How many of {something} does it take to kill {something}....?

I was watching a playthrough of Doom this morning and it got me thinking: How many Zombiemen would it take to kill a Cyberdemon? How many Zombiemen would it take to kill an Archvile? How many Revenants would it take to kill a Spider Mastermind? Etc. and etc. ad nauseam. I've seen people post videos on Youtube where they have like, basically have a Coliseum-type setup where they have monsters battle each other until there's a winner. As I've never really seen people talk about this on here, I was wondering if anyone has anything close to definitive about this?

Now I know Doom is very RNGeezus based so it'd probably wouldn't be the same number every time, but if done enough times a good solid average should come up, right? I dunno.

::shrugs:: I'm very curious about this and wonder if anyone's experimented with this before.

There was a demo somewhere where a Mastermind got stuck and killed by 3 Imps. It's quite relative.

18 minutes ago, bzzrak said:

There was a demo somewhere where a Mastermind got stuck and killed by 3 Imps. It's quite relative.

Yeah, I suppose any creature can get stuck and get killed by "Creature X, Y or Z". I was playing yesterday and a Cyberdemon was stuck between a doorway and a Pinky was biting on him and the Cybie just sat there and took it for a good minute, minute and a half and then became un-stuck and launched a rocket up the Pinky's ass. In Tatsardcacocaco's run of Alien Vendetta, Map 18 (?), a Mastermind got stuck on something and was munched to death by a Pinky (I think?). So it does happen, if they're stuck on something and not fighting back.

I guess I was meaning in a combat setting where Creature(s) A were fighting Creature(s) B. I probably should've been more specific.

6 hours ago, bzzrak said:

There was a demo somewhere where a Mastermind got stuck and killed by 3 Imps. It's quite relative.

A few years back I was playing PSX Doom, on the Doom II map Tricks and Traps the Cyberdemon got stuck in a doorway and a very angry Imp clawed him to death. 

It only takes one Lost Soul to kill a Cyberdemon, as proven by the pacifist run of E2M8.

7 hours ago, Chewyninja69 said:

Now I know Doom is very RNGeezus based so it'd probably wouldn't be the same number every time, but if done enough times a good solid average should come up, right?

The results would depend not only on RNG, but also on placement of the monsters, the surrounding architecture, distances, whatever. While it is possible to average the results of a fight with a specific setup repeated a million times, it is not possible to somehow average all possible monster placements and map architectures.

Thought some list might help here - surely someone gotta link to some stats showing like:

Weapon/Pistol shots/ Shotgun Fullhit/ Rockets/ImpFireball/whateverhurts

Cyberdemon 500/ 100/ 25 /100 /

SpiderMastermind 400/..

anyMonster x/y/z/...

wouldnt that do the job?

A "Combat setting" might be difficult to get and its outcome largely depend on surroundings yes, especially Cyberdemon vs Zombieman..^

And imho in comparison with other games I think RNG here got a rather small role, if any at all - if you dont want to let your local archviles open doors for you, ok.  But the small differences in between aiming and timing of monstershots, ok small - but you just avoid getting hit in general, then RNG then doesnt matter anymore.  And monster placement actually is fix - no RNG, and no reason to wait and see where they walk usually ;) 

These types of numbers have been on the wiki for years. Of course, there is no longer a link to the wiki on Doomworld... https://doomwiki.org

When threads of this sort pop up, anywhere and about anything, people are quick to point out that 'it depends', which is true and 'the correct answer' but is also not very interesting.

It also should be possible to set up factions of monsters that immediately infight in ZDoom, and post images of starting configurations in this thread, which, sure, would only answer the question 'X of {something} seems to be evenly matched against {something} in this particular arrangement' but would still be cool to look at. I'll try to look into that later, if I can get it working. 

For now, I remember this thread, which looks at evenly balanced one-on-one match-ups. 

Yes it does depend, and I agree its not an interesting answer. There are certainly lots of conditions that affect this. We would need to agree on a specific base line in which these monsters are fighting each other. What is a fair distance these monsters can be from one another? If its a battle of one monster and many others, are the others grouped together or spaced out? Are they coming from opposite directions? This type of stuff needs to be solidified before we can have a definitive answer, and since the OP is asking, he/she should be the one clarify.

3 minutes ago, 40oz said:

We would need to agree on a specific base line in which these monsters are fighting each other.

I think it's fine to set up certain parameters. Like sending a hallway of "X" against a cybie, or having X surround the cybie so all things can shoot at it at once. Pretty sure there will be vastly different results. Problem is hitscans dying to infighting though, so in that case the hallway method may need some tweaking. Assuming someone sets up a ZDoom script that does what rdwpa suggests, it should be quite interesting to see the numbers.

Well how about This question, Can a Cyber agro a zombieman without killing it, Then if it where to precede to get stuck, how long would it take our singular heroic zombieman to Kill the Cyber.

43 minutes ago, MrGlide said:

Well how about This question, Can a Cyber agro a zombieman without killing it,

In theory, a Zombieman doesn't need to get aggroed by a cybie to hit it, so if we remove this idea from the equation, we can move on to:

44 minutes ago, MrGlide said:

Then if it where to precede to get stuck, how long would it take our singular heroic zombieman to Kill the Cyber.

