And if you don't feel like re-reading the thread and want a simple summary. Nothing like changing the raid leader, the master looter, the moon phase at the time of the raid or any of the above will have any effect on what drops in your raid. And yes, I'm 100% metaphyically certain.

(just my theory)
At the time the first person enteres a raid (ID) instance, most mobs and bosses get spawned (some bosses are random (opera, ZG) aren't visible and maybe they are spawned later when you trigger something, but at the end it doesn't make a difference if they are spawned upon entering instance or when you talk to an npc.)
Every spawned boss gets his own little ID wich saves his status (alive/already killed), loot and some other things wich you should not be able to change with soft-resetting the instance (rnd bosses like opera (wich one you get) in KZ, nefarian adds, chromaggus breaths, ...).
Maybe not the loot for the bosses is saved but there is a reliable formula wich determines the loot by the bosses id and always determins the same loot with the same boss id, like it determines the same bosses abilities when the boss id is the same.

What happened at Sulfuron? There was complete rollback wich rolled back the save data of the instance ID and the loot in the peoples bags. And a soft reset happened, so the boss was alive again with the same boss id that he had before the rollback and wich was generated when the instance id was generated. So the loot of this boss was the same again. (Although it also could be luck)
At Aran was just the bug, that he didn't got an ID and wasn't saved in the instance. He was handled like a normal trashmob. That's also why the trash normally related to aran also respawned after a soft reset when you had this bug.


Solutions for bad loots/luck from bosses?
There are a lot of possibilities but you have to be very careful that there is still some surprise with loot so people not just farm some sort of marks from the bosses and that the system cann't be exploited.
E. g. A system wich prevents mail armor dropps when there are no Hunters/Shamans in the raid, can easily be exploited with bringing only cloth and plate classes to a certain booss to enforce specific loot. (10 man raid instances)

However, I think there are some solutions possible:
-Removing bad items and duplicate items(e. g. in KZ there are 3 different cloth spell dmg shoes) from the tables and shrink them.
-Increase the dropchances for every item so that series of 20 or 30 weeks where a specific item doesn't drop, gets much more unlikely (there will always be a small chance that an item doesn't drop in 100 runs even when drop chance is 90%)
-Shared loot tables between several bosses (similar to opera event or most trash in raid instances).
-A possibility to exchange unneeded loot to not dropping loot. E. g. a npc in every instance where you get 1 mark for every dropped epic you sell him. 4 such marks together with another mark wich dropps from the endboss you can exchange for an item you wan't but doesn't dropp. (Maybe exclude somt items like mounts, enchants, other endboss dropps)

I see this crop up quite a fewtimes, and I wonder - how exactly are you tying the frequency of drops per week to anyone? If it's per raid force - well, my raid force changes every week. There's always some people sitting out a week, some on holiday etc.
If it's tied to a specific guild - what about guilds that run multiple instances? (My Karazhan groups have never had the Lightning Capacitator drop ever. The other Kara team had it drop multiple times).

There's nothing persistent about raid forces between weeks in WoW, so trying to change the drop tables week after week is never ging to work. The best you can do is implement the system they recently started in EQ2, wherebye if you have already have specific items on classes in the zone, then those items won't drop any more. (At least, I think this is how it works).

There's no facility or mechanism that could realisticaly be implemented in WoW that would allow any form of dynamic raid tracking to ensure that over time each guild gets the average set of drops.

He's not saying that it should be persistent in any way. If there are less drops, each item can have its drop chance increased, and larger drop chances reduce the chance of streaks (cf. Gamlber's Ruin, variance of streak length on win chance).

You're making the assumption that the loot seed is based on raid ID. this is not what the theory says.

the theory is that the loot tables are determined as soon as the mobs spawn. it's what determines the loot tables that is in question. when you killed aran, and then "respawned" him, that didn't prove the theory wrong.

if the raid ID does not determine the loot tables, then of course you'll get different loot. but that doesn't mean that the tables weren't chosen when he was spawned.

Please read practically the entire portion of my post before trying to be an e-badass

I have read the whole thread multiple times. All that is mentioned is possible theories by other people who haven't even posted on this thread. Since you didn't read my post very well I'll repeat it. I was asking if anyone had any IDEAS to what APPEARS to work, NOT THEORIES to the EXACT loot system.

Durandal, I think the above post is clear but let's try again:

You cannot possibly influence the drop tables no matter what you do.

