it (Say you were forced to wake them up by shooting, they infought and one of them fell off before you reached the room), even then I think those cases would be debatable

1 hour ago, Keyboard_Doomer said:

The Journey MAP31 happens to have more monsters on the way to the secret exit which seems to be the ideal case. The other way around you have a pretty bizarre scenario: you'd want to go for the normal exit in an IL but for the secret exit in D2ALL/episode run. There are a lot of possibilites for varied scenarios as well - for example one route can be a lot harder despite having fewer monsters or one route can be a lot slower than the other. I would go as far as to say that it might make sense to have separate runs for these scenarios in some cases.

Dsda has separate entries for normal and secret exits, I think there isn't a more correct way to max the map since they won't be in the same table anyway.

Only "speed" categories have separate entries for normal/secret exit, categories that require 100% kills/secrets don't. But either way, you could have similar situations to what I mentioned with just one exit in the map or with multiple exits of the same type.

1 hour ago, Keyboard_Doomer said:

I think getting 100% kills/secrets better captures the spirit of "playing as intended" than using the compatibility that authors provided

I agree with this (and the rest of the complevel stuff in KD's post) and it's the principle I have been generally following myself when doing demos. Obviously the provided complevel is the preferred one but if it leads to issues which can be solved by using another compatibility setting, then I think switching to the other complevel should be the baseline way to go. More often than not that'll be perfectly reasonable and doesn't really change the maps in any meaningful way (and no, I don't consider slightly annoying boom monster behavior as meaningful). Ultimately it should be a case by case evaluation of course since there are probably cases where switching complevels does fundamentally change some other part of the map (e.g. trivializing some encounter due to boom ledges, insane lost soul flooding of some area) but I mean, reporting and discussion is encouraged if the runner has any doubts.

I can't say I agree with counting monsters that don't add up to 100% kills as monsters. IMO, I wouldn't host maxes for something like Batman Doom. If it were possible to create a Dehacked to tracked that, maybe I could reconsider, but without that, it would be horrible for archival purposes, and I don't necessarily think even that makes much sense. In any case, who's to judge what is or isn't a monster anyway? If Keens had the monster flag removed from them, would they count? They do no damage and do not move. And additionally, who's to judge if a monster spawn wouldn't cause problem. Take All Hell for example, which has helicopters that spawn non-counting plasma marines. If the plasma marine falls into an inescapable pit where you can't kill it, surely, it would be the same problem as PEs spawning lost souls out of bounds? So, why would this case be required and any other not?

WRT running on the wrong cl, it's certainly tricky, but I wouldn't necessarily give preference to using the wrong cl just to fix spawns vs now. I personally prefer running in the intended cl whenever possible; if the cl is broken, ideally the mapset should just be fixed through fixwads or letting the author know (preferably). If not, it is what it is IMO. Mostly I want to avoid scenarios where wads from 1995 are run in ZDoom because they happen to be maxable in ZDoom and not in cl3, which would be pretty annoying. 😅 I suppose case-by-case calls might be possible to make; for example PEs that are meant to be threatening might be a case for cl9 (although then why not run the original Doom 2 map 9 in cl9 as well?). Besides, in the case where one map of a megawad gets fixed with cl9, but otherwise the set works with limit-removing, should you use cl9 for all maps or just for that one map? It's confusing either way, and I think it would be best to stick with the original cl even if spawns are broken and whatnot in most scenarios.

WRT maps like Sunder 5 where monsters fall off or 13 where they don't spawn, I dunno, I am not keen on redefining max to add these exceptions. At that point, you start having tactical approaches to prevent monsters from spawning or pushing off cybs intentionally, and that would make no difference. Plus, if a map has very slow teleporters, but consistent ones, why not skip monsters then too? At the end of the day, max is going to be annoying whether a monster teleports in rarely or always but after 6 minutes of waiting. At the end of the day, there's also the fact that if a map has a bad max, it's kind of the player's choice to run it and deal with the consequences as well. :^)

7 hours ago, 4shockblast said:

In any case, who's to judge what is or isn't a monster anyway?

You should not have gone there. This is trivial - WE judge what is or isn't a monster. Obfuscating the definitions like "we may never know" is stupid. This is literally the thread where we decide what we do know. The very idea of this thread is that we can't ever get foolproof guidelines, and therefore we will need a table of case-by-case exceptions. I'm surprised you of all people would make such a trivial argument.

6 minutes ago, dew said:

You should not have gone there. This is trivial - WE judge what is or isn't a monster. Obfuscating the definitions like "we may never know" is stupid. This is literally the thread where we decide what we do know. The very idea of this thread is that we can't ever get foolproof guidelines, and therefore we will need a table of case-by-case exceptions. I'm surprised you of all people would make such a trivial argument.

