Curse

SotiCoto · May 15, 2007, 11:08 AM // 11:08

Ok... I figure some things need to be said about drop rates.... data needs to be compared and things need to be analysed. However, I have noticed quite a few things during play that many people might not due to... well... their being more rhetorical and less rational perhaps. There definitely do seem to be certain unspecified things that influence drop rates.... and quite a lot that probably don't.... so... I'm going to try and put them together.

#1. Populace Overfarming.
Myth. Total and utter myth, of that I am certain. Yet it is a totally prevalent one. People complain about certain spots being overfarmed because it is the easiest thing in the world to blame other people because you don't get good drops. I've tested this one a lot... farmed in areas where other people are swarming... and farmed in areas where there is nobody else around at all. In both cases, I've had times where I was getting tons of gold and loot abounding, and others where I got nothing much. Who else was there didn't seem to influence the results at all. Not in the slightest. You can even do the same farm as the bots at Granite Citadel and get just fine drops (provided you don't mind being mistaken for a bot yourself).

#2. The Anti-Farm Code.
~~ 2a. Zoning Repeatedly.
A-Net removed this. You can zone in and out as often as you bloody well please and it won't influence the drop rates. It doesn't matter if you're killing the first one or ten enemies, or going in there three times per minute. This won't change drop-rate.
~~ 2b. Killing Same Enemies Repeatedly.
Possible, but it seems unlikely at present. While my Necro's efforts always seem to wane over time, I've had other farming runs that show no diminishing returns over time, even when I'm killing just the one group.
~~ 2c. Getting Rares Repeatedly.
If any sort of anti-farm code is in effect, this seems the most likely.... though probably weak at best. With many characters I have had quite a lot of golds initially from farming, and then had them wane to a low level and stay there... and stay there until I haven't had any golds for a while despite killing a lot (for whatever reason). O'course that doesn't stop there being an occasional gold-spike...

#3. AoE Slaughter vs Individual Kill.
This almost certainly IS a factor... at least when solo-farming. I have reason to believe that when a lot of enemies die in very rapid succession or simultaneously, that the drops are intended to cycle between the people present to keep it more fair.... but if only one person is there then some drops may not happen at all. There is quite likely a timer in effect on this which resets after a few moments... In other words, killing enemies one at the time should yield better results than AoEing them down.
[Edit]: Spacing out your kills seems to drop more scaled loot, but not more rare stuff. This pattern is seen consistantly by many people now. If you're farming for golds alone, it is apparently still alright to just nuke the fetchers!

#4. Loot-Scaling + Hench-Flagging.
As far as I can tell, how much chalk / blue / grape stuff you get is going to be the same irrespective of whether you're solo-farming, 8-group farming, 8-grouping with the herohench parked out of compass range... Hard-mode OR Normal-mode. You will get one person's worth, and that seems to be it.
As for non-scaled stuff... you will get 8 peoples worth if you are the only person in compass range, irrespective of what is flagged outside. Normal-mode and Hard-mode much the same. Alive or dead doesn't matter; only whether they're in range or not.
Just remember though... herohench will take their share of any raw cash you find, no matter where they are on the map and whether they're alive or dead.... so only bring them if you absolutely have to.

#5. Hero/Hench versus Real Players.
The same. Totally the same. If you're in a party of 8, you will get 1/8th the rare drops and the same number of normal drops irrespective of whether the other 7 are meatbags or AI. Any suggestion to the contrary is a myth... Totally.

#6. Enemy Variety.
Can't really be sure on this one, but I don't think it makes any difference at all. I seem to get exactly the same number of golds, other rares, etc from farming runs with only one type of enemy as I do with every species under the sun.... Only thing enemy variety influences is WHICH items you get (and they will tend to have the same probability of a rare from one to the next).

And I won't patronise anyone any further by noting that Hard Mode gives better quality drops but not better quantity...

If anyone can add anything to this, please feel free to tell me.

