Guild Wars 2

THIS IS NOT A THREAD TO POST ABOUT RNG BEING UNFAIR OR FINE. THIS IS STRICTLY A POST ABOUT HOW TO CONDUCT RESEARCH AND GATHER REAL FACTS BEHIND RNG. SAYING THE DEVS SAID ITS FINE IS NOT A REAL REPITIABLE FACT. WE DO NOT HAVE EVIDENCE OF WHAT HAPPENS BEHIND CLOSED DOORS.

Please visit: https://forum-en.gw2archive.eu/forum/game/gw2/2-accounts-2-different-loot-experiences/page/4#post5030396 to continue discussing Views on RNG.

Everyone, including me, says that RNG is unfair. However neither side can provide any real proof that their side is correct (RNG is flawed/RNG is fine)

What we should do is conduct a small experiment. For example 5-10 people will record their experiences fighting the same world boss. Will do this for a week everyday at the same time. And just do a small comparison between accounts. Of coarse we would need the participation of 5 “young” accounts and 5 “old” ones. All data will need to be recorded and time stamped. If anyone’s down to try this, I suggest you get a team together to help prove or disprove RNGs flaws.

Recorded and taking a small dip of info from this can shine some light on what direction this is truly going. Will it conclude anything? No. But it IS a start in finding out what’s going on. This could lead to more research in the future.

Rules:
1: record video proof
2: if the game is patched, the experiment must be repeated/restarted.
3: once experiment starts, subjects cannot altar their MF at all. This means not using essence of luck/food/boosters. If done, the subject voids results.
4: all subjects must do the same world boss while recording.
5: at the end of the week, we shall compare results. If all subjects get around the same loot, the experiment can conclude or continue to a valid point of either repeating or ending.

This is just my idea on what to do. It involves a bit of work, but that’s what research is. Work. So if we wanna prove, or disprove something, work needs to be done. I would love to be a part of this and wouldn’t mind doing this however I don’t have the time at this very moment. (School finals) however sometime in the future I could do so if anyone’s interested in helping.

(Come forth my little asura friends!!)

What do you guys think we could do to help get some stats?

Crimson Clouds.4853’s idea

Rhyse.8179’s idea

You need more rules because of DR. And since we don’t know how it works exactly this will make your results less valid but we do know some ways to prevent it. The character doing this should always be the same and played the same amount of time in game and zone and do the same amount of events outside of world bosses or loot gathering. Furthermore, to make sure DR is reset I would propose this character has been sitting for at least one week without use. I wouldn’t have these characters salvage or buy anything either from like a vendor. Only sell or save items you obtain during this time.

Really? Just 5 or 10 accounts will give reputable evidence? It does not seem likely, to me, but, of course, I am no expert.

Andraus.3874:

You need more rules because of DR. And since we don’t know how it works exactly this will make your results less valid but we do know some ways to prevent it. The character doing this should always be the same and played the same amount of time in game and zone and do the same amount of events outside of world bosses or loot gathering. Furthermore, to make sure DR is reset I would propose this character has been sitting for at least one week without use.

I like it. We could do world bosses, and possibly have another set of players in another experiment do mobs in a certain zone for like an hour, log off. And repeat.

I also agree, characters should be parked for a week prior of experiment.

Inculpatus cedo.9234:

Really? Just 5 or 10 accounts will give reputable evidence? It does not seem likely, to me, but, of course, I am no expert.

If we get more that would be even better, but managing 5-10 accounts at a time is far easier then say 30.

TheFool.4589:

Inculpatus cedo.9234:

Really? Just 5 or 10 accounts will give reputable evidence? It does not seem likely, to me, but, of course, I am no expert.

If we get more that would be even better, but managing 5-10 accounts at a time is far easier then say 30.

Except that’s not enough to provide reliable results.

Yes and no. The whole point is to say " ok 10 accounts reported this, we should expand the research to xxx accounts now because of the results."

