Talk:Anti-farm code

Something doesn't make sense here. If time between kills affects drop rates, how can the timestamp drop generation note be true as well? ··· Danny Pew Pew 19:32, 16 September 2009 (UTC)

Well, Drop generation and drop assignment are two different things. The loot is generated when you enter the zone, but the drop assignment seems to be done at time of kill. This is of course only based on my observations, I've tested the loot generation theory a while back. My brother and I did always get the same rare loot (not loot scaled, so we always get itsince we played solo), but the oher drops were randomly assigned, so we did not always both get the exact same non rare (loot-scaled) drops. Zophar 15:40, 17 September 2009 (UTC)

Hmm. I guess that makes sense. I was thinking specifically about Vaettir farming, where I never noticed a large decrease in the number of drops, but I reconsidered my drops and realized that the only things that might not have been dropping were the very things I wouldn't bother picking up - white items. ··· Danny Pew Pew 19:42, 17 September 2009 (UTC)

Player Tests

This needs clarification. The second statement contradicts the first statement. How does time between kills affect drops does it increase the rate of drops with an increase in time between kills or decrease the drop rate? --The preceding unsigned comment was added by 68.143.166.98 (talk).

There are several factors if you want to calculate the optimal amount of items per time and they are both mentioned in the player tests. In first place it is the time between kills. If decreases rapidly if this is low. From then you have the time it takes to setup, the amount of kills you can make etc.. If you only have 30 percent change to get an item if you kill per 5 seconds you will get 60 items and 200 kills after 1000 seconds. If you have 80 percent change to get an item if you kill per 10 seconds, you will have 80 items and 100 kills in the same 1000 seconds. (Not looking at a few of the other factors) There is an experimental curve to be determined and based on that you can calculate the optimal killing speed. -- Karasu (talk) 21:11, 1 November 2009 (UTC)

Nicholas Gift trophies

This rule needs to not apply to those items. It will severely cut down on the time it takes someone NOT INCLINED to spend their game time farming to get their 5 Gifts and will drop that farming market off the face of the planet, good riddance. I'm tired of getting 2 trophies, then 1, then 0, then 0... I feel like I'm watching someone being auto-attacked under SoA. 141.165.170.184 20:45, 29 April 2010 (UTC)

I think the point of Nick was to cater to the casual farmer. You still spend a couple of hours over the time of a week farming for the items, or longer if you want to sell some items later, but it's not something that everyone wants to do. Datrulegend 21:18, 29 April 2010 (UTC)

I was under the assumption that trpohy items were not included in the code... I've farmed tons of area's and always seens the trophy items continue to drop even when my other drops have fallen to unworthy.... 65.6.156.101 11:26, 30 April 2010 (UTC)

Actually, I think nicolas is one of the proofs of many players farming the same area--84.196.119.34 22:48, 2 February 2011 (UTC)

Minute delay "resetting" drop rate?

I saw on another page that If you were to wait about a minute after entering a location, the drop ratio would in fact be similar to normal. Here is my source [1]. I think it could fall under the third * under player tests, but it may be beneficial to define if this too is indeed a valid method to "reset" the drop rate. If true, I think it could mostly help boost the individual players that farm such items, like Nick's requested item of the week and all, along with helping the run-of-the-mill farmer.--Neithan Diniem 05:30, 10 June 2010 (UTC)

I think this might also be affected by the part of the code that seems to reduce drops when too many are being killed per allotted time. By waiting, it reduces the number of foes being killed per time. It is something I noticed as well, sometimes I'll leave my character for several minutes or just AFK for hours outside of the outpost, and always yields better drops per run. --Lania Elderfire 06:13, 10 June 2010 (UTC)

I would be interested to see this in some sort of data form. If we could decide on a way to track drops prior to pausing and then post pausing this could be semi-validated. Anyone with more data-experience have any input? Siris/talk 06:24, 10 June 2010 (UTC)

The only way I can see this working is if a person sync-entered an explorable a few times near a boss, and after drops decline one wait a minute before killing said boss. That would provide for a controlled environment for experiments... Wish I knew this when I did all of those raptor farms... --Neithan Diniem 15:01, 10 June 2010 (UTC)

Disputing "timestamp" test results

The "First (few) foe(s)" and "Time between kills" seems probable and it should also be easy to gain substantial statistic evidence to back it up. But I strongly dispute the "timestamp" theory. I have tried (several times) to set two computers up beside eachother, on the same (wired) connection, with an identic character on each account and I have made sure to enter the same area at the exact same time on each account. I have then slowly and deliberately killed the same mobs simultaneously on each computer and I have NEVER experienced drops to be even remotely similar. And even if anyone tries to tell me, it's because I have to enter at the EXACT same millisecond on each account and thus claim that the game calculates full drop tables for every instance in the game 100 or even 1000 times each second, then I will call the bogus card on this whole theory. Can anyone provide screenshots of the actual drop assignment to back this far fetched theory? --Manassas 10:17, 14 October 2010 (UTC)

