Random Number Generation, Loot, and Superstition

[MacFeegle]

I've seen a few posts recently about Superior gear, and a lot of speculation... I figured it'd be best to compile a number of responses together, for the sake of convenience.

1. Random numbers are random:
Well, okay, this isn't technically true in all cases, but the details of why is beyond the scope of this post. For the purposes of DD loot drops, random is random, which means that what is true for one player will not necessarily be true for others. Loot is determined dependent on the wave, stage, and stage settings, but is still significantly random.

In practical terms, this means that running 15 waves of Teh Bestest Lewt Stage is not a guarantee of getting anything useful, which includes the bonus pets. In a Random Number Generator situation, your chances of an upgrade (or even anything useful) are almost never 100%.

2. Having all players in a game be the same class increases loot quality:
Let's examine this idea rationally.
First, only weapon type drops are effected by hero class - armor is generic.
Second, enemy spawns change (increase) with more players.
Third, weapons that are upgrades over an equipped weapon only display to those who can equip it.

Therefore, having multiple players increases enemy numbers, which increases possible drops, which increases the number of random loot spawns, which provides more chances for upgrades... but those upgrades will only be easily visible if they either an armor (which spawns for anyone) or a weapon type that your class can equip. In short, there are more chances for good gear drops; having all the same hero classes only impacts whether or not good weapon drops are usable by you.

TL;DR: This is a superstition.

3. "Added New Loot Tier" actually means:
... that the upper end of possible loot drops has been raised. No, seriously. There isn't really a lower end to stage loot quality in DD, as evidenced by finding the occasional low-level loot in NMHC. This means that any time a tier is added on top, the overall percentages of loot quality by tier changes.

Specifically, it gets smaller. It works like this:

Stage A has 4 possible tiers of loot drops. This means that Stage A has a base 25% chance of dropping any given tier of loot.
Stage B also has 4 tiers, because stage B is intended to allow all possible tiers of loot to drop. It also has a base 25% chance of dropping any given tier of loot.

(And now, the change...)

A new tier of loot is added to the game.
Stage A still has 4 possible tiers of loot drops, which doesn't include the new top tier. Stage A still has a 25% chance of dropping any given tier of loot.
Stage B includes all tiers, and now can drop anything from 5 tiers, meaning it now has a 20% chance of dropping any given tier of loot.

Remember, random is still random - and when increasing the range of random results, it decreases the chances of any given set of results occurring. Allowing the possibility of a result does not mean that result is guaranteed, unless other results are eliminated. In DD, this doesn't happen - the "best" way to improve gear drop chances is via adjusting the ranges by stage selection, higher Survival Waves, HC and Mix Modes, etc. But random is still random.

4. "X doesn't drop!" "Yes it does, I have X!"
Anecdotal evidence is anecdotal. In a random situation, what is true for you will not necessarily be true for anyone else - and anyone can claim anything they like. If you want to be certain that X drops, find out if anyone in Ranked has one - if they do, it exists. Whether or not it drops for you does not "prove" the non-existence of X, only that you have bad luck.

Anecdotal evidence also doesn't necessarily "prove" that any given stage and setting combination is "better" - consistent good results from multiple sources and tests proves that. Science!

Besides... if anecdotal evidence was acceptable as proof, I'd have "proof" that Mythic gear always has at least one significant minus, usually to resistances, regardless of descriptors like Enchanted.

5. Superstition is superstition:
... which is kinda the point of this post, in general. May as well claim that TE nerfs loot drops for people wearing pants while playing, or on Tuesdays, or that there's a guy at TE who monitors everything anyone says or does in DD and manually edits every drop based on whether or not he likes a player.

Personally, I find the last one much more believable than "I can't get it to drop, it doesn't exist!"

[PenguinPower]

Great thread, interesting read :P. I must say I agree with your view on the 'I can't get it, it doesn't exist', but mainly because I think its aimed at the next stage. Trans items drop a lot on aquanos compared to moraggo, so I think supremes will drop a lot on sky city compared to aquanos. That being said, it'll still be all random of course.

[ssjrockyrambo]

I'm not a coder, but what I do know is, as you said, "random isn't random," and because of this, there could be parts of the code Trendy has created that, if enough people play with varying playstyles, are eventually going to benefit some players more than others.