Indefinitely, the results would help open the door to expanding the research. This is not conclusive research rather a gateway. Say all 10 accounts found that RNG is bugged, now that doors open to expand it, cause yes 10 out of Xxxxxx accounts is not a valid/realiable result. BUT there is some reason to expand the experiment to see if the results are truly flawed or by coincidence.

I commend your idea to research this; however, the drop rates on many items are in the fractions of a percent. Therefore, to get the appropriate resolution, you will need several orders of magnitude more data to draw any sort of respectable conclusions. It is just not feasible to do with a handful of data points.

I suggest taking a course in statistics before trying to set up experiments, because it would let you know that this has been framed extremely poorly and any conclusions from it will be gained incorrectly.

So, if the 10 accounts found that they all received nearly the same loot? Would this be enough to abandon the experiment? And what, exactly, determines whether loot is within range, or not? If someone gets a rare item, does that automatically mean that RNG is ‘borked’? If no one does, is RNG not ‘borked’? How often should a rare item drop? Do you know the absolute drop rates, and what factors apply to drop rates, if any?

I’m just not sure something like this would definitively (or even remotely) answer any questions.

Still…good luck (if applicable in such a case).

I’m going to try to nitpick in order to make this experiment as foolproof as possible- I suggest others do this too. It’s better to find flaws in the experiment before it has been conducted rather than after (and find all of your data fall through the holes).

-Subjects should avoid the area of the worldboss before recording. Even better, all subjects should join the world boss map at the same time. This is to reduce the chance of DR affecting one player and not another.

-All subjects should achieve gold reward from the world boss event. If they do not, their results are discredited for that boss (considering event reward level scales reward?)

-All subjects should have a clean inventory or a designated bag for loot they obtain.

-Would game chat be a good addition to recording data? This means that it is more obvious which loot item dropped from which mob.

-The sample size needs to be larger. 5-10 people over 7 days only means 35-70 attempts at data. More people need to be involved and over a longer period of time. Alternatively, start off small as a pilot experiment to see if you can find any problems with the experiment.

-Patches happen every week (usually on a Tuesday). Ceasing the experiment just because of this would be crazy. I think the experiment should be conducted over a few weeks- it would be interesting to see if patching even changes anything.

-Controlling magic-find over a long period of time would be very difficult. I would suggest that people refrain from using MF boosters (food, banners, guild buffs, asc. upgrades), but record their magic find at the time of each world boss.

-Find a statistician. Somebody who has experience with numbers, at least. In a perfect world, number each participant so that the statistician/“experimenter” isn’t aware of which data comes from which participant. (To reduce experimenter bias).

Crimson Clouds.4853:

I’m going to try to nitpick in order to make this experiment as foolproof as possible- I suggest others do this too. It’s better to find flaws in the experiment before it has been conducted rather than after (and find all of your data fall through the holes).

-Subjects should avoid the area of the worldboss before recording. Even better, all subjects should join the world boss map at the same time. This is to reduce the chance of DR affecting one player and not another.

-All subjects should achieve gold reward from the world boss event. If they do not, their results are discredited for that boss (considering event reward level scales reward?)

-All subjects should have a clean inventory or a designated bag for loot they obtain.

-Would game chat be a good addition to recording data? This means that it is more obvious which loot item dropped from which mob.

-The sample size needs to be larger. 5-10 people over 7 days only means 35-70 attempts at data. More people need to be involved and over a longer period of time. Alternatively, start off small as a pilot experiment to see if you can find any problems with the experiment.

-Patches happen every week (usually on a Tuesday). Ceasing the experiment just because of this would be crazy. I think the experiment should be conducted over a few weeks- it would be interesting to see if patching even changes anything.

-Controlling magic-find over a long period of time would be very difficult. I would suggest that people refrain from using MF boosters (food, banners, guild buffs, asc. upgrades), but record their magic find at the time of each world boss.

-Find a statistician. Somebody who has experience with numbers, at least. In a perfect world, number each participant so that the statistician/“experimenter” isn’t aware of which data comes from which participant. (To reduce experimenter bias).

Do you mind if I put this in the original post?

Not at all I’m pleased I could be of some help.

