Random Damage and Competitive play: Why they don't mix

[Blasphyx]

I can see that people put up with it, but my view is that "luck" should not be a factor in determining who is better.

Everyone is subject to the same level of randomness. If you compare the usual priv players to one another, you can accurately claim who is better. The universe doesn't just tilt odds for or against specific people's favor.

EDIT: at least we unrealistically manage to reload the SSG in the same amount of time every time.

[Spottswoode]

Randomization is (in the RPG statistics sense) an abstract representation of where the bullet hits and how much damage it does. This method allows for smoother combat (rather than calculating what body part is hit every time an attack is made) than attempting to do combat with fixed damage parameters. Plus, especially with weapons like the chaingun and ssg, the amount of damage that each pellet would do would vary in reality with the amount of energy each projectile carries. (Plus there are some people who get shot in the head and live, pass out from losing a finger and bleed to death, etc.) Fixed damage isn't really a better indicator of skill because in real combat relative skill may or may not matter depending on the circumstances. (Plus in real life people only die once.) All of that being said, even modern damage calculators suffer from the same basic problem. (Meaning hitboxes don't translate significantly better to real life.) So your actual skill in the game comes from your tactical choices based on the circumstance. ( I could go further into combat theory, but the point is made.)

Latency and mechanics are an issue for balance reasons. However, narrowing the damage spread may make more sense, in terms of realism, than making damage constant. Constant damage can get you killed from a player who is lagging and not even in the same room just as easily as variable damage.

[Theshooter7]

Your premise is flat out incorrect. While I think that a mod that 'fixes' random damage ranges and shot patterns would be an interesting experiment, making blanket statements like that is disingenuous, and it is my belief that random damage in and of itself is not "anti-competitive".

You're right, my statement was a bit broad and I should perhaps apologize on that, since clearly, despite the random damage, the competitive scene goes on, one way or another.

Competitive play ultimately boils down to "good" players being able to consistently get better results than "bad" players.

Yes, however what about comparing the best of the best? Does it mean that if one pro player beats another pro player, the loser pro player is, in fact, a bad player? It shouldn't, it should just mean the other is better. But when it comes to the best vs. the best, I dislike the idea that "luck" can factor into deciding who is actually better than the other. Though to be fair, Qent had pointed out that being able to make the call between a sure victory and when you should back off and play it cool does weigh into deciding who is a better player. Conversely, when to be aggressive and put your opponent on the run, or when to lay low for any reason is the same deal.

The problem is if you want to introduce a new competitive standard for something with an already existing standard, such as Doom or IDL-style CTF, you need to take into account the very best players who have played by the old standard for so long. You need a not-trivial number of them on board, and even take their input into account, otherwise your competitive standard will at best be regarded as the "kiddie pool", with the "real competition" elsewhere (which is something Zandronum already has issues with). Or simply not played at all (look at what happened to ChainDM, and Chaindude is a very skilled and well regarded player).

First off, I am not trying to introduce a new standard to competitive Doom by any means (especially since Zandronum lacks any method of even implementing said ideas; in particular a missing DECORATE flag and some such). I don't mean any offense in this, but it is simply not my aim at current. These are just my views on the subject.

Secondly, yes, I was already quite certain there would be some spark or conflict with this idea amongst veterans, simply because this is how it has been since the beginning, changing little if at all. It's similar to the CS-1.6 vs CS:S debate; many players of the classic 1.6 refuse to touch Source because it feels entirely different and plays different. And I'm no pro CS player, hell I'm terrible at the game overall, but even I can tell the differences between the two and understand how it would affect players from one game to the other.

Also, Mobius agreeing with me? Never thought I'd see the day, heh. :V
But to state something he mentioned, this discussion is focused almost SOLELY on the random damage factor; ping/latency is a separate issue altogether apart from this. But in all, Mobius p much summed up my stance on the issue in a nutshell, for other game genres at the least.

[X-Ray]

I think it's fine as is why change it? Sometimes you get lucky i have got some really lucky moments when i spawn killed people with the shotgun 3 times in a row which was funny but that's how the game is. Also the more you play the more you get used to the weapons just saying..

[Dark-Assassin]

I hate getting insta killed by a SSG yet it takes 2-4 shots to kill someone else.
I always seem to be on the bad end of this broken RNG system. Broken meaning it's not actually completely random afaik.