I suppose, yeah, that's what this thread would be for, i.e., providing wads and asking which things count. That said, the tracking argument still stands; it's complicated unless there are official Dehacked patches to decide what is or isn't a monster (or cl0/Doom v1.2 for lost souls, if this were the agreed direction). Without those, it would be painful for DSDA maintainers to verify maxkills for any wads with questionable monster kills. Perhaps something like Batman Doom is a fair exception if such a Dehacked were made, but overall, I would probably prefer just having no maxes for maps like that, especially when it's not clear from the wad textfile why the monsters were exlcuded from the intermission counts.

Let's call a spade a spade. If the archive no longer wishes to be an all-encompassing archive, because inconvenient demos are hard to verify and the staff insists on having verified every one of them, then just drop the charade. I don't particularly give a shit about Batman Doom. I don't care how its case gets resolved, honestly. But making it a platform for some general decision making that goes beyond the max ruling AND doesn't follow the exception principle, BECAUSE it is so vewy hawd to vewify, that makes me fucking mad. State your rule change, or state your exception list, or don't bother.

To be clear, I am not saying that my reasoning for not wanting stuff like Batman Doom to not have maxes is purely effort of uploading, although it might have sounded like that in the post. I certainly would prefer for demos to be verifiable easily, it would make life easier, but if not, it is what it is I guess. As a whole, though, I generally prefer treating WADs as they are released, so if something doesn't count for the intermission screen, then it doesn't count to me. :P

Ya, if a thing isn't flagged as the kill counting, then the kill doesn't count. That seems clear cut to me. Such things aren't monsters. The only person who should own that choice is the wad author, and if it seems wrong by either choice or misadventure, it nevertheless is what it is. If you're going to start upgrading objects to monsters, why aren't you also opening unmarked secrets for instance? Those are as much of a glaring omission for the spirit of max, if you want to bring spirit into it instead of going based on what the engine counts.

One other thing that I would like to add to this is that lost souls arguably aren't the only monster that is excluded from intermission counts; the other monster is Romero heads. I suppose the original ruleset for Compet-N just never accounted for them, because in most of the cases in those WADs they are required to kill (outside of TNT AFAIK, where it is much faster to kill anyway). As far as I know, I'm not aware of any rule that requires players to use Romero head exits for max, despite them counting as a monster, e.g., in the pacifist ruleset. Either way, this is a place where we probably want to clarify the existing rules or, if we change them to extend the definition of monster to other cases, we should also consider whether Romero heads should be killed.

I think I would have to agree with kraflab, anything without the countkill flag shouldn't have to be killed. It's the most basic way to track what is or isn't necessary to kill for a max demo. I think in good faith, anyone doing dehacked stuff and messing around wouldn't remove a countkill flag from any particular monster for no reason, much like how the lost soul is essentially half projectile half monster because of pain elementals. Say there's a turret monster with 30000 hit points, but doesn't count as a kill; technically it could be killed, but is there any competitive or reasonable reason it should have to be, especially because it doesn't actually count on the kill counter?

Also I do believe that Romero heads don't have the countkill flag, so with the countkill flag declaration, they wouldn't have to be killed anyways.

When it comes to maps with a head and another exit, it seems reasonable and simple that the standard max rules should apply: Exit with all monsters dead (except lost souls) and all secrets collected on skill 4. Then it essentially it doesn't matter how you exit the map, as long as you kill all the monsters and get all the secrets.

Basically, I guess, declaring that things without the countkill flag don't have to be killed future proofs any dehacked modifications from generating more exceptions, solves the issue of Romero heads needing to be killed or not, and also maintains simplicity when verifying max demos via the counters.

The barons in magnolia seem like a better case to consider than batman doom. In batman doom nothing is tagged for kill%, whereas in mag the barons specifically are exempted, and I think it's pretty clear that for max you shouldn't have to kill them, because that's exactly why (correct me if i'm wrong ribbiks) they're not tagged as kills. If you're going to start killing the barons because they seem monstery to you, you might as well kill the evil eye turrets as well, I'm sure there's some extremely inconvenient way of killing those.

Edit: I think kraflab's point about unmarked secrets is very pertinent also. If you're going to start deciding for yourself which things count as kills and secrets, you're setting yourself up for a huge number of annoying edge cases and stupid decisions.

Why not just say that *in every case* whether or not something is tagged as kill% is whether or not it counts as a kill for max, and if people want to record maxdemos on batman doom (and other loopy cases) they should make an edit of the wad that restores kill tags to monsters.

