Guild Wars 2

Despite claims from an arenanet game designer posting on the forum, the mystic forge is indeed suffering from (unintended?) diminishing returns.

Since the game designer in question claimed that the mystic forge is not intended to suffer from diminishing returns, this has to be a bug of some kind in the game programming.

Large scale testing has shown this bug is easily reproducible.

All you need to do is stock up on a few stacks of T5 fine crafting materials and T5 dust, and have some spare skillpoints.

Put the T5 fine material, the same material of T6, dust, and philosophers stones in the mystic forge, and confirm you want to forge them.

Now repeat this process 20 times.
You will notice a clear downward pattern in the return of T6 fine materials, until you eventually no longer get more than the bare minimum possible crafting materials from the conversion.

Now get 50 of your friends or guild mates to test this as well. They will all get exactly the same results.

This doesn’t only apply to fine crafting materials, it applies to EVERYTHING that involves a random element in the mystic forge. Including making exotics using rare equipment, or using exotic weapons to make a legendary base weapon. everything is affected by these diminishing returns

Bugs in code happen, and some can be hardly detectable, but have an immense impact.
Maybe the game designer intended it to be no diminishing returns, but somewhat it happens due to a bug in diminishing return code.
Suffice to say that statistics is a valid clue that says there is obviously a problem, provided the samples are large enough to be conclusive.

Because of the randomness and the HUGE amount of money that needs to be invested to test this it was hard to confirm that the pattern was indeed there. Unfortunately by the time I was able to dump enough money(well over 100 gold in one go) into the mystic forge to be completely sure that this was happening every single time I already possible lost hundreds of gold in materials.

Alright well this doesn’t happen to me.

Wahaha.7938:

Alright well this doesn’t happen to me.

How many times did you do the conversion and how much time did you have between the first and last conversion?

I hope this isn’t like the dungeon DR glitch where some people were getting DR without running any dungeons.

My samples are too small so I won’t pretend they matter in any way, less than 20 tries per. In those samples I didn’t notice any trend however. The best piece of info I have is mystic clovers crafting which I attempted 107 times at once and didn’t get any dr (and got a 37% success rate).

If many people have done this and everyone has gotten the same result this is worth considering, although it’s still directly in contradiction to what the developers have said.

However, you haven’t really presented any evidence for your claim that this DR effect applies to every type of conversion.

When are you claiming that this DR came into effect? I converted ~800 batches of sigils a few days ago and I didn’t see any evidence of DR.

Wahaha.7938:

My samples are too small so I won’t pretend they matter in any way, less than 20 tries per. In those samples I didn’t notice any trend however. The best piece of info I have is mystic clovers crafting which I attempted 107 times at once and didn’t get any dr (and got a 37% success rate).

That isn’t a valid sample. That is a 1→1 conversion, of course it can’t be dimished. What the OP is talking about is when you convert something and get a random number back.

for fine crafting mats its 5-12, for common mats its 10-40 (? I think)

I haven’t tested it for the claim about exotic weapon conversion, but I know for a fact it exists for Armored Scales, and gossomer scraps.

lackofcheese.5617:

If many people have done this and everyone has gotten the same result this is worth considering, although it’s still directly in contradiction to what the developers have said.

However, you haven’t really presented any evidence for your claim that this DR effect applies to every type of conversion.

When are you claiming that this DR came into effect? I converted ~800 batches of sigils a few days ago and I didn’t see any evidence of DR.

Several members of my guild who were making money through sigil conversion reported that they stopped getting higher tier sigils after enough attempts. I didn’t do any testing on this myself though.
I finished my clovers before the mystic forge patch so I can’t get any data on those, but they use a different kind of lookup table to determine the results, mystic clovers probably aren’t at the top of the table, so even if there’s a wall that blocks you from getting “better” results after a certain threshold, it would still be possible to get them even if they’re lower on the list.

What was the date of the patch you’re claiming brought in DR?