Koning · May 15, 2007, 11:25 AM // 11:25

I will post here the same as in that other thread than:

Quote:

Originally Posted by Koning

Anyway I've found another interesting thing which has to do with droprate; When we kill a mob at the same time all together (lets say with ss/sandstorm/soj and the like) it seems that you get ALOT less drops than when you kill them seperatly (Warrior, dervish etc and sliver armor/spoil victor in some extend).

For example when I go to UW with my sandstorm e/me, and kill 1 group of smites, there's normally only 1 white/blue drop or coin drop. (rares and ecto's excluded, you just have to be lucky to get those)
When I lure 2 or more groups together and kill them, I STILL get only 1 white drop/coin drop (again, rares excluded). This way it seems its more profitible to kill groups seperatly, and not lure them together.
However, when I go to the smites with my warrior (w/d most of the time) he seems to get more drops than my ele. I think the reason is that he doesn't kill them all at the same time.
Btw, HM or NM is the same.

Same drop behavior with minotaurs/trolls and all other monsters.

Any1 more noticed this?

So, I'm also noticing that killing all enemies at the same time gives less drops than when you kill them seperatly. And I did enough runs to say this is reliable information. Tho I also want to know if more ppl experience this.

okidoke · May 15, 2007, 11:43 AM // 11:43

Quote:

Originally Posted by SotiCoto

#4. Loot-Scaling + Hench-Flagging.
As far as I can tell, how much chalk / blue / grape stuff you get is going to be the same irrespective of whether you're solo-farming, 8-group farming, 8-grouping with the herohench parked out of compass range... Hard-mode OR Normal-mode. You will get one person's worth, and that seems to be it.
As for non-scaled stuff... you will get 8 peoples worth if you are the only person in compass range, irrespective of what is flagged outside. Normal-mode and Hard-mode much the same. Alive or dead doesn't matter; only whether they're in range or not.
Just remember though... herohench will take their share of any raw cash you find, no matter where they are on the map and whether they're alive or dead.... so only bring them if you absolutely have to.

What does non-scaled stuff mean eg. rare materials, armour?

I agree and understand the rest of your views and farm accordingly.

Guildmaster Cain · May 15, 2007, 12:06 PM // 12:06

Hmm nice, I hope this bans some of the Myths that some ppl spread.

AoE Slaughter seems interesting to investigate... So does Henchflagging if you dont care about money sharing.

Though you cant base anything on a few runs in a certain area, as some ppl try to do. Sometimes I get 5 runs no rare at all and another run in the area suddenly 4 rares drop and more good stuff. Its all calculated in the chances of loot-roll.

FoxBat · May 15, 2007, 12:22 PM // 12:22

Let me clarify by combining your points; loot scaling is connected to kill rate (which is noticable when killing monsters quickly all at once, instead of slowly one at a time). You will get nearly full drops killing elona minotours with -10 degen than say halfish with sliver armor. (It will take ages, but you will get more drops per critter.) Therefore, kill slowly enough and you won't be affected by loot scaling at all. Note by "kill" I mean deaths within a short time window, so if you take 5 minutes SoJing a group of 10 down and then they die at once, you will get less drops then if you spent 30 seconds on each one with a sword warrior, even though the farming time is the same. One person claimed ~30 seconds between kills was the sweet spot for full drops but I haven't tested that number.

As far as I can tell, this means that things on the loot scaling exception list, should *not* be affected by kill speed, which would be confirmed by all the rares and ectos still dropping in speed farms.

KartMan · May 15, 2007, 12:40 PM // 12:40

Quote:

Originally Posted by SotiCoto

#1. Populace Overfarming.
Myth. Total and utter myth, of that I am certain. Yet it is a totally prevalent one. People complain about certain spots being overfarmed because it is the easiest thing in the world to blame other people because you don't get good drops. I've tested this one a lot... farmed in areas where other people are swarming... and farmed in areas where there is nobody else around at all. In both cases, I've had times where I was getting tons of gold and loot abounding, and others where I got nothing much. Who else was there didn't seem to influence the results at all. Not in the slightest. You can even do the same farm as the bots at Granite Citadel and get just fine drops (provided you don't mind being mistaken for a bot yourself).