TheFool.4589:

Yes and no. The whole point is to say " ok 10 accounts reported this, we should expand the research to xxx accounts now because of the results."

Indefinitely, the results would help open the door to expanding the research. This is not conclusive research rather a gateway. Say all 10 accounts found that RNG is bugged, now that doors open to expand it, cause yes 10 out of Xxxxxx accounts is not a valid/realiable result. BUT there is some reason to expand the experiment to see if the results are truly flawed or by coincidence.

Not yes and no. Just yes, it’s not enough. You cannot get a conclusion from 10 accounts. I mean you can think you get a conclusion, but statistically, no, you cannot get a conclusion. If the purpose of this experiment is to show anet there’s something wrong, the second they see “10 accounts only” they are going to dismiss your claim (rightfully so).

It might be worth trying to set up a guild for this experiment. Communication between participants would be easier along with organisation and controlling guild MF buffs.

Sorin.4310:

TheFool.4589:

Yes and no. The whole point is to say " ok 10 accounts reported this, we should expand the research to xxx accounts now because of the results."

Indefinitely, the results would help open the door to expanding the research. This is not conclusive research rather a gateway. Say all 10 accounts found that RNG is bugged, now that doors open to expand it, cause yes 10 out of Xxxxxx accounts is not a valid/realiable result. BUT there is some reason to expand the experiment to see if the results are truly flawed or by coincidence.

Not yes and no. Just yes, it’s not enough. You cannot get a conclusion from 10 accounts. I mean you can think you get a conclusion, but statistically, no, you cannot get a conclusion. If the purpose of this experiment is to show anet there’s something wrong, the second they see “10 accounts only” they are going to dismiss your claim (rightfully so).

Indefinitely, the results would help open the door to expanding the research. This is not conclusive research rather a gateway. Say all 10 accounts found that RNG is bugged, now that doors open to expand it, cause yes 10 out of Xxxxxx accounts is not a valid/realiable result.BUT there is some reason to expand the experiment to see if the results are truly flawed or by coincidence.

Yes I can read.

This: “Say all 10 accounts found that RNG is bugged”

You cannot make this conclusion because your sample size is too small. Period.

Crimson Clouds.4853:

It might be worth trying to set up a guild for this experiment. Communication between participants would be easier along with organisation and controlling guild MF buffs.

I wouldn’t mind setting one up.
Name:The subjects
Tag: RNG
Haha

You would need 100’s or even better yet 1000’s of accounts to even come close to achieving results that could be considered statistically significant.

Sorin.4310:

Yes I can read.

This: “Say all 10 accounts found that RNG is bugged”

You cannot make this conclusion because your sample size is too small. Period.

It’s only 10 for right now. Just to see. A month from now we can do 500 accounts all we want. There’s nothing wrong with starting small and expanding out. We can even do 10 at a time and have 50 a week.

It’s about time someone on the home forums demanded a proper test. Until someone does some kind of statistically rigorous experiment, or until Anet releases their source code, all the chatter and whining is meaningless. We need an experiment under controlled circumstances with a sample size of no less then 200 and ideally greater then 1000.

Suggested experiment formats:

Experiment 1: Gather 100 champion chests of the same type (say, gilded coffers). Open them all with the same character. Record drops. Repeat on 9 other other accounts, for a total sample of 1000 chests. 1 sample = 1 chest.

Experiment 2: Do an identical World Boss cycle for an identical period of time. Say, 1 week, 5 different bosses per day (NOT 5 kills of the same boss, to prevent diminshing returns), the same 5 bosses each day. That’s 35 samples per account, so we will need at least 10 accounts (350 samples) and ideally 28+ accounts (1000 samples). 1 sample = 1 WB kill.

Experiment 3: Kill 1000 level 80 veteran mobs, in a level 80 zone (eg, southsun, cursed shore), with a level 80 character. Record body drop loot, exclude chests and coffers. Repeat on 9 other accounts. Because MF doesn’t affect chests, this is the only way to test MF. 1 sample = 1 kill.
—Because the drop rate is so low, it’s unlikely that 100 veteran kills (1000 samples over 10 accounts) would give a meaningfully large result. So I bumped this one up by an order of magnitude to a total sample of 10,000 kills.