Has anyone actually looked through ALL the code to make sure there aren't any weird things in it that might increase good loot or decrease it? Is it truly "random"?

[Loperz]

Your point 2 is incorrect.

Those of us that support the All of the same class for survival, so it because it increases the Armor/Weapon ratio. This can be proven quite easily. If you run 1 of each class you will have the floor literally covered in weapons, with almost no armor. If you run 4 of the same class you reach a ratio where their is still less armor than weapons, but it is not so one sided. So the idea is the more armor that you actually get to drop, the better chance some random item you want will actually occur. + more armor drops is more shop fodder, since armor is easier to sell than weapons.

How it basically how it appears to work is that for each additional class u add you get more weapons. So with 1 char u might get 30 weps 10 armor before hitting the drop count. While with 4 chars of the same class you might have 80 weps and 20 armor. But if you run 4 of the same class you get closer to the 90 weps and 10 armor or even worse. I can't explain why this occurs, but it is my observation after observing what i feel are enough actual survival waves.

The rest of the points are valid. Basically the amount of players farming high end survival has increased, but the newer people reaching this tier haven't realized just because you farm waves 25-35 or whatever of survival, most of the loot is still crap.

At the end of the day if you pick up 50-100 random weapons, and trans armor while farming survival if 1 item is actual an upgrade consider yourself lucky, and be happy.

[Iyashii]

Interesting read.

More players should have a greater understanding of fundamental game play mechanics.

[CrzyRndm]

Is it truly "random"?

A properly coded RNG will work off a multitude of factors. These can include (but are not limited to): undisplayed digits from a previous output, time (normally undisplayed digits again), Hardware (eg. current rate of data transfer from the hard drive), or other unrelated internal factors (eg. a random stat on a randomly selected item of loot currently on the ground).

The chances of this not behaving indistinguishably from a true random number in a game that only uses 3-4 digits of the output are so tiny. So long as the majority of the factors are unrelated to what the RNG is being used for, the randomness of psuedo-randomness is not worth worrying about

[kc7zzv]

A properly coded RNG will work off a multitude of factors. These can include (but are not limited to): undisplayed digits from a previous output, time (normally undisplayed digits again), Hardware (eg. current rate of data transfer from the hard drive), or other unrelated internal factors (eg. a random stat on a randomly selected item of loot currently on the ground).

The chances of this not behaving indistinguishably from a true random number in a game that only uses 3-4 digits of the output are so tiny. So long as the majority of the factors are unrelated to what the RNG is being used for, the randomness of psuedo-randomness is not worth worrying about

As a programmer, I totally agree that the RNG they draw numbers from is close enough to random that we don't care, but...

They might not have coded it to be a totally random (Flat/Uniform) distribution. For all I know, it might be a bell-curve, or the theory that any piece of equipment knows how many pieces it and its ancestors replaced, and calculates quality based on that might be true. (Both of these are random ideas that I assume are false. The first, because I see no reason for it, and the second, because nothing I've seen supports it. ) Sorry, but you've given my no reason to believe it is truly random, in the way I think you are implying.

[MacFeegle]

@Loperz:

You are correct, sir... having more weapon drop options does indeed force a lower armor drop percentage... and having a Series EV counts as two different characters as well. I phrased it how I did specifically to respond to players who apparently believe that hero class has a direct impact on what armor types and quantities drop, even when playing solo.

In truth, All Same Class Survival groups aren't doing much more than shifting drop ratios - providing more chances for a good drop, but not actually impacting good drop percentages in any direct way. Rolling the dice more often, if you will.

@ssjrockyrambo:
Depending on what specific RNG function is being used, "random" results can be ridiculously predictable. I'm not a programmer myself, but I've done some scripting in the Unreal engine, among others... and while I can notice certain patterns, I don't know what TE uses as its variables, drop algorithms, etc. All I've got is educated guesses and anecdotal evidence.

@kc7zzv:
I generally assume that any given RNG loot setup does not have sophisticated determination of drop options. Most of the time, there's the simple Effort of Creation factor. From an entirely subjective position, there's also usually the WTF factor, as in "WTF?! Why would the game think I actually need something like this? I'm an X, X doesn't need this crap!"

Also from an entirely subjective position, I don't actually like RNG loot... but I like being kicked from someone's online gear farming game on account of ridiculous superstitions even less.