[Ijon Tichy]

With the SSG at least, random damage isn't as much as a factor as you make it out to be.

Code: Select all

>>> shots = std.seqMany(std.shoot, 1000000, 5, 20) 
>>> std.mean(shots)
200.004425
>>> std.std(shots)
18.25494331460996
>>> std.deviations(shots)[1]
{1: 0.660833, 2: 0.961353, 3: 0.996687, 4: 0.999959, 5: 1.0}

As you can see from this, your average SSG blast will only go 18 damage off from the average. 96% of SSG blasts will do 163 and 237 damage (which with a freshly spawned dude is ALWAYS enough), and only 0.002% SSG blasts will do 109-127 damage. That is still enough to kill a fresh marine. If all pellets connect, a freshly-spawned player will ALWAYS die.

Code: Select all

>>> pellets10 = std.seqMany(std.shoot, 1000000, 5, 10)
>>> p10mean = std.mean(pellets10); print(p10mean)
100.015675
>>> p10std = std.std(pellets10); print(p10std)
12.913234656529246
>>> p10dev = std.deviations(pellets10)[1]; print(p10dev)
{1: 0.663214, 2: 0.969866, 3: 0.997725, 4: 1.0}

The standard deviation drops, as is expected - not by 50%, though, which shows that the spread is definitely widening. Not very much, though - 66% of your shots will do 87-113 damage. 97% will do 74-126 damage, and all of 0.23% of your shots will do 48-61 (actually 50-60) damage. Still, about half the time, 10 pellets won't do in a freshly-spawned marine - 14 pellets seems to be reliable enough to be legitimately surprised when it doesn't (99.7% chance of killing a fresh marine).

Basically, it's a very reliable and very potent killer at close range, and an unreliable piece of garbage at long range. We all knew that, though.
Spread is the main problem, not damage rolls. Whether a pellet hits or not is much more important than whether it rolls 5 or 15 damage. If anything should be made certain, it's how the pellets spread.

Edit: oh yeah I'll roll up some numbers using the Doom 'random' number generator in a bit

Edit2: Well, I rolled up some vanilla-style SSG blasts, and here's my results:

Code: Select all

>>> shots = std.seqMany(std.oldShoot, 1000000, 5, 20)
>>> sMean, sStd, sDev = std.mean(shots), std.std(shots), std.deviations(shots)[1]
>>> sMean, sStd, sDev
(207.03125, 18.889455350472655, {1: 0.59375, 2: 0.984375, 3: 1.0})

>>> for i, key in enumerate(sorted(counts[0]) ):
...     print("{}: {: <6} {: >5}%".format(key, counts[0][key],
...                                       round(counts[1][key]*100, 2)), end="  ")
...     if (i+1)%4 == 0:
...         print()
... else:
...     if (i+1)%4:
...         print()
...
170: 15625   1.56%  180: 62500   6.25%  185: 109375 10.94%  190: 109375 10.94%
195: 62500   6.25%  200: 78125   7.81%  205: 125000  12.5%  210: 46875   4.69%
215: 78125   7.81%  220: 46875   4.69%  225: 46875   4.69%  230: 109375 10.94%
235: 62500   6.25%  240: 31250   3.12%  245: 15625   1.56%

>>> std.percWithin(shots, hi=199)
0.359375
>>> std.percWithin(shots, lo=200)
0.640625

The standard deviation isn't too terribly different, and the mean went up a bit... but notice how the range got tightened so that it all fits within a 75 damage range (and skewed towards the high end)? Switching to the old RNG actually makes the randomness a less important factor for hitscans!

Okay final edit I promise: Then again, the Skulltag SSG does feel like a frickin' pop gun, so who knows.

[AlexMax]

Competitive play ultimately boils down to "good" players being able to consistently get better results than "bad" players.

Yes, however what about comparing the best of the best? Does it mean that if one pro player beats another pro player, the loser pro player is, in fact, a bad player? It shouldn't, it should just mean the other is better. But when it comes to the best vs. the best, I dislike the idea that "luck" can factor into deciding who is actually better than the other.

Except "Who is actually better than the other" is not something that is determined by the victor of one match in a vacuum.

[Catastrophe]

.

[StrikerMan780]

The random bullet spread is only logical to me, most FPSes, competitive or not, have random spread. Quake 1/2/3, UT, etc. IRL, a shotgun doesn't have a perfectly, 100% uniform spread pattern either.