Anyways, I have another bunch of sigils I didn’t bother to do earlier because I got bored; I can still run those through the Forge if we’re looking for data.

ZudetGambeous.9573:

Wahaha.7938:

My samples are too small so I won’t pretend they matter in any way, less than 20 tries per. In those samples I didn’t notice any trend however. The best piece of info I have is mystic clovers crafting which I attempted 107 times at once and didn’t get any dr (and got a 37% success rate).

That isn’t a valid sample. That is a 1->1 conversion, of course it can’t be dimished. What the OP is talking about is when you convert something and get a random number back.

for fine crafting mats its 5-12, for common mats its 10-40 (? I think)

I haven’t tested it for the claim about exotic weapon conversion, but I know for a fact it exists for Armored Scales, and gossomer scraps.

You say a 1 to 1 conversion isn’t valid and then proceed to talk about testing weapon conversion which is also 1 to 1.

Since we have no idea what the loot list for the mystic clovers looks like we have no idea if they are affected by DR or not. For all we know clovers themselves could be the “worst” result in terms of the list, which would mean even if other items get their chance for a successful roll reduced to 0% clovers would still be a possible result.

We do know that for transmuting items and for rerolling equipment there is a list where good results are at the top and have a low chance and bad results are at the bottom and we get those when the results at the top fail their dice roll. When DR kicks in the top results get their chances reduced to 0% meaning they always fail their dice roll, and then it keeps going down the list like normal until we get a successful dice roll on an item with more than a 0% chance.

ZudetGambeous.9573:

That isn’t a valid sample. That is a 1->1 conversion, of course it can’t be dimished. What the OP is talking about is when you convert something and get a random number back.

How about 2-4 T6 materials with that recipe? Judging by data I saw, these were not affected.

Unless we know exactly what the loot lists look like(which will never happen) it will be impossible to pinpoint exactly where loot gets effected.

Remember a dev posting a while ago that an anomaly with the loot lists caused the GS base weapons and dagger base weapons to be far too common because of the way the loot lists were set up, but since the other weapons had their loot lists set up differently they were affected to a different degree.

I did 1000’s of blue dye and didn’t notice any DR, but no yellows anymore? defo something nerfed. You can buy 2k dye for 1.6g try it yourself.

@Wahaha – The reason why what you’re describing with Mystic Clover won’t experience DR for a couple of reasons:

• There are no ‘lower quality’ Mystic Clover.
• You can only ever receive one Mystic Clover.
• The chances of getting a single Mystic Clover is fixed for the recipe.

I’m not sure that this would even be a valid test, but a pattern of DR for the Mystic Clover recipe might be observable in the number of T6 materials yielded per attempt at creating a Mystic Clover. Since that is a variable number of materials, you might see the amount yielded diminish to the minimum with a large enough series of attempts.

But ignoring the Mystic Clover recipe, the circumstances where I choose to believe that a pattern of DR is observable is:

• The number of upgraded materials received when upgrading a stack of Common/Fine materials.
• The chances of receiving a high-level or -quality item from four random items.

If a couple of people who choose to challenge this are willing to upgrade, say, 15 stacks of the cheapest Fine materials you can find and Fraps the results, we could see whether that series of upgrades is affected. It should only take a couple of series of results being observably unaffected to disprove the theory that DR is applied to upgrading Fine materials in the Mystic Forge.

I wouldn’t ask anyone to sit around trying to upgrade a series of Rares to Exotics just to try to disprove that possible avenue of DR. Plus it’s more difficult to set up those trials. You would need one person to try thousands of consecutive upgrades and also thousands of people to each try a single upgrade to get a large enough sample size to sufficiently reduce noise.

On a side note, I don’t even salvage in a series now; I salvage one or two thing at a time as I loot them, and I will log out and play an alt or wait until tomorrow between salvaging rares trying for ecto. Yes, I am that convinced that DR is being (un?)intentionally applied to every random source of loot.