I don't agree with you on this point, please check my following post
http://www.guildwarsguru.com/forum/s...&postcount=162

SotiCoto · May 15, 2007, 12:49 PM // 12:49

Quote:

Originally Posted by KartMan

I don't agree with you on this point, please check my following post
http://www.guildwarsguru.com/forum/s...&postcount=162

I see your counter-theory... but what is it based on?

You see... there is this nasty little problem with the human mind known as the cluster illusion. It is a side-effect of a mind designed to identify patterns.... because sometimes a seqence of events can occur randomly in such a way as appears like a pattern. We therefore can only work in terms of relative probabilities... Randomness is in fact capable of replicating patterns by accident.
Furthermore... the human mind is more attuned to finding things wrong or undesirable than the other way around. You may register good findings as "normal" or "default"... but when things start going a bit wonky, the average person will assume something is wrong somewhere... that there is some sort of deliberate interference.

Now.... what you need to keep in mind is that just randomly.... you may be denied any decent loot for days on end of very frequent farming runs.... or alternatively you might get absolutely tons of gold.... and believe it or not, that can still be well within the bounds of random chance.
Now... if it gets toward weeks, maybe then you can start suspecting foul play.... but even then it has a probability of still being random.

Now... on the last triple-green-drop weekend... I was considering blaming my inability to get the Scar Eater on overfarm.... but thinking about it rationally, and taking into account the number of bosses I've soloed since.... 13 kills (about 40 attempts total, though I died more often than I succeeded) to get one green... even in a triple-drop-weekend, is certainly not vastly improbable. I was simply unlucky...
Later that day I teamed up with other people to farm greens repeatedly in Arborstone, and I personally got 2 greens the first run, and one each the next two as well (total of 3 bosses)... while nobody else in the group got any. It was just random. And for the record, tons of people were doing the same farming run.

So.... what makes you suggest otherwise?

FoxBat · May 15, 2007, 01:03 PM // 13:03

I'd be interested in seeing your data which supports an overfarm nerf, but your theory of the mechanics behind it is lacking. There isn't really any way for drops within seperate instances on the same server to be related unless Anet had specifically coded it to work that way. If they went with the common number generator which they probably did, it is just a seed based on system clock to fractions of a second precision. How often a new seed is created, whether its just one per drop or simply one used for your entire stay in that instance, doesn't relate at all to multiple instances on one server. Servers are not bound to in-game locations, so normally a server is running many, many instances of all sorts of different missions and areas at once, with none of those people complaining about their drops.

Koning · May 15, 2007, 01:05 PM // 13:05

The old griffons also were dropping frequent goldies while tons of people and bots were farming them...

KartMan · May 15, 2007, 01:08 PM // 13:08

it's solely based on my own (little) farming experience, example zehlon reach, I think I was among the firsts to try there, gold galore and nearly every monster did drop something, now that it is swarmed with Mo/D, I barely get 3 or 4 items (including gold coins) from the whole run.

what makes me think that everything is pre generated ?
I was doing a run in gates of kryta HM, in one monster pack, I received 5 life steal from one archer (normal monster) when he attacked me, guess what, he did drop a gold dead bow with a vampiric string on it.
My guess is that there is a large part of "randomizing" in Guild Wars and with loot/drop, and probably a test when you kill a monster to determine whether he will drop something, and if he did what part of "his inventory" it will be.
This could explain why if you AoE kill some mobs, you only get a few drops, since the "test" for drop or not will likely get the same "nodrop" result.
(I can't remember in which post I read this, it was related to random numbers I think)

hope it helps.

edit : I found the topic : http://www.guildwarsguru.com/forum/s...php?t=10153337

also you may find it interesting the following link http://www.rohitab.com/discuss/index...howtopic=24727

believe what you want to believe, I just wanted to give my point of view

cheers

Patrick Smit · May 15, 2007, 01:22 PM // 13:22

SotiCoto, the exact consequences of point four are not completely clear to me.

