www.auroraglade.eu – Community Site for Aurora Glade!
Anti Farming Mechanic - Facst and figures.
Interesting experiment, thanks. I was wondering though if you should not have gone on for a little while longer, the fact that a certain type stopped dropping may have been the rng. Though for actual items not dropping for 30 minutes at the end of your experiment sure seems an indication that something has kicked in.
But good to have an indication of when to move on.
My experience just in both general playing and actively farming, has been to see a noticable loot drop off after roughly 30 mins. Although that could be due to playing a glass cannon thief and average killing a mob in less than 3-4 secs. However this timer seems to be as low as 15 mins or so on the defend DEs in Cursed Shore, if I am spamming explosive arrow to tag a many mobs as possible, (often 8-12 mobs per wave).
This has to be the worse data analysis ever. How can you say the rate of no drop almost doubled when in the first two ten kill segments there are 7 no drops, and the last two ten kill segments are six and seven no drops. Also you say after 70 kills it doubled, but according to your data set it actually more than halved at 70 kills.
In my opinion you never even got affected by the DR mechanic based on your very limited data set.
(edited by Jester.3265)
I agree with Jester. I’m not seeing any “dropping to nothing” or “doubling” or anything of the kind. Ever piece of data in chart seems to fall within expected deviations.
Won’t take any player data as fact. Only way to get facts is to get a official answer from Anet. What we need to know is if the % chance of, say a blue crafting material decreases after x amount of kills. Not how many blue bones you got after killing some mobs.
@Jester
Apologies for confusion, but its in how you read the numbers. The rate of No Drop compared to the rest of the drops changed considerably.
If we look at the sections that contain the two ’6’s.
The first time 6 No Drops scores, the rest of the items (Junk+Trophy+Item+Craft) add up to 9. For the second 6, the rest of the drops score 10. When 5 No Drops score, the rest of the drops add up to 10.
Then there is a huge spike in No Drops – 11, and the rest of the drops only score 3.
The spike continues with 9 No Drops and the rest combined only scoring 5.
The start of the chart looks a little odd at first with 9 and 8 No Drops, but its not that alarming when we compare that number to the other items that drop 4 items and 7 items.
One aspect I didnt take into account was ITEM VALUE in terms of selling to a vendor. That may have effected the drop rates and distorted the start (or even the rest of the results), either way we can see a threshold was reached.
www.auroraglade.eu – Community Site for Aurora Glade!
I do not understand your numbers, according to you interpretation the loot rate (no drops) is appr. halved after 50 minutes, yet reaches the highest value after 70 again? Also. the numbers do not seem to be consistent between the categories. Seems more like normal statistical variation in a too small sample to me – which I suspect anyway because 103 repetitions is really low for such an undertaking, not to speak of methodological problems (eg. arbitrarily set time sets).
@Jester
Sure, the statistical sample isn’t large enough to have conclusive proof of anything.
However, the OP has provided far more meaningful and useful data than you have. Gain some perspective and get off your soap box.
In other words, if we look at it as %, every kill either scoring for ‘No Drop’ or for ‘Drop’
The chart would look like this
70/30 – ‘No drop’ is more than double the ‘drop’. Maybe it gave me a valuable item?
53/47 – The rate is balancing out
40/60 – Now ‘drop’ is higher than ‘no drop’. I am finding more items, more times
37/63 – Almost the same as last
33/67 – Again not much change
79/21 – The ‘BAM’, ‘No Drop’ more than doubles from previous results
64/36 – ‘No Drop’ remains higher than ‘Drop’
When looked at it from this angle you should be able to see that it appears a threshold was reached and anti farming kicking in, considering each section is 10 minutes of game time (or average 15 kills), thats 20 minutes or 30 kills where No Drops vastly out wieghed the items I recieved.
www.auroraglade.eu – Community Site for Aurora Glade!
It is clear that no drop does indeed double past a certain kill count, not time.
Followed by worse quality (vendor trash) while crafting mats stay the same.
This leaves a conclusion of, a) logout for an hour to reset your DR or b) continue farming and getting trash.
Since I love farming I myself farm through the Anti farm code and stock pile the heavy moldy bags which are worth nice money.
Of course from the very limited sample you cant get a precise figure, it could be 10% off the mark, it could be 30% of the mark. But there are the numbers from the first sample.
If anyone wants to repeat the experiment then we will have 2 sets of data and a better picture.
I said it before in my last post, I play on Aurora Glade server, anyone is welcome to hit me up and we will work together
www.auroraglade.eu – Community Site for Aurora Glade!
Didn’t they say that the anti-farm code only kicks in after an hour (or 25 minutes if you try hard enough)?
How would the game know? I could just stand in 1 spot and not kill 1 mob, then when I do decide to kill a mob I get nothing? it’s based off a kill count since that’s easy to program.
Ironangel is probably right, number of kills would probably be the trigger.
I have gone back over my chart and looked at patterns for ‘No Drop’.
From kills 1 to 76, the longest run of ‘No Dropss’ was 4 (which appears 4 times)
From kill 77, the longest run is 7
Up until kill 70, the chart doesnt look too bad with the rate of No Drops.
There is a good chance somewhere around 70 is the magic number. Factor in that during DE’s you get a lot of kills really quickly, this could be why we see reports of 30 minutes then the anti farm kicking in
www.auroraglade.eu – Community Site for Aurora Glade!