In all cases: Use multiple accounts with different amounts of MF and of different ages. Rigorously document all your samples and the results of each one. Post the data. For bonus nerd points, make a spreadsheet/graph and post that!

Rhyse.8179:

It’s about time someone on the home forums demanded a proper test. Until someone does some kind of statistically rigorous experiment, or until Anet releases their source code, all the chatter and whining is meaningless. We need an experiment under controlled circumstances with a sample size of no less then 200 and ideally greater then 1000.

Suggested experiment formats:

Experiment 1: Gather 100 champion chests of the same type (say, gilded coffers). Open them all with the same character. Record drops. Repeat on 9 other other accounts, for a total sample of 1000 chests. 1 sample = 1 chest.

Experiment 2: Do an identical World Boss cycle for an identical period of time. Say, 1 week, 5 different bosses per day (NOT 5 kills of the same boss, to prevent diminshing returns), the same 5 bosses each day. That’s 35 samples per account, so we will need at least 10 accounts (350 samples) and ideally 28+ accounts (1000 samples). 1 sample = 1 WB kill.

Experiment 3: Kill 1000 level 80 veteran mobs, in a level 80 zone (eg, southsun, cursed shore), with a level 80 character. Record body drop loot, exclude chests and coffers. Repeat on 9 other accounts. Because MF doesn’t affect chests, this is the only way to test MF. 1 sample = 1 kill.
—Because the drop rate is so low, it’s unlikely that 100 veteran kills (1000 samples over 10 accounts) would give a meaningfully large result. So I bumped this one up by an order of magnitude to a total sample of 10,000 kills.

In all cases: Use multiple accounts with different amounts of MF and of different ages. Rigorously document all your samples and the results of each one. Post the data. For bonus nerd points, make a spreadsheet/graph and post that!

May I put this in the original post?

Why should players invest this much effort? The burden of proof that the RNG is working properly and fair is on Anet side.
They have all the data of every player because they keep very detailed logs of everything.

TheFool.4589:

Rhyse.8179:

It’s about time someone on the home forums demanded a proper test. Until someone does some kind of statistically rigorous experiment, or until Anet releases their source code, all the chatter and whining is meaningless. We need an experiment under controlled circumstances with a sample size of no less then 200 and ideally greater then 1000.

Suggested experiment formats:

Experiment 1: Gather 100 champion chests of the same type (say, gilded coffers). Open them all with the same character. Record drops. Repeat on 9 other other accounts, for a total sample of 1000 chests. 1 sample = 1 chest.

Experiment 2: Do an identical World Boss cycle for an identical period of time. Say, 1 week, 5 different bosses per day (NOT 5 kills of the same boss, to prevent diminshing returns), the same 5 bosses each day. That’s 35 samples per account, so we will need at least 10 accounts (350 samples) and ideally 28+ accounts (1000 samples). 1 sample = 1 WB kill.

Experiment 3: Kill 1000 level 80 veteran mobs, in a level 80 zone (eg, southsun, cursed shore), with a level 80 character. Record body drop loot, exclude chests and coffers. Repeat on 9 other accounts. Because MF doesn’t affect chests, this is the only way to test MF. 1 sample = 1 kill.
—Because the drop rate is so low, it’s unlikely that 100 veteran kills (1000 samples over 10 accounts) would give a meaningfully large result. So I bumped this one up by an order of magnitude to a total sample of 10,000 kills.

In all cases: Use multiple accounts with different amounts of MF and of different ages. Rigorously document all your samples and the results of each one. Post the data. For bonus nerd points, make a spreadsheet/graph and post that!

May I put this in the original post?

Of course. You may want to summarize and link or something so it doesn’t get too bloated. Do whatever you like with my post.

Malediktus.9250:

Why should players invest this much effort? The burden of proof that the RNG is working properly and fair is on Anet side.
They have all the data of every player because they keep very detailed logs of everything.

