Curse

Hanging Man · Oct 07, 2008, 08:56 PM // 20:56

damn man
very good, well thought out and planned thread
it shows you've done your homework, and a nice touch with the graphs really puts things in perspective

but i have noticed, whilst using the A/E sliver armor build that killing a group did decrease the drops
thank you for confirming my thoughts
:]

MisterB · Oct 07, 2008, 11:09 PM // 23:09

Quote:

Originally Posted by My Lipgloss is Cool

I mostly agree with you, save one point.

I believe nothing changes the droprate of Event Items. I've done some experimenting in both 1k and AoE (I went to extremes. Luxon Assassin Halcyon Job farm. For solo, I killed them with SV and necrosis, and for AoE I balled every sin (except the boss) and killed them all at once with SS)

While I got less drops in my runs, I averaged about the same number of Chocolate Bunnies (It was Easter Weekend) per run.

Also, grats on the massive wall of text

_|\
\|

I'm not sure if event items were mentioned in that massive wall of text, but they are indeed exempt from loot scaling. They are on the list in the update notes.

Many people on Guru have been pointing to the SV vs. SoJ thread as proof that rate of kill does not influence drops when the builds used there simply do not kill fast enough.

Good article; thanks for posting. I look forward to your results on party size.

**Marty Silverblade** · Oct 07, 2008, 11:55 PM // 23:55

Interesting. I'll definately be looking at this again, sometime later.

PS: If you plan to add to your post, make sure you have a backup copy (in Microsoft Work or something) incase Guru chews up your post.

relt · Oct 08, 2008, 05:01 AM // 05:01

And people still try to say video games rot your brain or some such nonsense.

You've done some great work here Nechtan, and it is nice to see a very logically thought out post that isn't trying to spark an argument. Just general observation over various farms agrees with most if not all of what you said (there's quite a bit there, so i might have missed a small point i disagree with

). This is definitely something I will want to keep in mind during future farms.

Nechtan Thaumaturge · Oct 08, 2008, 09:01 PM // 21:01

First, Thanks everyone for the comments! I really appreciate them.

Quote:

Originally Posted by Striken7

I think your a bit off on how Loot Scaling works in general. It becomes very easy to see that it IS based on party size when you look at solo farming immediately before and immediately after the LS update.

Before LS, a solo farm run would produce a certain average number of drops. After LS, the same farm run with the same build (same time between kills, same number of foes, same everything else you discussed) would produce significantly less drops (obviously, NLS drops would remain unaffected but the total number still decreased greatly).

Considering all other factors remained unchanged, the only possible conclusions are that Loot Scaling lowered the amount of LS drops regardless of any criteria at all, or that it is based on some factor completely unrelated to the farm itself (skills, speed, foes, etc). Knowing that, and based on evidence collected since then, it does appear that party size has a direct effect on drop rate.

I agree with you that drop rates changed with loot-scaling, but I would counter by saying that ANet probably installed the kill rate code at the same time as loot scaling; therefore, you're seeing the net effect of both here. Unless you are talking about a fairly slow build (at least 20s between kills) then the kill rate effect should dominate.

If all goes well, then I'll have more to say about this next Monday.

Quote:

Originally Posted by Fay Vert

Q. Why do people farm?
A. To get lots of gold

Q. Why do people want lots of gold?
A. To pay for the gold sinks!

Gold sinks drive the demand for gold, if there were no gold sinks there would be no desire to farm.

I'd agree with Senrath that that isn't the entire story. Without gold sinks, prices would tend to inflate, which would necessitate more farming, which would tend to cause prices to inflate, and so on. So, we'd be farming to afford 10 million gp Superior Runes of Vigor or some such thing. However, I think there is some truth to what you say, especially since the prices of armor and basic weapons (just about everything you truly need) are fixed.

Quote:

Originally Posted by Shursh

second, I'd be interested to know a couple things about how you gathered your data:

1. you did all your test runs with either a "1k" or "aoe" build, but were they all done with the same character? the same account?

2. at what time of day and what day of the week did you get your data? were a majority of your drops/data gathered in one sitting?

I used my Elementalist for all the trials, so same account, same character. I generally play at night, but I didn't make any effort to do all the trials at a similar time, some were done in the early afternoon. Also, I gathered the data over the course of about a week, so different days, different times, and probably seven sessions.

