YOUR ANECDOTE SUCKS.

SUCKS.

Read the above line again a few times. Yep, your anecdote still sucks.

I consider myself an open-minded kind of guy. However I have yet to see anything that -- even remotely -- resembles the data one would need to make a case that something other than a well-seeded PRNG determines what loot is on what boss. It's possible to gather this data, but so far I haven't seen any that would stand up to any kind of scrutiny or even the most rudimentary application of the scientific method. There is more evidence that the American government planned the September 11th attacks -- by which I mean very, very little -- than there is that WoW's loot is generated in a non-random fashion.

Come on people. Most of you are reasonably smart and well-educated. Wikipedia will tell you how a pseudo-random number generator works and you all know at least a little about the internet and WoW's server structure. You all got the basics of scientific data analysis beaten into your thick skulls in the NINTH GRADE (if not much earlier.) This is not complex, but the majority of this thread is full of mouth-breathing retards who want to whine about their awful luck, as if that somehow, in contravention of all principles of reason, means anything at all.

News flash: If you have done this or plan to you have failed at coherent, rational thought (and would thus make an excellent candidate for the presidency of the United States, as an aside.)

Anecdotes suck because they reinforce beliefs that may or may not be founded in reality. Please for the love of all that is good and holy STOP POSTING THEM.

Edit: For some reason I thought I was posting in the other thread about loot generation. My injunction about mouth-breathing retards applies mostly to that thread and only in small part to this one. Apologies.

It's Nefarian, by the way.

I'll let the rest of your post go, and simply state that the purpose of my anecdote was to show why I, personally, would approve of and/or prefer a tokenized loot system for weapons in addition to armor.

Are you fucking dumb? Did you come here to post an endless rant about politics and our assumptions of a non random number generator? Are you trying to contribute anything or just troll? What exactly is your poiint of being on this forum?

People are not saying loot is determined by your raid leader, people are saying there are too many items in the loot table that have zero value to a guild, taking up drop space and creating a situation where the 'good' drops are never seen.

Take your bullshit anti America and whatever side point you were trying to make about this community and get fucked.

He admitted he overreacted to the comment about preferring loot tokens.

And also, while this is wildly offtopic, his post is not anti-American. He correctly notes that this is next to no evidence the US Gov't planned the 9/11 attacks. And, in fact, there is next to no evidence of it. But the conspiracy theorists there have a more plausible story than the loot conspiracy theorists in WoW. In fact, that'sl probably true. As I had a friend die on 9/11 and blame the people who flew the planes and arranged the whole thing, I got his point. Didn't perceive it as anti-American, either.

If your random number generator is incapable of spitting out a string of 20 zeros in a row, then it's not a very good random number generator. The thing is, never seeing Ashkandi drop is a perfectly logical outcome of having a real random number generator.

The problem with token drops is that they don't reduce randomness. They only change the scope. Think about the number of complaints people have made about how they had "horrible luck with champion drops, especially the helm and shoulder" or whatever. There were only three option on the loot table, and a good random number generator could possibly generate the "never drop item 3" case. It's a real case.

Here's what all of this really comes down to: People don't want real randomization. They want only the random outcomes with low variance from the expected average.

Did you know that slot machines don't use true random numbers? In addition to having their payouts carefully controlled to make sure they always hit 97% payout per hour or whatever, they actually choose their payout frequencies so people never have long strings of no payouts or many payouts. Casinos discovered people would pay slot machines for far longer when the the actual rate of payout roughly matched the expectation of how often you should win. Many people who hit exceptionally bad streaks on the slots would just leave them. (Not the addicted gamblers of course, but the casinos know they are getting money from those people regardless of what payout frequency they use for the slots.)

If a lot of people need a particular drop, wouldn't it actually be more logical to put those drops on the early bosses, you don't actually start out BT farming illidan so those early bosses will get killed more times and the in demand items have more chances to drop.

The additional caveat is that people want low variance from the expected average, only as to the "good drops." When I kill Illidari Council, I don't want an even spread of his loot table. I want very few [Veil of Turning Leaves] and tons of [Madness of the Betrayer] and [Cloak of the Illidari Council] and enough [Belt of Divine Guidance] for our holy priests to each get one.

It is inevitable that individuals will always want their item to drop, but the key really is minimizing outcomes that leave the entire raid grumbling. Yeah, when you've been killing a boss for 4 months, only one or two drops might still be useful and you'll be disappointed a fair bit. That goes with the territory. When we first killed Illidan all his drops (except cloak) were amazing. Now that we have ~15 kills notched, our tanks all have his helm and while it's alright to see a DPS warrior get a good tanking piece, it's clearly not as exciting as a [The Skull of Gul'dan]. That's fine. Illidan's drop table is great, actually -- almost everything he drops is best in slot or close to it. When we click on his corpse every week it's not a question of whether we'll get lucky, but rather of which particular players will get lucky and see their drops. (Let's entirely set aside the issue of Warglaives, which are stupid.)