Everyone who thinks they had figured out some system or way to do that is wrong. They don't understand random numbers, chance etc. So they concoct theories that are completely irrelevant. And through the magic of the internet, those theories persist. But they are still wrong.

If you had read the thread, you would realise that your question was not a good one. His response was completely justified. You cannot influence the loot tables. Any "theory" given would be bunk.

You can just make the players saved in the same instance for 4 weeks instead, then have all bosses respawn once a week.

Another way is to to have like "raid ID", if more than half of the player who enter the instance at the first time was the same as last week. The raid ID will be the same and certain loot will have smaller/higher chance to drop. Of course this can easily be exploited, but what would the point be? You don't want to get the same loot every week, do you?

I think what people miss are the huge deltas between items having a drop rate of 5%, 10%, 15% or 30% from a boss.

5% drop rate -> 90 kills required for a 99% certainty to see the item, 45 kills for a 90% certainty

10% drop rate -> 44 kills required for a 99% certainty, 22 kills for 90% certainty

15% drop rate -> 28 kills required for a 99% certainty, 14 kills for a 90% certainty

30% drop rate -> 13 kills for a 99% certainty, 6 kills for 90% certainty

If you look at weekly RaidIDs the conclusion is that items with around 10%-15% drop rate are really not guaranteed in any sort of way during the normal time a guild farms an instance (around 20 RaidIDs as a rough estimate).

And that's all for one drop from one boss. Taken together with every item for every player (some of which you want multiple of), your chances of having a few important items never drop is pretty high.

You guys have done a good job of showing what is indeed a bit broken. It really doesn't make sense that someone needs to spend >5 months in the same raid instance. It was "soul draining" back in the Molten Core days and perhaps the only thing worse was repeating it again in BWL. I think Raiste may be onto something very intriguing, however, although I'm not sure his idea goes far enough.

For argument's sake, let's presume it takes >15 weeks to beat an instance entirely and kill the end boss once (or a small handful of times). This would be an average of 2 1/2 weeks per boss and in an instance like SSC, I'm not sure it's a terrible estimate. Yes, the Nihilum's of the world (and the EJ's and many of you) can do it faster, but figuring on an average of 1 1/2 weeks for the first 5 bosses and 4-6 weeks for Vashj is not so awful a guesstimate.

Now, in a world where that represented some kind of endgame, you'd kill Vashj 12-14 times to get all the set gear from her. It's a bad example with Vashj because of the relatively easy progression in Hyjal/BT's early sections but bear with it. Once upon a time, we all accepted the mechanic I'm describing as "part of WoW". So if it was 16 weeks to first Vashj kill and 14 weeks to farm her to death for set pieces, you'd be at 30 weeks for the instance or ~7 1/2 months.

Essentially, based on the math above off of Vashj you'd have seen the 15% drop item with a 90% certainty once and if the dice played kind with you, you'd have seend it twice in most cases (non-lumpy distribution). But 7 1/2 months is a WoW anachronism for hanging out in the same place absent there being no place else to go.

To this end, chest pieces that drop before end bosses -- and helms for that matter based on the TBC model of 2 instances = 1 tier -- make an awful lot of sense. Guilds that are now in SSC and TK are going to kill Vashj/Kael even less than those that came before as they try to play catch up. There's a guild on our server that just got Kael after 9 weeks. It means they have an exceptionally high number of Vashj kills (10 already?) and are attuning a few new recruits and killing Kael at least a few more times. But I doubt they get past 6 Kael kills and I doubt anyone on our server really does.

Given the "pyramid model" of boss kills -- Lurker dies a lot to Guild UberLeet but Vashj dies relatively rarely -- maybe the loot tables should also pyramid. Put a lot of the quirky stuff on Lurker who has a relatively long loot table and put a shorter table on Vashj. In fact, maybe Vashj drops 3 tokens and 2 items that are from a not-particularly-long list of particularly wonderful items. You want them all? You're gonna have to farm her. But with the 3 tokens, you are looking at a rough maximum of 9 weeks for your first 27 to get helms (the 3 tokens are simply one of each... you've beaten the big boss, the big boss is stocked like a Nordstroms with all the goods).

I'm not sure it would be so awful to have this bottom up approach to loot. Cross class items sit on the bottom bosses where you are getting in a lot of kills, increasingly the likelihood of the loot dropping not just once but a few times.