IOf u flag hench far away and they grey out, would that result in mor drops for me? I interpret it like that, but maybe I'm just reading it wrongly.

letshuntx · May 15, 2007, 01:24 PM // 13:24

I believe that ANet has put in code to encourage spreading out your kills.
Instead of nerfing drop rate in certain farms that drastically improve it for a short period when you change campaign or area or something like that.
I found that when I started hard mode in tyria drops were amazing then went to crap. then when I started factions hardmode on the first missions the drops were amazing again and then got reduced again.

I could be wrong but thats what I believe

Bug John · May 15, 2007, 01:39 PM // 13:39

Quote:

Originally Posted by SotiCoto

I see your counter-theory... but what is it based on?

His theory is based exactly on the same as yours, your personal feelings.

I tend to agree with him, as I'm personnaly convinced that the number of instances in a zone has an influence on how good drops are.

I play on euro servers, and I've never had such good drops than since I came to taiwan for a traineeship : I play less, but I play when european players aren't farming.

Haven't you noticed too how good drops are when a new campaign has just been released ? And how fast this quality disappears as more people reach the new zones ?

I know gaile said this isn't true, but do you really think anet would explain how the anti farm codes work ? That would make these codes useless.

Believe me or not, I don't want to convince anyone (and you should do the same as long as you don't have read the source code yourself

).

XvArchonvX · May 15, 2007, 01:48 PM // 13:48

There are 101 myths out there that people have made up to either explain why someone didn't get the drops they wanted from farming or to just blame A-net. Almost all of them are pretty much like superstitions in that they are simple rules conjured up by the simple minded to avoid accepting the fact that random drops mean that you just might farm for days on end and never get what you want.

KartMan · May 15, 2007, 01:48 PM // 13:48

Quote:

Originally Posted by FoxBat

I'd be interested in seeing your data which supports an overfarm nerf, but your theory of the mechanics behind it is lacking. There isn't really any way for drops within seperate instances on the same server to be related unless Anet had specifically coded it to work that way. If they went with the common number generator which they probably did, it is just a seed based on system clock to fractions of a second precision. How often a new seed is created, whether its just one per drop or simply one used for your entire stay in that instance, doesn't relate at all to multiple instances on one server. Servers are not bound to in-game locations, so normally a server is running many, many instances of all sorts of different missions and areas at once, with none of those people complaining about their drops.

ok this is how I think about the server part of the game :

as I said, the "instances" are just data, whatever its form, to put it bluntly an array, containing various informations (coded or not, example int value 13 for dead bow). Each array corresponds to an instance, which are unrelated amongst themselves, but the functions that "uses"/"works" those arrays are the same for all those data.
so to speak if you have a function that spawns the monsters on a map, it may call another function that determines (probably using a random number generator) the drop and quality the given monster has.
It is exactly here where I think the skew is.

if you now consider the Guild War server you're as a whole (and not only in your own instance), the server may host 1000, 10000 instances or even more, so possibly hundreds or thousands of call to the same function in a very short time window.
I don't think we will ever see this server part code but that's at least how I guess it could be like.

FoxBat · May 15, 2007, 02:14 PM // 14:14

Quote:

Originally Posted by KartMan

This could explain why if you AoE kill some mobs, you only get a few drops, since the "test" for drop or not will likely get the same "nodrop" result.
(I can't remember in which post I read this, it was related to random numbers I think)

You're right about inventories being generated before drops, but the second part doesn't follow. The inventory generation has been that way since the beginning, but the reduced drops on kills in a short time frame is a) an extremely recent change and b) likely an intentional one as Gaile did mention a "complex formula" being used to adjust loot scaling. It's a sensible way to put a "hard cap" on the amount of gold bots can generate per time, much like the soul reaping adjustment.

Also if the game is using a single seed for 8 or 9 drop chance checks in any normal way, the first call will set the number for the second call regardless of the time difference between the two, even "simultaneous" because single-threaded processes can't do anything simultaneously and must make the calculations in order. Additionally if your theory was correct, we should see simultaneous kills trigger all-drops-at-once or no-drops-at-once, mistakenly using the same generated number, and not a consistent no-drop behavior.