Also, speaking generally now, it is possible that some of the factors that have been mentioned (number of players online at a given time, time since last logon, etc.) bear on drop rates, but I probably won't look into them, for a couple of reasons. First, it's difficult or time-consuming to test a lot of these hypotheses. For example, who really knows how many people are online at any given time? You could make the assumption that more people play on the weekends or count districts or some such thing, but that's not very exact. Second, certain of the factors might not be all that useful, even if they do prove true. Let's say logging out for half an hour makes your drop rates double, would you really want to do one run, then log out, then wait half an hour, then log back in, then do another run, and so on? Lastly, the effect of kill rate on drops is rather pronounced, so there's not a whole lot of opportunity for other factors to influence the drop rate much (assuming they act independently). If you delay twenty seconds between kills, you're already getting a 90% drop rate. Assuming double drops are a fairly rare thing in any circumstance, that only leaves 10% or so that you can gain.

Quote:

Originally Posted by A Simple Farmer

Question, how big was your sample size for NM and HM

raptors? You say, a number of runs. But how many?

I alluded to the numbers in the post, but in re-reading it I wasn't very clear. So, to be precise:

Normal Mode:

0s --> 10 runs
5s --> 5 runs
10s --> 5 runs
15s --> 5 runs
20s --> 5 runs
30s --> 5 runs
50s --> 3 runs

38 runs, 911 kills

Hard Mode:

0s --> 20 runs
15s --> 10 runs
20s --> 5 runs
25s --> 5 runs
30s --> 5 runs
40s --> 5 runs
60s --> 3 runs

53 runs, 1267 kills

I only did three runs for the long intervals because they take awhile to do and they aren't very interesting from a profit standpoint anyway.

So there it is; mind you, that's not a huge amount of data, but the pattern established itself pretty quickly and without much perturbation, so I decided to pull the trigger and write up the post. I will be adding to the data set in the future, but for the time being I'm more interested in looking at the party size question.

Quote:

Originally Posted by My Lipgloss is Cool

I mostly agree with you, save one point.

I believe nothing changes the droprate of Event Items.

Yep, that is correct.

Event items are non-loot scaled, or what I've been calling NLS. To model event weekends, you would add about 5% to the NLS drop rate for each special item available -- I don't remember the exact number, but gergnad had a thread on that recently. You would also have to adjust the value for the NLS equivalency depending how valuable the event items were. In any case, I can't imagine a situation where you would *not* want to kill everything as quickly as possible during an event.

Yawgmoth · Oct 09, 2008, 01:24 AM // 01:24

Great job there.
It confirms what I kept saying since after several days post the update, but I was too lazy to make a long detailed post with enough research data to back my theories up.

Just one thing - I don't see any mention of the First Minute Drop Reduction (FMDR) antifarm code (significantly reduced drops during the first ~60 seconds after instance creation) in your post, which could have affected some of your results.

To clarify for everyone: Back in June'05 Anet made first attempts to fight extreme farming and botting. They added the well known antifarm code (repeated farming of the same instance multiple times reduced the number of drops), which was removed with introduction of LootScaling, but they added a much less known antifarming code that reduces drops in the first minute since instance creation. Since the most popular farming runs of old times took at least a little over a minute from start utill first kills, most farmers simply never noticed it.

Now the most important part: FMDR was never removed and exists upon today!

*It affects Solo farmers just aswell as full parties 8/8 Human Players.
*It affects Normal Mode just aswell as Hard Mode.
*It affects areas designed for 8 playes as much as 2-man areas of Pre-Searing.
*It's not always active - may be not active at all when you start playing but can activate as early as the 2nd time you enter the same instance.
*Once activated it's very likely that it will affect you even when you'll enter a different zone, even if you do it for the first time.
*I'm not sure how to get clear of this code, maybe it's about time, maybe the old methods of doing missions work, hard to test that as this code is so easy to be activated again. I know that simply logging off for a day doesn't clear the code.
*The drop reduction is very strong, at least 75% chance for a NoDrop added on top of usual lootscaling.
*I can't say it's exactly 60 seconds - it seems to vary +/-5seconds and after that time FMDR simply turns off.

It's VERY easy to test and prove the existance of this code. Just find a place where you can make very fast kills in fist seconds after entering an instance, either solo or in a full human party, and do a few runs where you kill right away or wait 60 seconds and then kill.