Millisecond actually does matter, but not because the game is calculating drops 1000 times per second. What it likely does is use the timestamp as a "seed" for what's called a "pseudo-random number generator (PRNG)." Randomized numbers in a computer application are traditionally spawned using PRNGs - which are typically fairly complex mathematical equations that produce a wide array of highly-varied (but evenly distributed) results over time. To work, however, they have to be "seeded," which means simply given a number to start with. So, for example, a PRNG seeded with the number "3" and asked to produce 4 random numbers would read in the numbers "3", "4", "5", and "6" and produce, for instance, "81", "288", "0", and "12," and will produce those same results EVERY TIME it starts with the number "3." Using a precise-to-millisecond (or even nanosecond) timestamp as the initial value fed to a PRNG is quite common, as it ensures the PRNG is nigh-impossible to "rig" and produces superbly and absolutely random results without affecting game performance (since one beauty of the algorithm is that it's not doing any extra work no matter how precise the timestamp is). Even better, this timestamp value is usually calculated, to the millisecond (or more), by the operating system... so it's not only possible to use it to seed the PRNG, it's the most convenient and best-performing option realistically available. --guest 19:41, 22 October 2010 (UTC)

Okay, so in that case it's nigh impossible to reproduce because milliseconds count and server lag even at best is about 30ms.... I think that note should be removed. I think the only way to test if this actually exists or not is to have the Anet Developers answer the question... --Lania 19:57, 22 October 2010 (UTC)

Quite difficult, but apparently not impossible. The first forum link shows two guys managing to do it not once, not twice, but three separate times. He mentions they repeatedly entered it until they were pretty sure they had it synchronized (using nearby spawns as a guide). With 30ms margin of error, repeated attempts, and coordinated effort, it does make sense that it's testable. It's not absolute proof, but I think coming up with a way to repeatedly cause drops (and chest spawn locations, spawn layout, etc) to synchronize is pretty strong evidence towards the theory. --guest 20:14, 22 October 2010 (UTC)

It's bogus --Manassas 10:36, 27 October 2010 (UTC)

"It's bogus" isn't much of a counterargument. Have you read the thread in question? --Irgendwer 16:53, 27 October 2010 (UTC)

Random numbers in video games are never random because a computer is unable to generate numbers independent of anything. To create a "random" number, it is derived from a non-random variable, usually time. Tool-assisted speedruns are a good example of this. Each frame that passes will cause a game to behave differently, and speedrunners make use of it by optimizing random enemy movement, random encounter intervals, loot generation and the like. GW is probably not that different. Of course, enemy behavior is not randomly generated but things like weapon damage, block chance, and drop generation may depend on the amount of frames pass. Unfortunately, with 30ms of lag it is nearly impossible to be certain. Koda Kumi 17:21, 27 October 2010 (UTC)

They could be random if ANet is using a hardware entropy generator. A geiger counter next to a few pieces of depleted uranium is the cheapest method, but you can have all sorts of fun with silly quantum effects if you've got the money to spend on hardware. I have heard rumors and seen videos of people receiving the exact same drop rates, so I'm thinking this is the result of one of any of the following facts:

1) The only confirmed occurences were bugs. This is perhaps the most likely explanation.

2) ANet, after discovering players could 'hack' the drop rate by entering simultaneously, updated the drop table seed (be it a RNG, PRNG or whatever else) to use an alternative method than time of entry, or altered the input to include more variables (milliseconds, microseconds, etc).

3) The drop table implementation was redesigned entirely, or has a new component which obscures data. If the original videos with identical drops were shown before the introduction of loot scaling, it is possible that loot scaling and the raw drop table are two separate subroutines, each quite possibly using different seeds. For instance: if the raw drop table is calculated based on the time of entry but the loot scaling is calculated by the time an enemy dies, it is reasonable to expect one party to get different drops than another.

Those are just my guesses. If you like, I can ask Joe to dig around for the drop tables in the source. –Jette 17:42, 27 October 2010 (UTC)

A long time ago

A long time ago, me and my bro went into urgoz and had 1 person suicide on the mobs, other pick up loot. this earned about 1k a run. however, after 6 or 7 runs, a message popped up saying mobs dont drop loot anymore

Any support for my theory?

I have developped this little theory of my own on a part of the anti-farm code.