Lucubration.8361:

But ignoring the Mystic Clover recipe, the circumstances where I choose to believe that a pattern of DR is observable is:

• The number of upgraded materials received when upgrading a stack of Common/Fine materials.
• The chances of receiving a high-level or -quality item from four random items.

If a couple of people who choose to challenge this are willing to upgrade, say, 15 stacks of the cheapest Fine materials you can find and Fraps the results, we could see whether that series of upgrades is affected. It should only take a couple of series of results being observably unaffected to disprove the theory that DR is applied to upgrading Fine materials in the Mystic Forge.

I wouldn’t ask anyone to sit around trying to upgrade a series of Rares to Exotics just to try to disprove that possible avenue of DR. Plus it’s more difficult to set up those trials. You would need one person to try thousands of consecutive upgrades and also thousands of people to each try a single upgrade to get a large enough sample size to sufficiently reduce noise.

I can understand believing the first case of DR, because it seems that a number of people have independently confirmed the same type of pattern. I won’t be fully satisfied without seeing the actual data, but it’s definitely something to be wary of.

However, I don’t think that you have much justification for believing there is DR in converting rares. Considering that this is completely different to the previous case this would essentially be a separate bug. Moreover, since it’s a binary success/failure of getting an exotic on every attempt, it’s very difficult to demonstrate any kind of DR at all. Where’s the evidence for DR?

Lucubration.8361:

On a side note, I don’t even salvage in a series now; I salvage one or two thing at a time as I loot them, and I will log out and play an alt or wait until tomorrow between salvaging rares trying for ecto. Yes, I am that convinced that DR is being (un?)intentionally applied to every random source of loot.

Nice superstition you have there.

Still, for your sake, I’ll link you here:
http://inquisitive-myths.blogspot.com.au/2012/10/gw2-salvaging-scraps-which-kit-to-use.html

Salvage isn’t affected by diminishing returns, and neither is opening loot bags.

All that proves is that four days ago, salvaging wasn’t affected by diminishing returns.

Guess what?

Four days ago, neither was the mystic forge.

I inveted a lot of time converting silver to gold. I recorded the cost (since I bought most of the silver I used) and yield of every attempt.

While some of my initial yields were very high (190+) I can’t produce a trend from this data. It is all over the place (looks like it might be a normal distribution, but I have low confidence in that) and appears to simply average to a yield of ~80 gold ore.

Approximately half the people (on average) that produce a random yield result will receive more than the mean, while the other half will receive a yield below the mean. One out of 32 people will have an above-average result on their first 5 combines. This is the perfect set up for perceived DR, but is really just a streak of good luck followed by arbitrary luck.

@Cerise: “A few stacks” is not enough to produce any statistically significant results. Any trend derived from fewer than 100 unbiased data points is sheer garbage from a statistical analysis standpoint. You need thousands of unbiased results to have any degree of confidence in your conclusions. What makes a result biased? In short, a data point is only unbiased if you decide to record it before knowing what the data point is—you have to decide “I am going to do this 5 times and record the results” as opposed to “Wow, that first one was pretty good. I’m going to keep track of these” or “Wow, I really got some bad results. I heard about a DR conspiracy theory, I’m going to report it to them.”

@lackofcheese – There are still two things unclear about that test:

• Gossamer, while more valuable and higher tier than Silk, is still a Common material. I’m specifically referring to the chance of Ectoplasm from salvaging Rare items, which being a higher quality material may be affected differently by DR.
• Concerning the linked test (and not DR), because Gossamer is a Common material, it may not be affected by the salvage kit’s, “chance to yield rarer items”. It’s unclear what exactly this percentage chance refers to, but “Rare” is also item quality and Gossamer doesn’t fall into it.

Edit:
@Fildydarie – These were consecutive upgrades at the Mystic Forge? Interesting. What day(s) did you perform these trials and how many upgrades did you perform consecutively?