But if you kill Teron and you get [Robe of the Shadow Council] and [Botanist's Gloves of Growth], no one is happy. It could be your first kill or your 20th.

The RNG isn't nearly as harsh a mistress when she's choosing among multiple favorable outcomes. It's when you have things like [Tome of the Lightbringer] lurking as a Whammy on the board that "luck" can get ugly.

I guess this begs the question, then - are we debating whether or not to solve the issue of 'strings' in loot randomization (which has largely been beaten to death - people "don't want random", they want "semi-predictable"), or are we looking at potential improvements to itemization?

Why is itemization so bad in the first place? Is Blizzard intentionally trying to give us garbage in the form of useless off-spec junk for slots that other classes simply don't have? Or do they envision purpose and use to each of these items under some bizarre circumstance that we, as a min/max-ing playerbase, simply don't see?

I think the problem with the mentality of 'just give us better loot' is two-fold; first, it removes choices for the player (why can't I choose whether I prefer to gear for stamina/armor or dodge/parry for myself?), and second, it doesn't in any way address the issue of randomization. Just because [Lightfathom Scepter] is amazing doesn't mean we want it for 17 weeks straight (purely hypothetical - we haven't killed Leotheras yet, much less Vashj).

So, back on topic - how could the game prevent 'streaky' loot, without becoming manipulated or purely tokenized?

I think there are a variety of good suggestions that have been raised in this thread and elsewhere:
- incorporate a calculation into the loot generation based on how many classes in the raid can use a given item (increases chance to see weapons/trinkets; decreases chance to see idols/relics)
- observe existing loot and factor it into the calculation (every Warrior already has [Band of the Abyssal Lord]? Decrease its presence in the randomization)
- retain a memory/snapshot of loot based on some player factor (guild tag, etc.) - note that this is the most likely to be subject to exploits, etc., and may not be viable

I'm sure there are dozens of other ideas that are equally valid.

To make a long story short, itemization sucks - however, I'm happy to shard every pair of [Windshear Boots] I see if we're getting an equal number of [Dragonspine Trophy] to drop (DST is the most frequent drop from Gruul according to Wowhead... we've seen 1...).

Blizzard can't write off "useless off-spec loot" without implicitly acknowledging that some specs suck at raiding, which substantially contradicts their design goals. And they're never going to relegate "useless off-spec loot" to badge turn-ins, because that's tantamount to saying that hybrid loot is an afterthought. And they can't just scale back hybrid loot until it only has some kind of pitiful 3% drop chance, because then nobody will ever get it. Nor are they going to put together some half-assed scheme for making the game try to decide what loot people in the raid need, because there are just too many factors to consider.

The only realistic solution to these is tokenization for hybrid loot. They need to improve and streamline the whole tokenization process, so that it doesn't involve lugging tokens over to a guy in a capital city with a million conversion options in his vendor table. There really aren't any other reasonable solutions that don't screw over either pure classes or hybrid classes.

There's actually a very easy solution. Design the loot tables so that things like relics are rarer drops than other things on the same loot table. For example, one slot of a boss' 6 possible drops might be:

Bow (Physical Damage): 20%
Cloth Belt (Caster): 20%
Leather Gloves (Physical Damage): 20%
Offhand (Healer): 20%
Trinket (Tanking): 15%
Idol (Moonkin): 5%

Right now if those 6 items were on the same loot slot, they would all have a 16.66% drop rate. It turns out they could implement this system RIGHT NOW with no coding required. (Since the algorithm picks with even probability, just add 4 copies of each item to the loot table except 3 for the trinket and 1 for the idol.) Tokens just mask the problem; they won't solve it.

I think people are a bit off topic in much of this thread. Crappy loot is going to be crappy whether it is tokenized or not, and therefore out of the discussion about random or not.

1.) Poorly designed items should be fixed via reworking the item, not the loot system (RNG).

2.) Multiple same slot same armor (SSSA) items with a different choice of stats with no clear advantage are fun. Multiple SSSA items with a clearly superior choice is silly. See point 1. This doesn't really have anything to do with the RNG, it is an itemization issue.

3.) Multiple SSSA items existing when there is a clear gap in itemization for another slot or armor is silly. Once again, this is not a problem with the RNG, but with itemization.