The birthday presents thread is silly because the # of presents any individual can open is typically much much smaller than the number of rares they can accumulate, it is unlikely most people can accumulate enough unopened gifts to have any kind of remotely significant data. There isn't a huge variety of miniatures in the first place anyway.

There is speculation about some widespread attenuation of rares in hard mode past a certain point. So to verify over-farming theories you need to, like Soto mentioned, spend some time farming similar enemies in unpopular locales and compare drops.

Quote:

so to speak if you have a function that spawns the monsters on a map, it may call another function that determines (probably using a random number generator) the drop and quality the given monster has.
It is exactly here where I think the skew is.

As I mentioned, in conditions of not overfarming, a single server is running tons of instances of people in various missions and areas. The same function is being called across many instances, yet reputedly drop rates are not being lowered. Why would this specifically target areas that happen to be the same and avoid areas that are different, unless Anet intentionally coded it to do so? Once a seed is set anyway, you could call it an infinite number of times and have a "random enough" behavior, there isn't some degredation of integrity over time, so one seed value being shared across instances again should make no difference. There simply isn't anything relevant for a RNG to share beyond a single seed and # of calls.

ReZDoGG · May 15, 2007, 02:24 PM // 14:24

I noticed that also. That taking on a huge group and killing them all pretty much at once, kills drops. I put together build for Luxon Assassin's farming where I can take on every Sin that pops up there at once, which is over 25 sins. I killed them all, and only got 1 gold armor, and 1 purple armor with like 3 other drops. Definately better to take them a few at a time.

Ringsgold · May 15, 2007, 02:38 PM // 14:38

#3 is hereby confirmed:

farmed the spiders outside rihlon refuge, alternating between using SS and SV to kill them

KartMan · May 15, 2007, 02:55 PM // 14:55

Quote:

Originally Posted by FoxBat

Also if the game is using a single seed for 8 or 9 drop chance checks in any normal way, the first call will set the number for the second call regardless of the time difference between the two, even "simultaneous" because single-threaded processes can't do anything simultaneously and must make the calculations in order. Additionally if your theory was correct, we should see simultaneous kills trigger all-drops-at-once or no-drops-at-once, mistakenly using the same generated number, and not a consistent no-drop behavior.

I'm not sure for the all-drops-at-once or no-drops-at-once, I was rather thinking :

CODE
10 -> nothing
10
10
7 -> luxon pendant
7
7
4 -> weapon
3 -> luxon garb
3
3

well maybe with 10 or 15 more "10"s ...

I only used the birthday's topic for its content regarding random number, I know that the sample one can get is really small for it to be significant.

Quote:

Originally Posted by FoxBat

As I mentioned, in conditions of not overfarming, a single server is running tons of instances of people in various missions and areas. The same function is being called across many instances, yet reputedly drop rates are not being lowered. Why would this specifically target areas that happen to be the same and avoid areas that are different, unless Anet intentionally coded it to do so? Once a seed is set anyway, you could call it an infinite number of times and have a "random enough" behavior, there isn't some degredation of integrity over time, so one seed value being shared across instances again should make no difference. There simply isn't anything relevant for a RNG to share beyond a single seed and # of calls.

that's because of "prediction", nowadays most CPU architecture has this feature I think, if you call a function with the same parameters, it is highly likely that it will return the same "value" or "array" in our case.

Anet doesn't need to nerf farm, their servers are doing it on their own

FoxBat · May 15, 2007, 03:12 PM // 15:12

Quote:

Originally Posted by KartMan

that's because of "prediction", nowadays most CPU architecture has this feature I think, if you call a function with the same parameters, it is highly likely that it will return the same "value" or "array" in our case.

That would be a complete failure of a random number generator. The standard, cheap generator everyone is using is very capable of spitting out near-random results when being fed the same parameters, and if something was fishy anyway, you would notice it affecting far more frequently occuring identical random rolls like damage ranges in isle of the nameless.

I'm not trying to put down the idea that there may be something going on, but blaming the solid, simple, time-proven RNG itself is based on human misunderstandings of randomness, which in fact include long runs of similar results happening sometimes.