If the sole criterion for deciding whether a monster is actually a monster is if it acts like a monster, then might as well ignore Commander Keens in levels that have them, despite them having the COUNTKILL flag. Would make levels such as pl2 MAP29 and sodfinal MAP31 better to max.

A lil late to the debate, but from what I've seen here so far my thoughts are that anything is flagged as a thing that counts towards the kill count needs to be killed, while anything that isn't flagged as such should remain optional for any runner. I feel like that is the only reasonable baseline from where to talk about when, where, and why exceptions should be made, if any...

If something is "hidden", but not tagged as a secret, then it's either a conscious decision made by the respective mapper, or an oversight, and in either case I'd argue that ignoring this hidden but untagged gizmo is fair game. After all, runners far and wide used and abused oversights of all sorts to their heart's content ever since speedrunning became a thing (sometimes even unbeknownst to themselves), and roping non-measurable "metrics" like "intent" into the equation is a recipe for disaster in the long run...

As far as exceptions are considered, I'm generally fine with exempting monsters that just won't leave their ramshackle teleport closets, provided it can be verified that the respective closet is indeed a problematic one, even if it doesn't "goof out" 100% of the time (a map from 50shades comes to mind where PEs sometimes just wouldn't leave their closets, not sure there was a fix for that, on that note). That said, if a closet malfunctions, then the remaining contents of the closet shouldn't matter, because who cares if 90%, 50%, or 10% of the monsters actually leave the closet when the malfunction is always the same, and occurs for the same reason in principle..? I don't expect any maxer to grind a map with a defective closet to the point where they get the least possible amount of teleporting monsters, it just seems so unreasonable to me that I think "a malfunction is a malfunction, move on and kill what teleports into the map proper"

Likewise, I'm A-okay with exempting secrets that could only be reached via some arcane manipulations like D2 map 15, as well as secrets that cannot be reached due to verifiable mapping errors (I think here's where magnolia gets another shout out, unless I happen to be out of the loop and a fix has been made available)...

Speaking of verifiable mapping errors, when it comes to things that work in complevels other than what the mapper stated, those should be exempt also instead of allowing the runner to use a complevel that somehow makes things work for ILs, because if that's the only way to go about "fixing a problem", then what's gonna happen when the same WAD is to receive a D2ALL played on a complevel that fixes one issue but changes the behaviour of things in every other map respectively..?

11 hours ago, 4shockblast said:

I certainly would prefer for demos to be verifiable easily

Personally I weight this the most in the discussion for sure, despite some decisions being controversial or illogical, but easily verifiable no matter through manual work or automation. On the other hand, I'm not really a fan of a list to cover exceptions because it's bound to have more in the future... Though unlikely to happen, I still really want to have a ruleset cover most of the edge cases. I also support the idea using the author labelled complevel for runs.

I guess times to make a mapset that's cl2, but in one map, 90% of the monsters won't teleport out on cl2 but everything will teleport out on cl9 :P

5 hours ago, Andromeda said:

If the sole criterion for deciding whether a monster is actually a monster is if it acts like a monster, then might as well ignore Commander Keens in levels that have them, despite them having the COUNTKILL flag. Would make levels such as pl2 MAP29 and sodfinal MAP31 better to max.

I think it makes sense to have a consistent definition of "monster" that applies for both Pacifist and Max. Note that the traditional Max definition clearly regards lost souls as monsters, while specifying that they don't have to be killed ("every monster ... except lost souls").

As an example of a wad where the kill counter flag was totally haphazard, there is aotw.wad. The author just made the monsters whatever way he felt like, with no real attention given to the countable flag.

BTW, I don't see that Max should be a sacred category where the rules have to be tweaked to make it neat, tidy and clear-cut. Plenty of other categories are either infeasible or messy on certain maps.

The thing with closets is that most of the time it doesn't make a difference what teleport, in the mayhem map the closet has hundreds of enemies and it's usually 1 imp that doesn't teleport in, it doesn't make any difference in the fight but invalidates your run, saying a run in this map would now aim to miss monsters to save time is false, and so is the case of most broken closets. Pushing cybers off ledges would be the case that would give advantage to the player if it was decided they can be skipped. It could be decided if a monster not spawning makes a difference into the run, I can't think of maps where this is the case, but it would account for possible future noye2 maps that have broken cyber closets just to be the exception.

There somewhat of a difference between slow teleporters and closets that are broken and the monster will never teleport. In sc26 max your time is almost entirely dependant on how fast the monsters decide to teleport, but since the teleporter actually works I think you should kill everything in this map.