A few of the "Qiraji Armaments" mechanisms would hardly be tragic as well as it worked pretty darned well in AQ40 I though. If it means the ridiculous off-spec/PVP/hybrid item that really can't be used by many people is replaced by a token that can be used for 1 of 5 items ensuring it's definitely useful, then all the better.

But a system that acknowledges how often the bosses are actually killed might help a bunch. And when you decide to stop killing the end boss, you at least walked away with set pieces if not all the items on the table.

Note that the better loot is, the shorter in supply void crystals will be and Blizzard ought to keep that in mind with any change to "better loot". Today, voids are a joke of course -- we got 30 in Karazhan the other night -- but if the first 6-8 weeks of loot in Lich King are all useful and there is no way to enchant things, well boo. I'd propose that enchanting has a "make me into something better or worse' as a matter of course in Lich King anyway. In other words, keep the Nexus --> Small Pris --> Large Pris mechanic in place for whatever the future holds but also include voids. So today, 3 large pris --> void. Yes, I know that's ridiculously expensive today, but that's not the point.

When TBC was new, voids were rare but prismatics flowed like water from all the 5-manning. If 3 prismatics could be made into a void, it'd be a nice perk for enchanters to do something else useful and also mean that Karazhan loot -- which if you all recall sucked incredibly badly at the beginning -- could've been enchanted without some of it reqiring disenchanting.

The problem is in convincing Blizzard that this is a good idea.

According to Blizzard loot tables:
* There must be 6 items per loot table (with rare exceptions). Tokens count as 2 items and a quest reward drop counts as 6. Legendaries exist in a loot table with a lot of blank spaces.
* An end boss will have 1 more loot table than any other boss in the zone, or 2 if one of the extra tables contains a legendary or a quest drop.
* Loot from an end boss must be about 1 tier more powerful than any other loot in the zone.
* There should be rarer drops that give players a reason to keep coming back to the zone.

The first 3 of those guidelines are derived from observing the loot tables themselves. The last guideline is based on CM/Dev comments of the past. Given those guidelines, what changes do you make that aren't negative to players?

If you take an item off each end-boss loot table to increase the drop rate of each of the remaining items, then some players will miss out on a nice item for their class to give the rest of the players a better chance. While you have to clear the zone fewer times to get all the "good" drops, that's because there are simply fewer good drops to be had.

A better option would be to take 2 items of each end-boss loot table and make a 3rd table. That would be great for players because you still have all the possible loots and they are dropping far more often. You could even make the third table a "single player quest starter drop" with known rewards. There may be some reluctance on the Blizzard side of things to give an extra loot per kill, but in the grand scheme of a zone, it doesn't really seem that terrible. It's important to note that you really wouldn't want to do this on the current final boss of a zone. It's only something you'd do when the next zone is released to encourage players to keep coming back a few more times rather than just dropping the zone completely.

That being said, the problem then becomes rare drops from earlier mobs that are just very well itemized and/or highly desirable. At some point, the players themselves have to realize that not every item is going to drop enough times to supply every player that wants one.

On a side note, the reason why Void Crystals are so undesirable at the moment has very little to do with the numbers dropping and a lot to do with their utility relative to Large Prismatic Shards. Void Crystals are used in 4 rare weapon enchants, an enchanter only ring enchant, the Surefooted enchant (which nobody really uses because it's simply not as good as the alternatives) and shadow resist gear. You only need the shadow resist gear once per character, so it has a very small impact. The enchants all take equal or larger amounts of LPS. LPS is also used in a huge variety of other enchants. So basically you use your Void Crystals in weapon enchants and nowhere else repeatably. Arguably, Void Crystals need a repeatable use to soak up the excess numbers there are. Whether that's a recipe to turn them into 3 LPS (and wouldn't that make the value skyrocket!) or a new temporary buff item that uses them, it would all help to make Void Crystals more useful.

I've only been in a position to "test" this once when we weren't able to start the opera event because our Moroes had evidently re-spawned. Second time around he dropped exactly the same loot. Coincidence? Possibly, but there were no subs, no-one left the instance, all we did was run back and kill him again 30 minutes later.

On the other side of things we've had Netherspite killed four times in a single week (Though with soft resets between instances), and each time he dropped completely different loot.

The situation is different though, in our case the boss didn't respawn until after a soft reset.

I think I've seen DST drop about 9 times now, while others barely get one. We also get 60% or so of our tokens as Protector tokens. Some guilds get a warglaive every other kill, we've had 2 offhands in ~2 months... and even that is a lot. We've also had one ranged weapon drop, the bow from Archimonde and that person is not even in the guild anymore. There is crying about loot not dropping every week, heck even I do it...