This is a “put up or shut up” post directed at the whiners. Either put in the effort to get the data, or stop complaining. Asking Anet for the data is useless cause game companies are annoying and keep that kind of thing internal.

It may also attract the occasional science geek, like myself, who like graphs. Mmmmmm. Graphs.

I wish we had the tools Anet has. We could use the same account tagging they use for checking for bots. We could mark similar accounts and sit back and collect the data. Then compare said data to see if there are any anomalies.

Unfortunately it is nowhere near as easy for us.

Rhyse.8179:

Malediktus.9250:

Why should players invest this much effort? The burden of proof that the RNG is working properly and fair is on Anet side.
They have all the data of every player because they keep very detailed logs of everything.

This is a “put up or shut up” post directed at the whiners. Either put in the effort to get the data, or stop complaining. Asking Anet for the data is useless cause game companies are annoying and keep that kind of thing internal.

It may also attract the occasional science geek, like myself, who like graphs. Mmmmmm. Graphs.

Sorry but any data collected from what 10-30-50 accounts out of the 100’s of thousands or more GW2 active accounts isn’t going to resolve or prove anything. It would be laughed out of the scientific community if the results were attempted to be published.

JustTrogdor.7892:

Rhyse.8179:

Malediktus.9250:

Why should players invest this much effort? The burden of proof that the RNG is working properly and fair is on Anet side.
They have all the data of every player because they keep very detailed logs of everything.

This is a “put up or shut up” post directed at the whiners. Either put in the effort to get the data, or stop complaining. Asking Anet for the data is useless cause game companies are annoying and keep that kind of thing internal.

It may also attract the occasional science geek, like myself, who like graphs. Mmmmmm. Graphs.

Sorry but any data collected from what 10-30-50 accounts isn’t going to resolve or prove anything. It would be laughed out of the scientific community if it were published.

Small taste of the pie mate, it isn’t a conclusive experiment. This is a experiment to see if there should be an experiment.

“Sticking our tongue on the tip of the pie to see if it taste good enough to eat”

TheFool.4589:

Sorin.4310:

Yes I can read.

This: “Say all 10 accounts found that RNG is bugged”

You cannot make this conclusion because your sample size is too small. Period.

It’s only 10 for right now. Just to see. A month from now we can do 500 accounts all we want. There’s nothing wrong with starting small and expanding out. We can even do 10 at a time and have 50 a week.

I mean the 10 won’t let you see anything but for the sake of discussion, what results would show you the system is bugged? Not generally speaking (ie “one person got less drops”). What specifically. Like numberwise, dropwise. What data would say the system is broken?

Ayrilana.1396:

TheFool.4589:

Inculpatus cedo.9234:

Really? Just 5 or 10 accounts will give reputable evidence? It does not seem likely, to me, but, of course, I am no expert.

If we get more that would be even better, but managing 5-10 accounts at a time is far easier then say 30.

Except that’s not enough to provide reliable results.

How many accounts were involved in the data that showed there was a problem with ecto salvaging? I can remember one person posting his data, but I don’t know if anyone else did.

This isn’t about proving anything, its about providing enough information to back up the claims and convince the devs to investigate because they have far more data to work with but need a reason to look at it.

The system would be “bugged” if there was a significant difference between a normal and “problem” accounts. In science, I believe a “significant difference” is a difference of 5% or more.

The problem here is, if the experiment was conducted and data appeared to be perfectly normal and reasonable, I wonder how many people would discredit the data purely because they have experienced otherwise. (People currently discredit A-net’s word).

Sorin.4310:

TheFool.4589:

Sorin.4310:

Yes I can read.

This: “Say all 10 accounts found that RNG is bugged”

You cannot make this conclusion because your sample size is too small. Period.

It’s only 10 for right now. Just to see. A month from now we can do 500 accounts all we want. There’s nothing wrong with starting small and expanding out. We can even do 10 at a time and have 50 a week.

I mean the 10 won’t let you see anything but for the sake of discussion, what results would show you the system is bugged? Not generally speaking (ie “one person got less drops”). What specifically. Like numberwise, dropwise. What data would say the system is broken?