Random Damage, is much better when compat_oldrandom is on (Like I said earlier in this thread, sorry to repeat... :P), and while it may add a small factor of luck, everyone has the same chances when it comes to what the RNG chooses, so, I personally view the competitive playing ground as being even in this regard. Observe your situation and adapt, as I always say. If your shot didn't do enough damage, change your strategy.

Now, I suppose it would be interesting to have a compat flag for damage falloff based on target range, similar to what TF2 does when it's random damage is off... Basically, your pellets at long range would do 5, 10 at mid-range, and 15 at close range. (Following the 1d3 concept.)

[Tosen]

Yes, but what about another example: The Chaingun(!!!). Say two opponents, 100 health, no armor, start sniping at each other at roughly the same time, timing their tapped shots carefully to keep it accurate yet shooting as fast as possible. Do you think the guy who shot first would win? That may not be the case with random damage, as the other guy could get the full 3x or 4x multiplier each shot and this guy could be getting 1x or 2x. If you ask me, that's something that probably should not happen.

Its called moving so they cant hit you

[Xenaero]

Your premise is flat out incorrect. While I think that a mod that 'fixes' random damage ranges and shot patterns would be an interesting experiment, making blanket statements like that is disingenuous, and it is my belief that random damage in and of itself is not "anti-competitive".

I'm going to start a little bit of an argument with this, which is ironic since I think you are arguing just for the sake of it. Plenty of games have evolved (Quake) or have the capability of adjusting the gameplay (From CS all the way to TF2) in a competitive setting to reduce all the random variables possible. This creates a solid base for competitive play between players and teams.

Competitive play ultimately boils down to "good" players being able to consistently get better results than "bad" players. Something as random as Rock Paper Scissors can be played competitively because of the amount of human psychology involved...a good RPS player can always outguess a bad one over the course of a series of games. To take video game examples, Super Street Fighter 2 Turbo and Defense of the Ancients both have random elements to them, and yet are played competitively in spite of this.

While correct, random elements introduce a byproduct of psychology in the community that resembles dismissive-ness and lays the groundwork for bad attitudes. Luck, for example, is the most common excuse made in games and it can most certainly be made in various source ports with varying gameplay 'styles' all introduced at the core level, be it with the ZDoom sound curve, blockmap hit bug, and so on. Competitive play at its basic premise is about skill, not random variables. It would be wiser to use FPS examples compared to others outside the genre.

That being said, I've had a similar idea to fix spread patterns to a static value, along with damage. It would be an interesting experiment to see in action at the higher leagues of Doom competition.

[werewindle]

That being said, I've had a similar idea to fix spread patterns to a static value, along with damage. It would be an interesting experiment to see in action at the higher leagues of Doom competition.

have fun getting anyone from the higher leagues playing it. i'm not kidding, you won't get anyone. do you even know what a bag of rabid flying acidsnakes you are opening?

ssg is a mess indeed. look at this, it's very hard to project a gauss curve on that. what are you going to use as the fixed damage value? the average? the median? will you tie the pellet spread to it or randomize it inside a cone of certain radius and spread? will you fuck with it till the end of times, because none of the values feel right? the final value will have tremendous impact on every single shot in the game and it will change how players move in every situation!

how about berserk? look at that CRAZY ASS damage distribution. you need to decide if your fixed value kills a healthy guy or not. or a healthy guy with a green armor. or, or, or, or... and yes, even that DOES matter, despite the limited usefulness of zerks in duels.

even the sg/cg damage spread matters. fuck, even pistol spread matters in map01! i'm talking about having 15 health and deciding if it's worth it storming a guy gambling on a lucky full damage shot. this mental border would get lowered for ALL WEAPONS universally and as a result you'd only boost quake-style map dominance, because stronger players wouldn't have to deal with lucky full damage shots. that's a resolute no from me.