Big deal, loot is random, deal with it :-)

I think the addition of new heroic rewards (and new ways to get the corresponding tokens) is a step in the direction that you mention.

Ive only heard heresay of double-spawned bosses occuring, such as 2 Nefarians in the same instance Pre-TBC, but im wondering if theres any record of this having happened and said bosses being killed. In this sense the player influenced events, such as RaidID, time of first player entering the instance and raid leader would be identical and show that loot isnt predictable simply from one of these factors.

Its strange though how the Blasted Land mobs seem to be the only ones in game to wear their loot as if it were predetermined. I guess though it would be possible to have variants of the mob for a few weapon types, or something that says for example spawn it with Weapon Skin #10001 and loot table #704.

From a programming/data point of view, is the concept of the server creating and remembering a specific drop for every mob in every zone on the server at spawn time, aswell as all the mobs in an instance when it is created not somewhat inefficient, as many of the mobs may never even be killed. Even as the small amount of data to store upto say 6 item codes, it would seem more sensible just to store the id of a corresponding loot table.

Im much more comfortable with the idea that a boss is spawned with limited data, an ID to draw his appearance/behavior/scripting from, location, and the ID of his loot table. The loot table would be the list of the 6 pools of things he can drop. Then on his death, the loot is randomly (seeded by time of kill and whatever else,) selected from the pools and hes looted.

The pumpkin bag in naxx, several days after halloween is interesting though, but i guess its also possible that that mob was somehow flagged for halloween special loot, or created and linked to a loot table containing the pumpkin bag.

Its a massive amount of speculation really, as we dont know when/if things behind the loot system changes, nor do i believe if we could ever proove that the loot is selected at spawn time, or if the mob is just linked to its loot pools at that time, for selection upon death.

I've heard about others having the same behaviour, but never bothered to check. I guess it's just a flag they can toggle.

Well every spawned mob needs to have some data anyway, such as location, type (shared spawn points), FSM-state, current health, probably an empty threat list, timers, etc. Aditional 0 to 10 itemids are not going to kill you there memory wise. And from a performance view: You really get killed if you allocate/deallocate lots of memory continously. It is in general better to allocate once and not change it. Allocating / deallocating memory is a rather costly operation.

But just for fun, let's say one drop is 2 32bit ints (one for the ID and one for count) and we have a maximum of 10 drops per mob. That's 640 bits or 80 Bytes per mob maximum. 1000 mobs thus require ~80KBytes of memory. That's in todays day and age of GB's really no problem at all.

Now I have no clue how many mobs are on a given world server, but I guess somewhere below 10'000, so that's less than 800KBytes and thus less than 1 MB memory required for all the loot in the mobs. Really no problem there.

Well, that's been beaten to death. Loot is determined on spawn. We have the Blasted Lands mobs, we have blue confirmation and it would be really stupid to have several loot generating mechanisms instead of just one.

What happens on kill time is that the loot becomes "materialized" if you can say that for virtual items, and also at this time if it is a boss mob, logs it somewhere. Or rather, it's logged in any case, as that top-list feature on the official website demonstrates, they log that stuff anyway.

It isn't any of those factors. There have been multiple spawns of many bosses with different loot.

The nutcases all then claim "those bosses didn't spawn with the instance" but others do. Then they argue its still based on Raid ID or some other gibberish.

Even if it were based on these things -- and it isn't -- you can't control the Raid ID. You can't enter the instance at a precise fraction of a second etc. The notion that it's set by the raid leader is the most idiotic of them all. Why doesn't the raid leader get the same loot every week from every boss? Then we here it's the raid leader plus some interaction effect with the time. Alas, we still can't control the time.

People need to stop believing they influence random events. It makes them sound stupid or insane or both. And people need to stop worrying about controlling the mythical "seed" because we've all seen the same loot from the same boss. And yet we've almost never seen all the same loot from every single boss. Nor have we ever had a boss that drops the same exact loot every time through an instance's history with our guilds.

Stop looking for something in the tea leaves that isn't there.

Loot is random. Stating that computers can't make "true random numbers" in no way changes this fact.

Someone pestered a GM a few days ago and that was the responce, it's fun >_>

I know its not a fact, just amusing

If I had a dollar for everything ridiculous said by a WoW GM.....