For starters, will record rare and up drops. Then in another week will only record exotics. If 8/10 accounts get nothing but blues, while 2/10 get exotics/rates every time, there’s a questionable call.

If 8/10 accounts report at least on exotic/rare every time, then that shows that it’s fine, it’s all luck. This experiment is really about whether or not some accounts are luckier then others. If those 2/10 accounts keep getting exotics say 5/7 days of the week while the other 8 keep getting blues/greens and 0-1 exotic out of that week, there’s a valid reason to investigate.

These of coarse are not real numbers, I just used these for example purposes…

Crimson Clouds.4853:

The system would be “bugged” if there was a significant difference between a normal and “problem” accounts. In science, I believe a “significant difference” is a difference of 5% or more.

The problem here is, if the experiment was conducted and data appeared to be perfectly normal and reasonable, I wonder how many people would discredit the data purely because they have experienced otherwise. (People currently discredit A-net’s word).

The only opinion that matters belongs to John Smith. I don’t know if the thread from the Black Lion forum survived the Great Forum Massacre a while back, but he once said something to the effect that you don’t have to prove there is a problem beyond a shadow of a doubt, just provide a real hypothesis for him to test and he will investigate.

I’m afraid that any experiment like this conducted by players will never give conclusive results. As Rhyse said, we don’t have access to the source code – and it’s VERY unlikely that drop rates as a whole are limited to a single percentage check. For all we know the calculations go like this..

Damage threshold met? TRUE
Phase of the moon? WAXING GIBBOUS
-> Number of Hylek defeated today is prime? TRUE
-> -> Probability of an exotic warhorn drop +0.1%

One could say that it’s not worthwhile for the devs to implement such a convoluted scheme for calculating droprates, but if I were a particularly sadistic programmer, it would absolutely be so.

tl;dr
It’s probably too complex for us to figure out without their secret formula.
They won’t tell us about the secret formula because we’d immediately try to find a way to exploit it.

Crimson Clouds.4853:

The system would be “bugged” if there was a significant difference between a normal and “problem” accounts. In science, I believe a “significant difference” is a difference of 5% or more.

The problem here is, if the experiment was conducted and data appeared to be perfectly normal and reasonable, I wonder how many people would discredit the data purely because they have experienced otherwise. (People currently discredit A-net’s word).

5% of what? # of drops? That is literally inherent to a RNG system. It’s expected. I’m pretty sure your problem with the system is you don’t like it, not that it’s broken.

tolunart.2095:

Ayrilana.1396:

TheFool.4589:

Inculpatus cedo.9234:

Really? Just 5 or 10 accounts will give reputable evidence? It does not seem likely, to me, but, of course, I am no expert.

If we get more that would be even better, but managing 5-10 accounts at a time is far easier then say 30.

Except that’s not enough to provide reliable results.

How many accounts were involved in the data that showed there was a problem with ecto salvaging? I can remember one person posting his data, but I don’t know if anyone else did.

This isn’t about proving anything, its about providing enough information to back up the claims and convince the devs to investigate because they have far more data to work with but need a reason to look at it.

It was the salvage rate and had nothing to do with accounts. That poster just needed a large enough sample size of salvage results to present their case to open the discussion. Very much different than what people are trying to accomplish here as people are arguing that there’s a difference between accounts. This requires a large number accounts as well as a large number of attempts due to the various drop rates which tend to be rather low.

Sorin.4310:

Crimson Clouds.4853:

Snip.

5% of what? # of drops? That is literally inherent to a RNG system. It’s expected. I’m pretty sure your problem with the system is you don’t like it, not that it’s broken.

There’s no point in making personal comments. I’m interested in experiments, sure. I don’t think the system is broken though and never said it was- I’ve always maintained that disparity as described through annecdotal evidence is due to cognitive bias and/or is well within the realms of RNG.

As for the 5% thing, I think it would mean that player X (under the same circumstances/variables as player Y) would receive a consistent average of 5%+ exotics.