11 minutes ago, Ancalagon said:

saying a run in this map would now aim to miss monsters to save time is false

well, I was thinking more of seriously borked closets (and I've seen some that would make your head spin), but what you said there otherwise is the same as with the 50shades map I was thinking of, where the closet would empty itself almost entirely, but sometimes you'd have one or two PEs stuck in there... Sure enough, those won't really ever affect the time of the run in a meaningful way, because it's basically a pure BFG spamfest... The point was that borked closets can sort of reach a state in which they "loop" without producing results regardless of how many monsters are left in them, it's just less likely to happen in a densely packed and neatly arranged - but still somehow messed up - closet (which is also why I said that I can't imagine runners would consider grinding maps for optimal "closet breakage")...

I think ultimately if a closet breaks it shouldn't invalidate the run, regardless of what's left inside, even if it only breaks occasionally for some arcane reason. If you get to take home a good run because of a lucky RNG roll, or extremely productive infighting, why shouldn't you take home a run due to lucky "closet breakage" as well..?

Hello, I've been told to post this demo here of Rush which is a D2All UV-Max up until Map12 since Map13 is a credits type of map, it does have an exit but you can't kill all of the archviles. The UV-Speed does not exit Map13 and I did watch it just to make sure I wasn't remembering something incorrectly. I guess the topic of discussion is, does this count as D2All UV-Max or should I have exited Map13 without even trying to kill stuff?

rud2allm10402.zip

On 12/13/2021 at 10:34 PM, Red-XIII said:

Same thing applies for pl2 map 32, which replicates rq23. Go4it has the arc trapped that can be killed by crushers, but it is slow and pointless. Should it be killed?

Ghost monsters yes, of course.

no way.. this would be longer than the run itself.

ofc the ghost -intended to be killed that way- should be taken care of in my humble opinion

On 2/17/2022 at 1:16 PM, blob1024 said:

no way.. this would be longer than the run itself.

ofc the ghost -intended to be killed that way- should be taken care of in my humble opinion

Yes that's what I meant, I see no point at all in killing these type of arcviles.

I think arch-viles that are intended to revive monsters, but can't easily be killed, such as pl27 and rq23 should not be required for a max run.

I wonder if it would make sense to have some kind of exception for slayer map02, since it's pure luck whether or not 100% is possible.

The Lost Episodes of Doom (JPTR_V40.wad) - E1M4:  There is a room with 6 Cyberdemons in the large pit with long and 64-unit-wide bridge going through it. The pit is 256 units deep and covered with 20% dmg floors. Additionally, there is no lift around, so once you fall off the bridge, you can kiss your run goodbye. All you can get is a Plasma gun with 600 cell units and a rocket launcher with +/- 16 rockets. There are also 2 BFGs and 28 cell charges, but they all are in exit pits with no way back up. 

51 minutes ago, Kam Tovalski said:

The Lost Episodes of Doom (JPTR_V40.wad) - E1M4:  

well i was curious enough and played the map and there is enough ammo to kill them but probably would take hours to play it safely and wait for the cybers to move in a position that you are guaranteed to hit...

but for any1s sanity skipping them should be fine IMO :^)

rough estimate on ammo: 600 cells, 150ish shotgun shells, 200ish chaingun ammo 20 rockets.. rest of the map you would have to chainsaw everything to get this ammo, probably more shells/bullets as I used some while playing the map.

Actually I thought of another way to deal with the cybers, running back and forth from 1 side to the other on the bridge ( which is 100% doable after trying) so now I' m not so sure if I'm in favor of being able to skip them :p 

I have a question related to this statement:

On 12/13/2021 at 12:16 AM, dew said:

Unkillable monsters: Monsters that can't be killed due to a mapping gimmick. Most notable scenario is turret snipers revived by hidden, unreachable archviles. Yet there might be extremely contrived ways to kill the archviles, and then the snipers. Another example might be out-of-reach sniper cubbies. Status quo says they're to be ignored, but it might be healthy to discuss that case-by-case.

Specifically, about Another example might be out-of-reach sniper cubbies.

Let's take a look at Congestion1024 Map30 layout:

There are 6 mancubi marked with red rectangles and they can not be killed as they are out of reach from the player's auto aim. However, this is map30, so it's an IOS map. Theoretically, these mancubi can be killed through monster infighting (for example, cacos spawned by IOS) but it is rather impossible.

What should we do in that case? Can we ignore them as written in that statement? Or will that map remain (most likely) non-maxable?

Status quo says they're to be ignored, but it might be healthy to discuss that case-by-case so i'm asking for some help.

I had played this map with IDDQD and the infighting happens, it doesn't have to be cacos. I agree they are (or seems to be) unkillable by bullets/rockets/plasma due to being too far away. Sorry but I have no strong opinion whether we should allow to skip them.