4.) Mid tier upgrades are fine, as long as they are actually upgrades. Seriously, not everyone is going to farm the last tier of content indefinitely, and yet they aren't going to blast to the end of content immediately either (T6 chest vs non-set loot). Options are nice to have, but they have to actually have a purpose. Items should not be designed for those farming the content, but rather, trying to beat it. Once again, this has nothing to do with the RNG.

5.) Low use items are a different story. Every time some one brings up [Tome of the Lightbringer] as wasting a loot slot it annoys me, because I would put it to good use. You have to be a protection paladin that does a lot of main tanking. Therefore, it is not even a generic paladin, or even a generic prot paladin piece of loot. It is way too strict of an item for a 25 person raid. There will only ever be AT MOST one. This would be a great place to have a turn in of some sort, with multiple options.

6.) Many high use items are only high use because you stack your raid a certain way. This needs to be considered when discussing loot table percentages. We would hope Blizzard takes into account an example raid when designing loot table percentages (its obvious they don't). However, just because your raid swaps people out for encounters to stack doesn't mean you should screw over the people that bring a consistent and more diverse raid.

7.) Not everyone NEEDS to have everything, but everyone wants to. An entire thesis could probably be written about this point, so I probably won't elaborate further.

All that being said, I think there is room for badge loot. I do not think everything should move to badge loot, and I think the items should be comparable to the drops, but different. This would basically be a prize for..."You've killed big scary guy 5 times and haven't gotten a shiny purple yet? You poor bastard. Here, this might help you."

It doesn't cheapen any of the raid loot. You've put in the time. You feel good about yourself because your raiding has progressed your character and you are guaranteed progression. And, though over a long amount of time if necessary, you know that you will complete what you set out to do. There is a tangible amount of progress towards a goal that can be cut short, should luck smile upon you.

Does it really seem reasonable to you to put a drop in with a 5% drop chance? That means, on average, it's going to take a player 20 runs to get an idol. It wouldn't be uncommon to go 50 runs without getting one (8% likely, in fact). And if your Moonkin is out that day, then not only do you still have a shard, but your Moonkin is going to go jump off of a bridge when he finds out it dropped. Meanwhile, wands are still dropping like candy.

Moreover, you get into issues of class distinction with things like "tanking trinket", because something with defense and block value obviously has no use to a druid, and thus is nearly as "hybrid" as the moonkin idol (significantly useful to 2 specs out of 27). That's something else they could easily itemize into a token.

Effectively controlling the loot table slots is the easiest and most efficient way to fix things.

Token, Token, Random :-
The random item should be filled with disirable loot, trinkets, rings, cloaks, amulets and higher use items like physical leather, caster cloth and such.

Random, Random, Random :-
These slots should be broken down into categories;
*Slot 1 - Weapon [Melee Mace, Caster Staff]
*Slot 2 - Random [Physical Cloak, Caster Boots, Tanking Shield, Healer Belt]
*Slot 3 - Offspec [Idol, Libram, Relic]

Not only should they be broken down smartly, but they should be minimized in size.
Having 6 items in one pool means around 16% per item, having 4 means 25% per item, to be frank I think 4 is the maximum any one pool should contain. Weapons having a small pool means that they arent a hugely limiting factor like they have been for so long, if 6 weapons willd drop over the instance then have 2 drop from 3 bosses (one being final boss) and its a fairly good chance your raid might get half of the amount they need before they move on to the next tier.

If they did this, it would mean a higher chance of those items dropping, and you would not need a second (and worse) item in the same slot to drop somewhere else to try and satisfy raid demand (akama bracers, council shoulders).

The other hopefull side-effect of having small item pools would be less chance of them needing to invent poor items (healing haste..) just to fill space.
At first when I saw how small the random item pool was for Illdari Council and Mother I was not impressed, and this was before seeing that half of it was shit in the first place, however if all 4-5 of those items were useful then it would be really quite good.


There really does need to be a few people set aside to actively go over the loot database and check items to make sure they are worthwhile, and not already existing in an equal or better state from an earlier point.
And as stated above, item classification requires indepth knowledge about things to be done correctly, or rather to our standards, is it really so easy to cut and dry things?

These seem to be arguments with the example and not the algorithm itself.

Regarding the Moonkin idol example, you ideally want it to drop once or maybe twice in all of your farming. But remember that our loot objective as players is different from blizzard's objective as game designers. If they shared our objective of making sure the item only dropped when you needed it and never dropped when you didn't, they would make the boss drop a quest item you could turn in, and one reward would be the moonkin idol. Variance, to a point, makes people more likely to keep raiding.