Seems like a lot of people here haven’t even tried it. If you have not tried upgrading T5 materials to T6 materials, you cannot have an opinion. Go try it for yourselves and you will notice a clear downward trend. Just because something is “random” (and its not, its algorithmic) doesn’t mean it doesn’t have any constraints acting on it.

Vorpal.4683:

All that proves is that four days ago, salvaging wasn’t affected by diminishing returns.

Guess what?

Four days ago, neither was the mystic forge.

People have been claiming some kind of DR on the forge since at least 7 days ago, see here:
https://forum-en.gw2archive.eu/forum/game/crafting/Mystic-Forge-has-DR-too

More importantly, though, you’re talking as though everything in this game has DR unless it’s specifically proven otherwise. That’s a laughable attitude, honestly. Have you seen any evidence that salvaging is affected by DR?

@Lucubrication
I do agree that we need to see more data on salvaging for ectoplasm. It is indeed unfortunate that people don’t seem to have done large studies on the issue. Nonetheless, in the absence of decent evidence suggesting that there is DR on ectoplasm, there is little reason to believe in any.

As for the “chance to yield rarer materials”, most of what I’ve seen suggests that this does, in fact, refer to the common material one tier up. It’s unfortunate that salvage kits are so badly documented in-game, though.

Lucubration.8361:

@Fildydarie – These were consecutive upgrades at the Mystic Forge? Interesting. What day(s) did you perform these trials and how many upgrades did you perform consecutively?

I converted 82 stacks of silver since Saturday. I did one batch each day, but don’t know exactly how many or at what times—I don’t have my notes while I’m at work.

Well, I did more testing. I didn’t FRAPS anything, but I feel like if there is a problem, it’s with weapons/armor, possibly just rare – exotic. I just ran 800 runes through (200 conversions), and even at the tail end, I had batches of three or four major – superior conversions in a row. It’s hard to think there’s any DR in play when that happens.

Sanctus.8350:

Seems like a lot of people here haven’t even tried it. If you have not tried upgrading T5 materials to T6 materials, you cannot have an opinion. Go try it for yourselves and you will notice a clear downward trend. Just because something is “random” (and its not, its algorithmic) doesn’t mean it doesn’t have any constraints acting on it.

If you want to be pedantic about it, nothing is random, ever. Everything is governed by rules that, given a sufficiently powerful analytic system, can be evaluated to predict the result of an event before it happens.

What we have is most likely a pseudo-random number generator, designed to produce a series of numbers that are uniformly distributed over a large set. Because it produces a repeatable sequence of numbers with a uniform distribution, it is hardly random, however what is effectively random is the sequence of numbers a single user extracts from this master sequence. Thousands of users are taking the next value from the sequence for their own purposes—crafting, AI, attack damage, etc. As the complex interactions of thousands of entities is prohibitive to predict, the sequence of numbers generated by a single user is effectively imposible to predict, even if you had a full understanding of the engine and state of the pRNG, resulting in a series of values which are, for all practical purposes, completely random.

Just because you are not seeing the sequence you want to see does not mean the results are based on anything other than a sequence of uniformly distributed numbers. How many people aren’t seeing the trend that you do and go about their business because their yields are increasing? Or people that don’t perform multiple trials because of how low their first yield was. Admittedly, had my first yield been unprofitably poor, I wouldn’t have performed a second, which is why I have to discard that result from my data as being biased. Just because there is a trend does not mean that the pRNG is producing biased results—with unifromly distributed data, the further one data point is from the mean, the greater the probability that a second data point will be closer to the mean.

I guess RNG is now DR.

I have converted literally over 200 stacks of silver into gold usually in bulk (30-50 stacks at a time). My running average was 250 silver > 89 gold. If I break them down into sub groups of 10 the averages typically range from 85-95 with no noticeable decline especially not towards the bare minimum (40 gold).

So I am highly speculative of your claim that the mystic forge is being affected by DR.

Promega.7628:

I have converted literally over 200 stacks of silver into gold usually in bulk (30-50 stacks at a time). My running average was 250 silver > 89 gold. If I break them down into sub groups of 10 the averages typically range from 85-95 with no noticeable decline especially not towards the bare minimum (40 gold).