FMDR doesn't usually affect Raptor farming as running to cave, gathering all the aggro and killing takes more than 60 seconds. But it is possible to run in fast using a speedboost without grabbing all aggro and just nuking 20+ of them, get 0-2 drops, type /age and see 0 minutes. (NM only, before Shadow Form damage reduction nerf also HM)

Oh, and if you need people for some testing things count me in.

Chthon · Oct 09, 2008, 03:31 AM // 03:31

1. Very well done.

2. [nvm: Upon re-reading, you're not saying what I thought you were saying. You would agree that longer non-fighting time points towards slower killing becoming more profitable?]

3. While I have no data to back it up, my impression has been that special event drops are handled by an entirely separate parallel loot system. Outside of quest items and holiday items, as far as I know, non-boss monsters do not drop multiple items -- or do so only very rarely -- but holiday items frequently drop from the same monster as regular items. From that I would infer that holiday items operate outside the normal loot system like quest items do.

Senrath · Oct 09, 2008, 03:49 AM // 03:49

Quote:

Originally Posted by Chthon

2. [nvm: Upon re-reading, you're not saying what I thought you were saying. You would agree that longer non-fighting time points towards slower killing becoming more profitable?]

That's only true to a certain extent. While it seems that waiting between kills will increase the amount of drops you get, if you wait too long the run becomes really inefficient, even with the increase in drops. That means it's no longer as profitable.

Dreamwar · Oct 09, 2008, 01:41 PM // 13:41

Awesome post, very interesting too

hallomik · Oct 11, 2008, 06:16 AM // 06:16

Your post provides some good data to support a position I've held for a little while now.

I was persuaded by TMakinen's original testing. He further speculated that the gold value of the recent drop also works with the time interval to determine if the next drop will materialize or not. That is, if a lockpick drops, odds are increased that the next drop will not materialize for a given time interval. Conversely, if a low value item like a 20g skeleton bone drops, the odds of the following item materializing are only slightly reduced.

Your work doesn't prove or disprove that notion, and I'd love to see the theory put to the test. If it is true that the merch value of each drop also affects the liklihood of the next drop, that suggests that the better farms would be ones where a low value item can be converted for much more gold. I've actually been operating under this assumption when I farm Pongmei valley for granite. Stone horns have a merch value of 20g, but they can be salvaged into an average of 6 granite slabs which sell for 0.6 times the trader value for 10 granite. I will sometimes get 2 or 3 of these items dropping in quick succession suggesting the interval effect is small on these particular drops.

_Nihilist_ · Oct 11, 2008, 11:41 PM // 23:41

Nechtan,

This post was absolutely WONDERFUL. I was trying to figure out how the SOJ/SV theory was working the way it was, it seemed a bit bugged (not that the testers blew any data or modified data - they had a great post as well), but the data was too similar, suggesting that it was actually the kill rate that was the determining factor rather than single kill vs. multi-kill data.

Thanks for the great work, your mathcraft/scientific method rocks my face.

Dae GW · Oct 14, 2008, 05:07 AM // 05:07

O M G

This is really wonderful! You must have spended so much time to write this for the community, i really votesticky this. It´s just amazing!

**Marty Silverblade** · Oct 14, 2008, 07:52 AM // 07:52

Quote:

Originally Posted by Dae GW

This is really wonderful! You must have spended so much time to write this for the community, i really votesticky this. It´s just amazing!

I've linked it in the new farming sticky already.

gerg-nad · Oct 14, 2008, 08:25 PM // 20:25

I think the article is very good the only thing its missing is a statistical analysis. What were your p-values and stated null hypothesis. With the raw data from your Drop% and Kill Interval(rate), one could run a regression, as these are both continuous variables. Also you could run a General Linear Model to compare HM and NM using the Kill Interal (rate) as a covariate.

Summary data can be misleading as its not accounting for the noise in the process, so it would be much more convicing if your conclusions had the statistical analysis where we use the appropriate F, t, or even Chi-square tests.

RudyNam · Oct 16, 2008, 04:51 AM // 04:51

Amazing, I loved reading it, even though I would have liked to have a small bit at the top summarizing it all.