I have a Pre-Searing char, where I used to farm all trophies at once, so I always had at least 25 of them when needed. Farming 25 of the trophies was no problem, it took an hour at most. Then the farming bored the hell out of me, so I just played that char once a day to collect the Gifts of the Huntsman. Then I ran out of trophies, and started farming the trophies that Nicholas collected that particular day, and noticed that the drop rate had significantly declined, and I hadn't farmed ANYTHING on my entire account for weeks. Now it took at least 2 hours to collect the 25 trophies needed for that day. Then the next day I started to farm "in advance" again, and the drop rate of the item needed earlier was restored.(I don't recall which item it was, but my best guess would be Skale Fins).

So my theory is: if a collector with a variable collectible item collects a certain trophy, that trophy's drop rate will decline.

Anecdotal. If you want to test it, find fifty people or so with a bunch of spare time and do some tests. Some players have hypothesized that drop rates decline when many players are farming an area, but there is no way to test this without very large sample groups. If that is true, then the drop rate for items asked for by either version of Nicholas would certainly decline. –Jette 13:52, 19 July 2011 (UTC)

The biggest issue with anecdotal evidence is that people remember confirming instances, but almost never pay attention to counter-evidence. When I tracked pre-Searing farming, I found no difference between advance- and as-needed-farming rates or total time. (I farmed perhaps an hour/day (if I logged on), building up stock in each item.)

If you want to provide enough evidence for your theory to get other people interested, you would need to do something like this:

• Farm the same trophy every day for an hour. Record the total amount collected. (Choose something other than Skale Fins, unless you avoid killing them as you run to Nick.)
• Wait two weeks (logon to the toon, but do not kill any of the foes that drop the trophy).
• Repeat at least 10 times.

That won't prove the theory, but it would provide at least a smidgen of evidence; it also provides a vague sense of just how much time it takes to find patterns in drop rates. Since the minimalist testing I've offered would take a single person 40 weeks, you might try to enlist a couple of friends to test with you. — Tennessee Ernie Ford (TEF) 16:39, 19 July 2011 (UTC)

There also appears to be some link to "tagged" accounts. I have 2 accounts that I've farmed with, both of which have a 55 Monk on them. Over 300 dead sword runs between the 2 showed marked differences. On the older account that I've farmed all over every campaign with, the drop rates were from 10% to 50% of the newer account. Both of the characters in question are identical in gear and skills, but on the account that I used to grind farms for many years the drops tend to be much poorer. Another anomaly of note on the dead sword farm, the dash to the ravines on the older account is always much easier with the wurm form taking a great deal less agro and damage getting to the portal. I found this curious, and had a guildie confirm this. He was amazed at how easily my character account always made the dash while his consistently got hammered. My two cents worth, but it seems probable that some of the "anti-farm" code that has been implemented is specifically account linked. ~~Ecks --The preceding unsigned comment was added by 68.2.241.211 (talk • contribs) at 07:18, 2 December 2011 (UTC).

Your using data from two accounts to draw conclusions about code that affects the entire game. It's easy for one account to be luckier over 150 runs than another (and since we don't have the raw data, it's hard to draw conclusions from "10 to 50%").

As a contrasting anecdote: the first 150 Royal Gifts I opened dropped 5 minis, but I haven't seen a mini drop from any gift (Royal, Traveler, or Strongbox) since; I also have friends who have never had a mini drop from gifts. If you look at the posted drop rates for GotT and RGs, you'll see people who appear to be very lucky and a lot more ppls without much luck.

So, while it's theoretically possible that accounts get tagged or that drop rates are reduced dynamically to counter extensive farming...no one has yet published anything close to sufficient evidence to support either hypothesis. For statistically useful data, we'd really need to see data tens of thousands of runs, e.g. hundreds of runs across hundreds of accounts. – Tennessee Ernie Ford (TEF) 17:31, 2 December 2011 (UTC)

Given the amount of time involved in tens of thousands of runs, it would be faster and more efficient to drag Joe into a dark room and beat him with a rubber hose until he talks. –Jette 02:21, 3 December 2011 (UTC)

'When possible, try to add leechers to a party.'