TheFool.4589:

Sorin.4310:

TheFool.4589:

Sorin.4310:

Yes I can read.

This: “Say all 10 accounts found that RNG is bugged”

You cannot make this conclusion because your sample size is too small. Period.

It’s only 10 for right now. Just to see. A month from now we can do 500 accounts all we want. There’s nothing wrong with starting small and expanding out. We can even do 10 at a time and have 50 a week.

I mean the 10 won’t let you see anything but for the sake of discussion, what results would show you the system is bugged? Not generally speaking (ie “one person got less drops”). What specifically. Like numberwise, dropwise. What data would say the system is broken?

For starters, will record rare and up drops. Then in another week will only record exotics. If 8/10 accounts get nothing but blues, while 2/10 get exotics/rates every time, there’s a questionable call.

If 8/10 accounts report at least on exotic/rare every time, then that shows that it’s fine, it’s all luck. This experiment is really about whether or not some accounts are luckier then others. If those 2/10 accounts keep getting exotics say 5/7 days of the week while the other 8 keep getting blues/greens and 0-1 exotic out of that week, there’s a valid reason to investigate.

These of coarse are not real numbers, I just used these for example purposes…

An RNG system, literally, allows this kind of result while working. Your problem is with the design. Not that it’s broken.

Crimson Clouds.4853:

Sorin.4310:

Crimson Clouds.4853:

Snip.

5% of what? # of drops? That is literally inherent to a RNG system. It’s expected. I’m pretty sure your problem with the system is you don’t like it, not that it’s broken.

There’s no point in making personal comments. I’m interested in experiments, sure. I don’t think the system is broken though and never said it was- I’ve always maintained that disparity as described through annecdotal evidence is due to cognitive bias and/or is well within the realms of RNG.

As for the 5% thing, I think it would mean that player X (under the same circumstances/variables as player Y) would receive a consistent average of 5%+ exotics.

The system they designed allows this. It allows 500%+

Sorin.4310:

TheFool.4589:

Sorin.4310:

TheFool.4589:

Sorin.4310:

Yes I can read.

This: “Say all 10 accounts found that RNG is bugged”

You cannot make this conclusion because your sample size is too small. Period.

It’s only 10 for right now. Just to see. A month from now we can do 500 accounts all we want. There’s nothing wrong with starting small and expanding out. We can even do 10 at a time and have 50 a week.

I mean the 10 won’t let you see anything but for the sake of discussion, what results would show you the system is bugged? Not generally speaking (ie “one person got less drops”). What specifically. Like numberwise, dropwise. What data would say the system is broken?

For starters, will record rare and up drops. Then in another week will only record exotics. If 8/10 accounts get nothing but blues, while 2/10 get exotics/rates every time, there’s a questionable call.

If 8/10 accounts report at least on exotic/rare every time, then that shows that it’s fine, it’s all luck. This experiment is really about whether or not some accounts are luckier then others. If those 2/10 accounts keep getting exotics say 5/7 days of the week while the other 8 keep getting blues/greens and 0-1 exotic out of that week, there’s a valid reason to investigate.

These of coarse are not real numbers, I just used these for example purposes…

An RNG system, literally, allows this kind of result while working. Your problem is with the design. Not that it’s broken.

Again, the whole idea is to prove that RNG needs to be investigated further after small experiments. This is going to help show some stats for both, in general, those who think RNG is fine/broken. This is a unbiased experiment that’s is NOT TO CONCLUDE OR MAKE FINAL RULINGS If the scale leans one way over the other though, there’s call for it to be investigated further.

So your criteria for the system is broken is a possible outcome of the designed system, and this will show that further investigation needs to be done?

I give up. I now understand why the devs stopped responding to RNG broken threads. They realized long ago that the players have no idea what they are doing.

Sorin.4310:

Crimson Clouds.4853:

Sorin.4310:

Crimson Clouds.4853:

Snip.

5% of what? # of drops? That is literally inherent to a RNG system. It’s expected. I’m pretty sure your problem with the system is you don’t like it, not that it’s broken.