Spiritz · Dec 09, 2008, 12:47 PM // 12:47

Ive not read all the replies on this topic yet as ive just woken up so if i repeat an idea someones said etc i do appologise.
From an outside point of view as well as no1 knows anets exact setup i look at things this way -

Often faster kills gives less loot ( note i said often ) - possible timeframe for kills ?
Also its stated the amount of drops in a zone are determined on entry but do ppl account for other variables - x% for a drop then routine chooses a drop at random , say zone has 350 foes in hm an thats a possible 350 items ( i can dream cant i lol ) - for all we know its % drop then it works thru a list working via drop chances ( % drop - fail and move to next % drop fail an next and % drop pass then end .. like saurian bones from raptor farming.

Ive also noticed that sometimes if you entered a map and for eg near start is 20 foes and 1 boss , say you went str8 for boss u may get a drop but if u killed the 20 first then boss sometimes u get decent drop - is loot scaling at times not only down to speed of killing but amount of kills.

As always we can only speculate as theres no given way to check results as too many variables - specially with ppl saying they did 10 uw runs and got 20 ectos and others sayin they only got 2 ectos for the 10 runs.

Then again we could all be searching for an answer which cannot be given - mainly because were told drops are random and randomness cannot be predicted ( if it could be id be playing an winner the lottery every week lol )

God created GW - satan created the drop routines

HawkofStorms · Dec 10, 2008, 12:50 AM // 00:50

While interesting, 3 points:

a) Your original post's order of the paragraphs makes it sound like Skinny Corpse's testing occured BEFORE lootscaling was implimented, but in fact it occured almost half a year after loot scaling occured.

b) The fact that "drops can vary by each run" means that there is no "definitive" way to know that your particular behavior is what caused the drop rate to vary.

c) While your pattern may be very real, its still not statistically signifigant (although I know doing this hundreds of times would be mindnumbing). 5 runs with each time intervel still isn't enough to "prove" anything mathematically, but it definately does shine a light on things and validate some theories.

Really, I think the best answer would be for someone to do Skinny Corpse's tests again, but this time throw in more variables. Have one person kill in mass and another kill 1 monster at regular intervals. Skinny Corpses original tests never really used a mass kill AoE build. They just showed HOW/WHEN drops are assigned. If drop rates vary after the two people running a 1k vs AoE build while synced so there drops are supposed to be the same, then it would be easier to definitively say that the rate of killing was determinitive.

dr love · Dec 10, 2008, 05:56 PM // 17:56

whats the build u used to kill the raps 1 by 1?
i only know how to kill them very fast almost all at once and would like to see if i can reproduce your results

Unreal Havoc · Dec 10, 2008, 06:25 PM // 18:25

I can add to this discussion, hopefully.

Mainly from farming Mindblade Spectres in UW on the Chaos Planes. I used to use the commonly used:

[build name=Chaos Plains Farmer prof=a/e shadow=12+1+3 fire=12 deadly=3][Glyph of [email protected]][Deadly Paradox][Shadow Form][Fire Attunement][Glyph of Lesser Energy][Mark of Rodgort][Lava Font][Dash][/build]

Then once out of sheer curiousity I used:

[build prof=A/E dag=11+3 sha=10+1+3 crit=10+3][Glyph of [email protected]][Deadly [email protected]][Shadow Form][Golden Fox Strike][Golden Fang Strike][Death Blossom][Asuran Scan][Dark Escape][/build]

and I found that I got alot more drops (as in I filled my inventory) using the dagger farmer than the firemagic farmer when clearing the Chaos Planes. It didn't affect the amount of rare drops I recieved however (ie: ecto and golds didn't really drop any more or less than usual) but it did affect the amount of normal drops I recieved. Also using the dagger farmer did take a fair bit longer to clear with than the fire magic farmer.

Fay Vert · Dec 10, 2008, 11:03 PM // 23:03

Quote:

Originally Posted by HawkofStorms

Really, I think the best answer would be for someone to do Skinny Corpse's tests again, but this time throw in more variables. Have one person kill in mass and another kill 1 monster at regular intervals. Skinny Corpses original tests never really used a mass kill AoE build. They just showed HOW/WHEN drops are assigned. If drop rates vary after the two people running a 1k vs AoE build while synced so there drops are supposed to be the same, then it would be easier to definitively say that the rate of killing was determinitive.

I agree, I did this and posted in that same thread

http://www.guildwarsguru.com/forum/s...66#post4288066

Loot scaling exposed, its down to drop rate, not party size. But disturbingly, I could not reproduce the sync entry phenomenon. I don't doubt it, but I could not reproduce it.