If leechers do indeed increase your drops, then why not just take 7 heros and flag them out of range? Saves a lot of time finding leechers. Random Weird Guy 11:51, 19 January 2012 (UTC)

I've been reading up on this a little, and was wondering the same thing. Anything not make this work?--Saera Serena 18:15, 4 August 2012 (UTC)

It should be easy enough to test. FYI you need at least 10-20 runs minimum, since the variation in the drops is enormous (i.e. it's really easy to have 4-5 bad runs in a row). Probably Vaettir farms are the easiest to use for testing. – Tennessee Ernie Ford (TEF) 18:40, 4 August 2012 (UTC)

Trying for a specific weapon

I wanted to get Wing's Axe for a FoW build, so I got my Rit out there and solo farmed Wing, Three Blade for about five hours. I got a reasonable amount of loot, but the axe just would not drop. Have I hit the anti-farm code, or do I just have terrible luck? Blue Totoro (talk) 14:18, 5 January 2014 (UTC)

idk if they still have the antifarm code, but if they do, you'll have hit it ;). how many kills is that over 5 hours? (only recent green farm I did was vera - and the boss dropped a green roughly once every 4 runs - believe same drop rate in nm and hm)-Chieftain Alex 15:30, 5 January 2014 (UTC)

I'm not sure, each run was kind of inconsistent. It was typically around 8-9 minutes or so, killing his group and one right next to his. It frequently would go 2 or 3 times as long when I died or when I was distracted. Blue Totoro (talk) 15:55, 5 January 2014 (UTC)

I got Wing's Axe the first time I killed his group today. If you want it (for free), pm me ingame or respond here. (my technique was poor, but I pulled wing into the area with all the canthan peasants + managed to spike down the rits with spirits + technobabble) -Chieftain Alex 16:29, 5 January 2014 (UTC)

Wow, that's amazing good luck. I'd like to buy it from you, but you're listed as offline right now. Blue Totoro (talk) 16:30, 5 January 2014 (UTC)

I hadn't logged in in a month, so I basically get the green guaranteed for the first boss I kill. (forgot that I had offline mode on ;) -Chieftain Alex 16:45, 5 January 2014 (UTC)

Items in inventory affect drop rates

Could there be a connection between the amount of certain items, say, Clovers, in the player's inventory and the amount of times that item drops from, say, Vaettirs? This would explain why people who repeatedly farm the same area get less and less drops over time. I had this idea and tested it a few times: I had roughly half a stack of Shamrock Ale in my inventory and was farming Vaettirs, every run yielded me only about 2-3 Ales for a few runs. Then I dropped my ales on the ground at the portal and repeated the farm, I had 5-8 Ales per clear. It could be just weird luck-based occurrence but doesn't sound completely impossible either. Needs more testing imo. Jeree95 (talk) 18:31, 19 March 2015 (UTC)

Just remove that section about "seeding"

There has been a lot of speculation about that theory, but I seriously question why anyone is spending time on it. Three reasons:

• firstly because it is 100% speculative (i.e. there is no actual indication that drops are generated at the time of entering the area instead of at the time of the actual monster kill);
• secondly because it's 100% unfalsifiable (i.e. you cannot prove or disprove this in any way, because even *if* we had a hypothesis of how the seeding works, it would still be impossible to test it accurately due to the fact that it's probably dependent on milliseconds);
• thirdly because it is 100% of no consequence (i.e. even if the theory is correct, there is nothing a player can do to avoid any disadvantages from it).

Can't we just limit this article to information that's actually relevant, and remove this part? -arnosluismans (talk) 10:10, 3 April 2019 (UTC)

Did you follow the link to see their results?

• It's more likely that they entered the zones at the same time than killed every monster at the same time.
• It's absolutely testable by repeatedly trying and seeing if you can generate the same loot drops in the first few mobs. It may be a 1/100 chance of your timing working out, but their methods can be repeated and shown to be false with a lot of effort.
• Finally, we document the game on this wiki, whether it has a meaningful consequence for players or not. Describing the underlying mechanics was/is a goal for some of our community members.

Fortunately for perspective, the section you're not happy with is sitting under the "disputed" section. G R E E N E R 17:40, 3 April 2019 (UTC)

As a slight aside: considering that death levelling was presumably unintended (I can't find a source for this offhand but think I remember reading it somewhere with respect to the old LDoA method) and that it doesn't affect drops at all, it could be considered evidence in support of the seeding theory - that is that drops were created for the original level of a creature, not the time it is killed. Loggy (talk) 04:21, 4 April 2019 (UTC)

Nah, could be any number of things. loot tables could be determined by the mob/level at spawn, or they could be manually set per-mob-per-zone. We simply don't have enough (any) information regarding how drops are determined. Thrain | contribs 03:10, 6 April 2019 (UTC)

I agree that a wiki should describe mechanics, even if they are of no direct consequence. However, as it stands, this point is not describing any mechanics; it only mentions a very broad possibility of a mechanic, whose source is speculation and whose tests have been inconclusive. Documenting each semi-interesting experiment that the player base has ever performed, is not what GWW should be used for. I therefore would like to reiterate my suggestion to remove this point.

If it needs to stay for one reason or another, let us at least move it to the "speculative" section. -arnosluismans (talk) 10:44, 16 March 2020 (UTC)