So I am highly speculative of your claim that the mystic forge is being affected by DR.

Were these conversions done recently and consecutively?

I’ve done 5 consecutive conversions each of the past two days, and on each day there was a monotonic downward trend in the number of resulting materials I got out of it. I know 10 trials isn’t fantastic for statistics, but it does make me wary of continuing to do them several at a time, unless/until someone else verifies that they’ve done a lot of conversions one after the other within the past couple of days and saw no DR.

Fildydarie.1496:

Any trend derived from fewer than 100 unbiased data points is sheer garbage from a statistical analysis standpoint.

You definitely don’t always need at least 100 trials to get statistical significance. If I flip a coin 10 times and it comes up heads each time, I will start to doubt that it’s a fair coin, because that has only 1/1024 chance of happening with a fair coin. It’s not proof, of course, but it’s a reject-the-null-hypothesis situation with 99.9% confidence.

I’d agree that you need a sample size about that large if you’re sampling from a bigger population (50 patients in each arm of the study is a good rule of thumb for preliminary clinical trials), but that’s not exactly what’s happening here.

For silver->gold, let us assume that there is a valid range of numbers, between which any integer value is possible (e.g. the yield could be any integer between 10 and 200, inclusive).

The lowest yield I’ve heard of is 40. The greatest is 192 (was possibly 194, I’d have to check my notes). This means there are (192-40+1)=153 potential outcomes at a minimum.

Flipping a coin has two potential outcomes.

10 samples is five times the size size of the set of posible outcomes for one, and 1/15th the size of the other.

10 samples is not nearly enough for establishing a trend when there are 150+ potential outcomes.

I saw this happen from throwing thousands of large scales and pieces of gear, total value of at least 300 gold, into the mystic forge.

Fildydarie.1496:

Flipping a coin has two potential outcomes.
…
10 samples is not nearly enough for establishing a trend when there are 150+ potential outcomes.

When the potential trend being suggested is diminishing returns, though, the outcomes are again reduced to just two: below the last one or above it (plus a small chance of the same number, for the discrete case we’re discussing here).

Suppose I only ever do two consecutive conversions at a time, and wait long enough between each pair to make sure there’s definitely no DR effect remaining. Suppose also that every time, the second conversion gives fewer materials than the first one. Now we’re basically back into coin-flip territory, albeit with varying probability depending on where the first conversion is on the relevant interval.

LinseyMurdock.4165:

If we had DR in Salvaging or the Mystic Forge, we would tell you, just like we told you about the DR in other areas of the game.

Thanks for your prompt reply to this issue, and for the reassurance that DR doesn’t exist on the Forge. For the record, I don’t think the OP or (most) others in this thread thought this was some secret attempt to sneak DR into other areas of the game, hence the “glitch?” in the thread title. It was more a concern that there might be some kind of bug in which intentional DR in one area had unintended DR effects in another.

I don’t really understand the reason for not having DR in the mystic forge.

Take for example large scales into armored scales. Putting in 50 large scales nets 5-16 armored scale. For me the average has been about 10 per conversion if I only count the first 20 conversion of the day. If I do the conversion 30 more times right afterwards, it drops down to an average of 6. This means that even though on average 10 armored scales are created from the conversion for most people, people who, lets say, do 100 conversions end up with an average of 6.8 armored scales. This means that for someone doing a large amount of conversions every attempt took 3.2 armored on scales out of the economy. This goes along with the goal to prevent too many items from entering the economy and piling up over time.

I want you guys to think about this a bit. Seriously. Why would we ever want to systematically discourage you from chucking things into the Forge or salvaging them as often or as fast as you want to? It is SO SO SO (I can’t stress this enough) important for items to be leaving the economy at a healthy ratio to the items being created and entering the economy, otherwise, there would be no economy. Everything would pile up and pile up until everything was worth nothing because the supply would be endlessly higher than the demand.