The fact is that a 16% drop rate on a prot paladin libram is ridiculous when prot paladins comprise 0% to 4% of your raid. A 4% drop rate on this, or a Moonkin Idol, is completely reasonable. Yes, maybe the Moonkin will be gone that day, or maybe it will never drop. But this case is much better than the chance of seeing it drop 5 time in 12 weeks, something that could easily happen with the current 16% rate.

Or they could clump the "quirky" loot accordingly. Put the RoS moonkin drops, the Gurtogg paladin belt, the Shahraz libram, and a couple of other such pieces on Naj'entus, and swap them with some of his highly-desired drops like the dagger, Slippers, rings, etc. Now every time you kill Naj'entus, you're going to get some fringe piece of loot for a particular spec that isn't too common, but you aren't left feeling like you just missed out on some amazing drop. At the same time, if you put Naj's dagger on Shahraz's loot table instead of the Relic, for example, it's pretty much a guarantee that your first Shahraz kill is going to make someone very happy.

Mathematically, spreading the loot around doesn't alter the situation much one way or another, other than eliminating the possibility of "unwanted loot" streaks, which statistically aren't very common anyway. Putting all the distasteful hybrid loot on one boss is just going to thin that boss' table out quite a bit (because there are many hybrid specs) and essentially force a scenario where any given piece of hybrid loot has a ridiculously low drop chance. Even if you significantly restrict hybrid loot options quite a bit such that one boss has a bunch of 15% loot drops, it's really not substantially better than 15% loot drops spread out across every boss.

It would be a very minor improvement over the current system in terms of streakiness, but God forbid you make that boss skippable.

No, I understand that a quest item is an easy way of accomplishing the token system you suggested. My fundamental point is that Blizzard hasn't done that because it makes raid drops too predictable. Players want items to be as predictable as possible, but that decreases the amount of time they spend playing the game.

If we want to help Blizzard fix the loot system, a good start would be discarding all options that effectively remove the carrot from the stick and give it immediately to the horse. No matter how badly the raiding population might want that, it is simply not an option. It's not even remotely being considered by Blizzard. At best 2% of drops fill this criteria-- that's how sparingly Blizzard uses it.

So a 4% drop rate is better that 16% for librams, but has other issues? Fine, whatever. It's not an option that immediately disqualified by the higher game design constraints Blizzard has chosen to follow, so it's still a good topic for discussion.

I'm pretty sure I addressed this pretty explicitly. I suggested a "high-but-not-100%" drop rate item, i.e. a 25%-35% chance token that thoroughly covers options for a variety of classes/specs, to make drop chances for every hybrid spec be approximately equal to what any core class would observe. In other words, exactly the same as what they already do for set items. But streamlined in such a way that it doesn't require taking a token to a loot vendor whose item tables are too packed to find anything. It wouldn't be hard, and it certainly doesn't violate the design principles you refer to. In fact, it strikes me as an obvious truth that hybrid classes should be afforded the *same* access to loot that any other classes have. Not better and not worse. Tokenization allows that; forcing them to chase after 4% drops doesn't.

I apologize if what I'm saying is either not clear or just too confusing.

It doesn't affect the mathematical averages, but it does minimize the variance of getting lots of good loot versus bad loot. Consider this. Suppose there are four pieces of loot, A, B, X, and Y. A and B are high demand items while X and Y are undesired.

Current Loot System:
Loot Slot 1: A, X
Loot Slot 2: B, Y

There are four option for loot payouts from this boss:

A,B: Great
A,Y: Average
X,B: Average
X,Y: Terrible

But imagine we swapped the items how Gurgthock suggested, with each slot containing items of relatively even demand:

New Loot System:
Loot Slot 1: A, B
Loot Slot 2: X, Y

Now all four payout options are of average interest:

A,X: Average
A,Y: Average
B,X: Average
B,Y: Average

The expected value of the payout is the same but the variance of possible outcomes has stabilized.

Not only are you essentially just agreeing with me, but you make the assumption that Blizzard puts certain drops into the game just to make people unhappy.

I really do not believe this because I believe their philosophy on items is that not everything can be a godly item and there does have to be some lackluster items but that does not really give them reasoning to put dupes and really useless items. Every boss should have those 1-3 really oh shit I hope it drops item but not everything (no matter the progression level) should be as godly as illidan drops (in their respective progression). Players should get rewarded so again I am not saying dupes and terribly itemized items should be put in but do not expect to see DST/Madness/etc on each boss.


On the topic of relics, I actually do not see a reason against making them token based as they are VERY specific and adding them in like say the AQ weapons (chance to drop off each boss). I would assume a raid would have about 6-9 hybrids so this is somewhat fathomable as a loot category of it's own. Possibly a 10-15% drop rate off each boss in that tier.