[Powerman]

Granted the first few examples aren't a game but real life, but when people in the old west did stand offs you mean to tell me there was no luck? That no one barely outdrew there opponent? That they barely Got the shot off in time to save there own skins? Now let's look at Doom, do you want to strip away all the DoomGods tittles and remake them get them? Doom is just a game, and being really awesome at it is cool and all, but no one will really care. I mean if you were really good or a God at Halo, or Call of Duty, or Team Fortress 2, and you lost to some Newb, because of Random Damage, then sure Take out the random and see how well those two really are. But a game partially based on luck and Random Damage should be just that a game based on luck and random damage, and if you want to change that, make a mod for it and run a server with it. but if all competitive game modes become that way a lot of people will get there panties in a bunch over it not being vanilla. don't fix what isn't broke or even wrong

TL;DR It's the way it is because in reality it is fair

[Watermelon]

The random engine has cost me games. Point blank with skulltag turning all your pellets into 5 damage duds instead of 10 or 15 means a well placed shot leaves them with enough health to capture when they should have died.
Plus the skulltag's whiffing (whifftag) makes life even worse.
Sometimes I feel like playing CTF is tossing a coin. You either win, or you die [/GameOfThrones]



My favorite was playing with Tau-cannon beasts on Half Life. It was either you hit them and one shot them, or die.
And I'd usually die

[Eonfge]

No, thanks. Luck is also part of competition!

I second this.

I have to agree that luck and unbiased* variety improve the game from a casual and spectator side. I will always prefer to play a competitive game that has those two elements in them. Numerical random, where there is a variety in outcomes in the same situation, is key to a great spectator and enjoyment sport.

I understand that Quake Live and Counter Strike can't have such a system, where random can help you, or fuck you over. That is the appeal of those games: In Quake, the Railgun does a certain amount of damage, so that 1 railgun blast and 3 machinegun pellets will drop a normal player. Doom is more arcade then both of those games and should allow itself some liberties on behalf of entertainment.

I would not like to see such a change in the official Zandronum release. Doom works as it is, and I would not like to see it tweaked out on behalf of a small group of competitive players that wish to out rule any change.

On the other side, if you were to make a PWAD that changes gun behaviour to support your taste of competitive play, then you're welcome to do so.

* Opposed to biased variety: see Call of Duty 4 support helicopter for a simple example.

[Llewellyn]

I think that it is fair because both players are subjugated to the same rules. That is fair by definition.
It would be unfair if someone always got max damage with no chance of ever hitting the min-damage, but since they don't that means you both had the same opportunities and the better player won out.

Also, being a good player doesn't require that you understand RNG or the components behind it, only that you can't count your shots. I think that competitively being able to count your shots for more than just a rough estimate is more of a gimmick and really promotes micromanagment, which isn't really a good mechanic for a fast paced shooter.

Fast paced shooters rely on several things: Environment awareness, prediction, and aim.
If you always have good aim but have poor prediction then you will often be behind the other player who has more time to aim because they know where you WILL be not where you are. Anyway, I don't like games that don't have some kind of random element in it. Randomness is what gives games a certain flavor and replayability.

[Spider]

I'm just going to stat something:

In the competitive doom the most used weapon is the SSG right? Well It's the most used in TDM and CTF so.
Every bullet does a damage out of the possible damages: 5, 10 and 15. It's 33.33% chance for each damage.
now the Super Shotgun fires 20 bullets of 5 10 and 15 damages, meaning it will do a total damage of 100 - 300 damage If you do the math.
The doom marine only has 100 hp, a point-blank shot will ALWAYS kill you no matter how different the damage of the bullets will be.

On other weapons like the Rocket Launcher. It does 20-180 damage with the projectile ONLY. Plus the radius damage of 128 damage AND 128 radius; It's a total of 148-308 damage. As said before the doom marine has only 100 hp and he will NEVER survive a rocket right in the face unless you have a Megasphere/Soulsphere/Armor/Megaarmor whatsoever.

Same with the BFG. It's projectile alone does an amount of 100 - 800 damage in multiples of 100^1-8(random), so the doom marine will never survive a BFG either. The BFG trail is a hitscan of the player's view which does a damage i don't really remember but the more closer of the degree your viewing the more damage it is, Not about luck this one.

Just my knowledge about projectiles, hitscans and damages.

[Jenova]

While we're at it, we can turn on spawn farthest because spawns are also all about luck.

[Qent]

Sarcasm? There was a video about how spawn farthest can be more like spawn worst.

[Ru5tK1ng]

While we're at it, we can turn on spawn farthest because spawns are also all about luck.

You're right! While we are at it lets enable spawn protection. It's not fair when I spawn without a weapon and have to face a guy with a BFG!

Maybe some of you should stop being terrible and adapt and overcome the randomness.

Otherwise go play Quake.