Bottom line, DR systems exist to prevent players from generating too much currency and destroying the economy. Why would we have DR on the things that balance and fix the economy?

There a simple answer to that last question: to limit income of new items/resources to players via Mystic forge. By your own logic: less items – healthier the economnomy is.

I dunno. I’ve seen reports of people’s income from the material upgrade recipe over 30 attempts and it looked pretty clearly like a definite downward spiral.

LinseyMurdock.4165:

It’s not really a reason not to have it in the Mystic Forge. It’s more the reason why we never needed to ask the question “should we have DR in the Mystic Forge or not”.

Good point. But that still means there could be a glitch that shouldn’t be there. Unfortunately this is really bad time to buy large scales so I’ll have to wait before I can record my findings on… a large scale.
I also have no skill points left to do any more tests because I used at least 200 of them on my main and 100 of them on my alt on conversions already.

Thank you for taking the extra time to look into this and making sure there is no bug, consider this issue closed at the time. I asked for the thread to be deleted since it’s just wasting peoples time now.

LinseyMurdock

Though then I would ask if that works when it also has the side effect of causing people to stop using it altogether.

Umm… To increase time of achieving something for players, encouraging them to buy gems to exchange them for gold and buy needed materials off TP to save a couple of days they would’ve spent @Mystic Forge.

^_^

LinseyMurdock.4165:

If we had DR in Salvaging or the Mystic Forge, we would tell you, just like we told you about the DR in other areas of the game.

Er… you mean after nearly a month of silence in the face of dozens of threads about it? That’s not exactly inspiring and isn’t something you should be bringing up as an example of how transparent about these things you are. It took dozens of threads and a huge general uproar to even get a response.

Also, as to why you’d add DR to the Mystic Forge in the first place, you could easily ask why you added DR that only affects legitimate players to the rest of the game, an obviously ridiculous and anti-player, anti-fun system to add at all. Easily one of the worst designs ever put into a game.

As for people not wanting to use the Mystic Forge, well that’s currently the case anyway due to Godskull weapons cheaters being given a free pass to cheat and make thousands of gold and dozens of precursors, and then the rest of us get told it’s supposed to be “fun and exciting” to throw hundreds of gold into the MF with absolutely nothing to show for it.

I’ve actually heard from no less than a dozen people that are quitting the game due to how ridiculous the Mystic Forge and getting a precursor is now (due to the Godskull weapons cheaters inflating the price, and the horribly designed garbage RNG system for getting it from the Forge), after farming and grinding for months to attempt it only to get nothing.

I’m curious if there’s been any thought given to implementing some Streak-breaker code to prevent these sorts of rampant bad-rolls—→that resemble D.R. so much….

I’ve experienced them as well when I attempted to salvage a bunch of Rares all in 1 sitting. And normal players don’t care about the science behind R.N.G’s and how they can resemble the most Malicious systems possible. All they they walk away with is the Perception of that Malice as being intended. And if a couple checks and balances could fix that perception… wouldn’t that be better for sustaining the long term population of the game VS. the risk of a bunch of rage-quits? ….the Hottest topic going on right now about crafting, is the R.N.G. factor for legendaries and overwhelming running commentary I’ve seen is that people just don’t trust RNG’s. Nor should they, since the science of RNG’s themselves dictates that massive amounts of disparity can result from sample sizes smaller than the 1000’s. So, Is that really fair to people who will never have 1000’s of stacks of anything? I don’t think I’m the only one, I seem to recall Mr. Smith also saying something about minimizing disparity where possible….

Thank you for taking the time to look into this and respond to us, Lindsey. The issue was, I think that these systems aren’t at all transparent to players (which is a good thing). In appearance, though, it all boils down to, “a window opens, stuff comes out of it and goes into my inventory”, so these different systems can appear entirely related. :p

Since Linsey has replied and provided a detailed answer to the topic of this thread, it will now be closed.