Xericor you seem to be totally ignoring the fact that you started out with a higher percentage of no drops, then the drop rate improved, then it got worse again. Following your logic, this would mean you started out with the DR mechanic in effect, then it stopped, then it started again. This is why you need a WAY bigger sample size. Overall, you get a 54% rate of no drops. But you also need to consider that some of the times you got drops, you got more than one drop.
RNG is random.
For example, last night I was farming skelks in a high level zone. There would be times when I’d get absolutely no loot at all for 6, 7, or more mobs in a row, and other times I’d get potent blood 3 times in a row (some times AFTER the streaks of zero drops).
You need a really big sample size to account for randomness. In other words, probably at least 30 sessions worth of farming longer than the anti-farm code is rumored to exist.
antifarming code is worse than bots
Has it been considered that a combination of # of enemies killed and the time lapse between the kills is the trigger, not simply time killing or # of kills? My understanding is that the code is similar to the one used in GW1, which could track how quickly a series of enemies died in succession to each other, with the intention of providing the DR’s to builds that rely heavily on AoE farming techniques. It would be interesting to see a test similar to what the OP did using both single kill techniques and AoE techniques to see how time was effected by the code.
My experience just in both general playing and actively farming, has been to see a noticable loot drop off after roughly 30 mins. Although that could be due to playing a glass cannon thief and average killing a mob in less than 3-4 secs. However this timer seems to be as low as 15 mins or so on the defend DEs in Cursed Shore, if I am spamming explosive arrow to tag a many mobs as possible, (often 8-12 mobs per wave).
same thing here. I’ve made a similar post pointing out how bad the game punishes you for killing a lot of things for an hour (mainly I pointed our how bad the drops and droprates are after half an hour in Orr (WHILE MOVING AROUND THE MAP). And yes I am a thief too, and I did notice the 30 minute “timer” on the code.
What we need to know is why is a dumb hidden anti-game mechanic implemented in a game advertised as free and diverse and open world and not limited by anything? This is the most ridiculous crap I’ve ever heard
Good to see some actual data. Sad that it’s necessary though, this is the kind of thing the game makers should share with us.
With a sample size as low as this (admittedly a lot of work even as is, but still low statistically), I’m betting that those results are well within error ranges and not statistically significant. As you can see you do have 9 “No Drop” in your first 10 minutes of farming, so seeing an 11 and 9 at the end really isn’t an outlier. There are also plenty of 1 “Item Drops” spread throughout, so seeing a couple of zeros at the end aren’t all that surprising either. Your highest crafting mat drop rates were towards the middle of the experiment as well.
This could be indicative of a decreased drop rate on certain items after a set amount of time or kills, but the data simply isn’t conclusive.
First, I would like to commend Xericor for taking the time to do something as tedious as kill the same few enemies repeatedly and record the data from it, and thank him for putting himself out there with an analysis that would undoubtedly get him a good deal of flak. While the data set is too small for definitive answers, It’s still very interesting.
There are some odd things happening, such as the increased rate of loot dropping in the middle (significantly higher than the overall rate). I would really like to look closer at that, but it seems accompanied by slightly higher kill rates (assuming the time numbers are always 1 item/kill, it could just be multiple drops per enemy throwing the numbers off), which actually runs counter to what most players assumed in GW1 about the kill rate. I would say it’s just the randomness if it didn’t seem to span half an hour and 46 kills.
Individual types is much less definitive because the numbers are so low for them. Trophies is in the single digits altogether, and tend to be more or less just randomly spread throughout the data, so we can’t even say for sure if it changes or not. Junk doesn’t seem to have much of a pattern either even though it is definitely more plentiful. Items on the other hand, have a pretty clear drop off for the last half hour. If you compare the first 50 enemies to the last 50, it’s 8 vs 3 and no items drop for the last 30 enemies. Again, small numbers, but it looks like something is happening.
Based on these numbers, I would hypothesize that kill rate actually is the biggest determinant over whether or not you get loot, but the rate of items will actually drop after killing enemies in an area enough times. I would also like to conjecture that we aren’t dealing with a single loot table, but with actually multiple. Each major type could be randomly selected separately (low probabilities in each table). This would make probability modifiers like anti-farming, magic find, etc. more manageable since improving the odds of one thing won’t push out another.
Thanks again for the data, I’d love to see more like this. If anybody plans on doing something like this again, I suggest a few things. First, take a sum total of the merchant value of all the goods on an enemy including coins. This can be used to avoid double counting issues and makes it easy to check for an overall change. Take the data down to individual enemies, and record time down to the minute or half a minute (This is known as the lowest level of granularity). This can make it easier to group the data in different ways. After that, record each type of item dropped separately, but associated with the specific enemy. And when you give it to whoever you want to analyze it, don’t summarize it ahead of time, it tends to make it harder to work with.
Also, keep it in excel, csv, or some other standard format. I would be happy to throw some data into the fancy software I play with for a living and post charts and things all over the forums but I’m not manually entering the data unless I am gathering it myself, which probably won’t happen because I tend not to have the time. But please, feel free to post more data like this.
Fort Aspenwood – [fury], [SAO], [NICE]
Fun on someone else’s schedule is not fun
Fun on someone else’s schedule is not fun