There’s no point in making personal comments. I’m interested in experiments, sure. I don’t think the system is broken though and never said it was- I’ve always maintained that disparity as described through annecdotal evidence is due to cognitive bias and/or is well within the realms of RNG.

As for the 5% thing, I think it would mean that player X (under the same circumstances/variables as player Y) would receive a consistent average of 5%+ exotics.

The system they designed allows this. It allows 500%+

Like I said, under the same circumstances/variables.

You could surely measure disparity even with the effects of magic find, anyway.

Simplistically, you could expect that a player with (+)100%mf gets 2x the amount of rares as a player with 0%mf. If (as suggested) a new account has better luck than a veteran account and Mr 0%mf gets 10x the amount of rares as Mr 100%mf, there’s a problem.

EDIT: If this happens consistently, then there is an obvious problem.
Sure, RNG is RNG and I appreciate that all outcomes are possibilities. When a wild possibility is replicable, then it is not RNG.

what’s the problem if people give it a go and there’s only a sample size of 10? Why get so upset about it? Sure it’s not statistically significant but who knows maybe it will get enough hype for Anet to look into it. And what is wrong with that? And if they don’t look into it then oh well try to get a bigger sample size and try again. It’s nothing to argue about. At worst 10 people wasted their time

Sorin.4310:

So your criteria for the system is broken is a possible outcome of the designed system, and this will show that further investigation needs to be done?

I give up. I now understand why the devs stopped responding to RNG broken threads. They realized long ago that the players have no idea what they are doing.

Pretty much this. The idea that 10-30 players farming the same world boss will provide data that may be worth further investigation is laughable at best.

Andraus.3874:

what’s the problem if people give it a go and there’s only a sample size of 10? Why get so upset about it? Sure it’s not statistically significant but who knows maybe it will get enough hype for Anet to look into it. And what is wrong with that?

I don’t see anyone here upset. That aside, the problem is that a small sample size most certainly will not generate “hype” for Anet to look into it. It’s just silly and a waste of time for all that participate. But hey, if you want to waste your time making videos farming a boss with 10 other people and post the results here, go for it.

Heck, I’m not on either side of the issue. I just don’t think there is any value on the purposed experiment.

This is the quote from John Smith I was thinking of earlier today:

“This is the kind of post I mean when I say evidence. I don’t require numbers I require a coherent idea that makes sense in this setting. If I think it’s valid I’ll be researching it myself (assuming I haven’t already).”

From

https://forum-en.gw2archive.eu/forum/archive/bltc/Increasing-trading-post-tax/page/2

JS has proven to be a reasonable and open-minded person. If you can show that there may be a problem, it doesn’t matter what the other players say, if you can convince him the issue should be looked at then he will do so.

Check this thread out. Apparently DR affects different accounts differently. Now figuring out why would be awesome. But again that could be so many things like how old the account is. How old the character is which way the wind is blowing ….. Etc

https://forum-en.gw2archive.eu/forum/game/gw2/Diminishing-Returns-in-the-Silverwastes/first#post5030253

This might be a better starting point because it would be more easily reproduced as in getting less karma crests and silver/copper if you can figure out why it happens.

Andraus.3874:

Check this thread out. Apparently DR affects different accounts differently. Now figuring out why would be awesome. But again that could be so many things like how old the account is. How old the character is which way the wind is blowing ….. Etc

https://forum-en.gw2archive.eu/forum/game/gw2/Diminishing-Returns-in-the-Silverwastes/first#post5030253

This might be a better starting point because it would be more easily reproduced as in getting less karma crests and silver/copper if you can figure out why it happens.

Or even this

This right here is the reason why this whole experiment should be done. There are multiple responses of people getting multiple pres, or multiple of one rare/exotic item.

Perhaps the first step is that all those participating should read up on the basics regarding probability. It’s helps to have a foundational knowledge about this when interpreting results.

This is a far better approach to the RNG discussion that any i have seen on this forum before. I sincerely approve (even if i’m sure there’s no problem to find in the first place).