I'm pretty sure somebody with better knowledge of the details in regards to zombieman behaviour than me could calculate this down to a +/-10sec accuracy, by figuring out how many times a Zombieman shoots per second/minute/hour, and using the average damage dealt per hit.

9 hours ago, Chewyninja69 said:

How many Zombiemen would it take to kill a Cyberdemon? How many Zombiemen would it take to kill an Archvile? How many Revenants would it take to kill a Spider Mastermind?

I placed a single cyberdemon randomly in a crowd of several thousand former humans with the MBF friendly monster flag*. It seemed fairly consistent that the cyberdemon kills 1200-1300 of the friendlies before it dies, although a couple of times the death count went up to over 1500.

A similar thing with one enemy archvile against a thousand MBF humans gave about 50-60 humans dead before the archvile died.

I couldn't get any reasonable result for spiders and revenants because the revenants surround the spider, which immobilises it, and punch it to death. It might down one or two before they reach it if you start them far enough away, but then it can't hit them very hard either.

________

* MBF friendlies don't shoot through each other. If you switch the alignments so the cyberdemon is friendly and the zombies normal, the zombies almost immediately infight each other to death and the cyberdemon mops up the stragglers.

Back to the drawing board! 

ya, but naturally it would have no reason to hit the cyber unless the player was in front of it, and the trooper would only be hitting the cyber when missing the play, which would increase the randomness of Time to Kill. So I figured the first logical question was, if a Trooper can even live from anything the cyber could do to aggro it.

Yeah, it's possible to do less than 20 splash damage. 

In my experience I've seen a single SMM being punched to death by just two revenants many times, actually by just two (or three) of whatever kind of non-boss monster. I think it's because her sprite is so big that she can barely move in a closed area, or when she gets corralled. Still I've also seen her surviving in a massive horde of lots of different enemies against her, so yeah, it all depends (yes I know it's a vague answer). 

It also depends on the type of attack of each monster and where are they positioned. For example in one occasion there were two SMMs from very distanced positions infighting each other. I don't remember the exact time when that was happening but let's invent, it was 30 minutes of playing the map. By the time I reached an hour they were still there fighting. In fact, I had to off them by myself because they could spend hours and hours and still no one would fall.

Cyberdemons are another story, there's this map where two cybers can infight a massive horde of (a hundred more or less) revenants and it takes some special movements and moments to make the revenants off them quickly. I've always wondered how an infighting is going when I'm not seeing it. In this occasion with the cybers and revenants, if I run away and take a teleporter to another side of the map, I would listen the cybers dying in a matter of seconds. But if I stay with the monsters, the cybers would win. I believe it's because of one's position in the map, the monsters will always follow you even if they can't see you, unless they are busy infighting. 

For an archvile to successfully kill a boss monster, there has to be a minimal distance between each other, so the splash damage doesn't affect the archvile himself. It will still take him a long time. I've seen two archies trying to kill a very weakened cyberdemon for like 10 minutes. I can't imagine how long would take a solely zombieman to kill a cyber, I would fall asleep on my keyboard...  

Quote

which is true and 'the correct answer' but is also not very interesting.

:D

Quote

Can a Cyber agro a zombieman without killing it

That should work, but the systematics/mechanics often get A LOT deeper/different than one might think at first when innocently asking those questions. Think about this 2x damage / line of sight / complex with BFG dmg calculation, guess theres quite some highermathematics stuff  going on with rocketblast calculations too...links are yet missing..but actually If youre interested in something (or why ask then?) than "depends" is a great (if not best) answer for both sides chance going in deeper..

Quote

specific base line in which these monsters are fighting each other. What is a fair distance these monsters can be from one another?

As "benchmark" maybe a plain room Matrix Style - without any walls and set all oppononents around according to some special distance, maybe minimised so all can just walknot stucking withhin each other - and a straight line teleporting in to simulate 1 on 1 StandOff and find average times till shots....I dont know but Im quite sure theres some MonsterArena-Idea already put together in one way or another. ;)

Quote

Back to the drawing board! 

wow that one came incredibly quick

Quote

By the time I reached an hour they were still there fighting

yes, they have a huge spread on distance

46 minutes ago, RjY said:

I placed a single cyberdemon randomly in a crowd of several thousand former humans with the MBF friendly monster flag*. It seemed fairly consistent that the cyberdemon kills 1200-1300 of the friendlies before it dies, although a couple of times the death count went up to over 1500.

 

video please :)

That's really cool RjY, what is that and how did you do that?

I'm inclined to think that it's a big enemy against little former humans, and the big guy still wins. That wouldn't happen in the former humans couldn't infight each other.

8 hours ago, MrGlide said:

That's really cool RjY, what is that and how did you do that?

Thank you for your interest.

What: see my previous post in particular the part 40oz quoted. 15000+ MBF friendly troopers, versus a single cyberdemon. In this particular case to facilitate recording I started the cyberdemon right next to me, so I die almost immediately -- thus the recording doesn't get interrupted by a switch back to player view halfway through (I cut off the very beginning), and so that when the cyberdemon dies the troopers stop moving, instead of continuing to mill around, which makes the point of its death more obvious.

How: errr. In late 2001 I got super interested in the behaviour and outcome of large monster fights (guess why, it should be easy if you remember what came out at the time...) So I made this program that generated a map and filled it with random monsters in specified ratios, and over the next few years wasted dozens of hours running it with different parameters and observing the results. So this thread is a nostalgia trip for me. :) The rest is just having my source port draw the automap with things visible and write the screen to a file with the game clock running at 10x usual rate, then turned into a gif with imagemagick.

8 hours ago, galileo31dos01 said:

I'm inclined to think that it's a big enemy against little former humans, and the big guy still wins. That wouldn't happen in the former humans couldn't infight each other.

No, the big guy loses. But he took me, and nearly 1500 of my allies, with him. My allies don't infight (much) because MBF friendly monsters are careful not to try to fire through each other unless specifically angered by a stray shot another, in which case they retaliate exactly once, then go back to what they were intent upon before.

@dew (below) Thanks :) I've found much enjoyment in watching the patterns develop.

Holy shit, RjY, that's soooo pretty!

10 hours ago, galileo31dos01 said:

For an archvile to successfully kill a boss monster, there has to be a minimal distance between each other, so the splash damage doesn't affect the archvile himself. It will still take him a long time. I've seen two archies trying to kill a very weakened cyberdemon for like 10 minutes. I can't imagine how long would take a solely zombieman to kill a cyber, I would fall asleep on my keyboard... 

Well, in a plain field battle with no architectural gotchas a vile will kill itself long before putting even a dent into a cyb's health. But theoretically that same vile might be able to take down up to 77 masterminds...

Quote

MBF friendly monsters are careful not to try to fire through each other

never heard of that, thought they shoot each other like hell all time.

- still wondering what your truly great vid looks like when switching just one single paramater (this MBF thingy maybe, or just difficulty settings), guess its a totally different electric sheep then?^

2 hours ago, PanterD2S said:

never heard of that, thought they shoot each other like hell all time

No that's the enemies:) These chaotic beings don't care who they hit when trying to attack you. MBF friendlies are much more careful not to hurt you or each other. This of course places more workload on your CPU - I expect Killough wrote them not expecting you to have more than a few allies / helper dogs against a hundred+ horde.

2 hours ago, PanterD2S said:

[...] switching just one single paramater (this MBF thingy maybe, [...] ) guess its a totally different electric sheep then?^

Right. For example this is what happens if you invert the alignments:

As expected the former human enemies immediately attack, regardless of each other, and ~80% of their initial 15000 are wiped out by less-than-friendly fire in around 15-16 seconds' gametime. The single cyberdemon, now allied with the player, cleans up the remainder in a couple of minutes.

^It's so pretty, can't stop looking at it. Also nice way to show thing-movement behaviour.

..and in a blimp they nearly all kill each other, strategically lose that battle.

Im totally stunned watching the dots -  you should really make a youtube serie about that, or some customizable utility so everyone can have his own unique sheep. And this is all precalculated here I guess ?- Cybie never gonna lose this by having real bad luck?, but just imagine a close fight set up and realtime calculation with different outcome every run...! 

Tons of ideas.., but what exactly are YOU doing with that incredible...Tool(s)?

 I mean, beside amaze people in Forums...^

ahh

Quote

over the next few years wasted dozens of hours running it with different parameters and observing the results.

obviously others want to do exactly the same ^^

- by any chance, is there a downloadable version anywhere maybe? And a collection of your favourite Battles ;)

9 hours ago, RjY said:

How: errr. In late 2001 I got super interested in the behaviour and outcome of large monster fights (guess why, it should be easy if you remember what came out at the time...) So I made this program that generated a map and filled it with random monsters in specified ratios, and over the next few years wasted dozens of hours running it with different parameters and observing the results. So this thread is a nostalgia trip for me. :) The rest is just having my source port draw the automap with things visible and write the screen to a file with the game clock running at 10x usual rate, then turned into a gif with imagemagick.

Nice! I was going to suggest something exactly like this but it appears you've already done it.

Ideally it would be nice to see this expanded and automated to test every possible match-up of monster types. Ideally this would take the form of something like:

1. N of monster type 1 vs. M of monster type 2. All of monster type 1 have the friendly flag set.
2. The test runs some number of times - headless, at maximum speed (like -timedemo, but not bothering to render the screen)
3. The test ends when every one of a particular monster is killed. (eg. in Sergeants vs. Lost Souls, if all Lost Souls are killed then the Sergeants win). Probably this needs a custom/modified source port to end the test and report the result.
4. Some kind of random element is introduced so that the test doesn't produce the same result every time. Maybe this is achieved by moving monsters around to different starting positions for example.
5. The results are examined in aggregate, and we use the results to adjust N and M (from step 1). Then we re-run the whole process again. The goal is to adjust the inputs until we end up in a stable situation where each monster type wins ~50% of the time. At this point we can say that they are evenly matched.
6. Repeat all of steps 1-5 for all combinations of Doom monster types and build up a complete table of how the monsters match up.