Guild Wars 2

ANET, Why oh why has ecto salvage been nerfed? or is it a bug – a very expensive one since I just bought 2500 ecto. I used to get 1.77 dust per ecto with mystic kit, now 1.07

WTH?

Use the copper fed salvage o matic, you’d be better off

Also used to get about 25,000 luck a stack, now only about 12,000.
???

Are you sure it’s not just RNG, like it pretty much always is?

2500 is a pretty big sample size.

People often think something is nerfed when RNG doesn’t favour them though.

Lazaar.9123:

People often think something is nerfed when RNG doesn’t favour them though.

Behellagh.1468:

2500 is a pretty big sample size.

Lazaar.9123:

People often think something is nerfed when RNG doesn’t favour them though.

2500 is a large sample size. That’s not to say that it isn’t possible to get the results he got without the chances having changed, but that’s a pretty low chance.

As a former mass salvager I have never encountered such a low return rate with that sample size. Not even close. This indeed deserves to be looked into, I hope John will pick up on it.

(On a side note: it would explain why the dust to ecto price is no longer in sync – dust should be a kitteneaper with the current ecto prices).

There is no nerf. If there’s a nerf the price should changed and nothing changed.

You would think the price should change already with all the flippers around. It’s been 6 hours.

laokoko.7403:

There is no nerf. If there’s a nerf the price should changed and nothing changed.

You would think the price should change already with all the flippers around. It’s been 6 hours.

Well supply is up and price down since the patch. That indicates on the surface that nothing has changed.

We need confirmation from other salvagers if everything is OK or not.

I tried it.

250 ectos ==> 434 dust.

2500 IS a large sample, but nothing stops you to throw a coin 1000 times and it gives tails 900 times for example.

2500 isn’t really a large sample size. 250,000 maybe, but not 2500. There are 100,000’s of ectos passing through the market every day, and a nerf would be evident in the prices of dust and ectos. Since they haven’t changed I can say with absolute certainty that it has not been nerfed, you just got a bad streak.

The one time there was a bug with ectos it was very quickly reflected in the market price.

Also… kit does not effect the ecto→dust conversion. You wasted a lot of money on kits.

Belzebu.3912:

2500 IS a large sample, but nothing stops you to throw a coin 1000 times and it gives tails 900 times for example.

While that is theoretically possible, the chances of that are astronomical.

Maybe he just deposit something by accident or didn’t count right. Just saying.

Windu The Forbidden One.6045:

Belzebu.3912:

2500 IS a large sample, but nothing stops you to throw a coin 1000 times and it gives tails 900 times for example.

While that is theoretically possible, the chances of that are astronomical.

Sure it is, but I gave an extreme example on purpose, but a lower average return in a sample of 2500 is totally plausible.

Belzebu.3912:

2500 IS a large sample, but nothing stops you to throw a coin 1000 times and it gives tails 900 times for example.

You could have every one on the planet earth flipping coins at a rate of 1000 flip per second from the start of the creation of the universe, and most likely that will still not happen.

The OP either lied, Anet random number generator is broken, or the OP messed up something(for example using the wrong salvage kit).

Because even if every single player salvage ectos non stop from the opening of guild wars 2 server, it is still highly unlikely what the OP said occurred unless there is something wrong with the random number generator.

Just because its improbable doesnt mean it didnt happen.

But just because it’s possible doesn’t mean you should take it into consideration. 2500 is a decent sample size.

Yeah im not siding with anyone. Just saying the arguement “its improbable” is pretty weak.

And anecdotal evidence is anecdotal. Lets ppl make more test and we’ll see. Do you guys know how much tread on salvage nerf of all kind were posted here? And as far as I’m concerned i never saw a single one being right. So maybe he’s right, but pass experience tell me that he just made a small mistake while counting his stuff.

spoj.9672:

Yeah im not siding with anyone. Just saying the arguement “its improbable” is pretty weak.

It’s not weak. Winning a lottery is improbable.

What the OP described is something “astronomically” small. Someone need to do the math to explain the situation.

spoj.9672:

Just because its improbable doesnt mean it didnt happen.

This.

If something have 1 chance in 100 billion, it still means it can happen.

His results arent as improbable as the coin example. Yes his sample size is large. But its not so large that its completely unrealistic. Discounting an arguement based on mathematical improbability is pretty foolish. You cant be sure that he wasnt just that unlucky.

Its possible he made a mistake but its also possible he was just unlucky. It doesnt really matter if other samples show that the rate is fine.

Belzebu.3912:

spoj.9672:

Just because its improbable doesnt mean it didnt happen.

This.

If something have 1 chance in 100 billion, it still means it can happen.

Right but if the number is like 1 in 1000000000000000000000000000000000……… with so many trailing zeros that is larger than every single character ever typed in the GW2 forums, most likely that didn’t happen.

You shouldnt discount results just because they seem unlikely.

spoj.9672:

His results arent as improbable as the coin example. Yes his sample size is large. But its not so large that its completely unrealistic. Discounting an arguement based on mathematical improbability is pretty foolish. You cant be sure that he wasnt just that unlucky.

Its possible he made a mistake but its also possible he was just unlucky. It doesnt really matter if other samples show that the rate is fine.

No you don’t understand. When people say flipping 900 tails in 1000 toss…

I dont’ think people really understand what they are talking about. That is a number so small that you will only expect to occur once when every single person on the planet earth start flipping coin at a rate of 1000 coin per second from the start of universe creation in a bazillion multi universe.

Wouldn’t be surprised at all if the salvage rates, loot tables, etc. are monitored and adjusted regularly. Maybe it’s an automated process to keep certain ratio’s in the economy?

ZudetGambeous.9573:

2500 isn’t really a large sample size. 250,000 maybe, but not 2500. There are 100,000’s of ectos passing through the market every day, and a nerf would be evident in the prices of dust and ectos. Since they haven’t changed I can say with absolute certainty that it has not been nerfed, you just got a bad streak.

The one time there was a bug with ectos it was very quickly reflected in the market price.

Also… kit does not effect the ecto->dust conversion. You wasted a lot of money on kits.

It likely is within the first standard deviation level of confidence that the chances for dust got changed. So about 68%. Not high enough to be absolutely sure, bit high enough it should be double checked.

laokoko.7403:

spoj.9672:

Yeah im not siding with anyone. Just saying the arguement “its improbable” is pretty weak.

It’s not weak. Winning a lottery is improbable.

Yeah, but if somebody comes along and says “Hey, I won the lottery!” Would you argue, “No you didn’t, that’s extremely improbable.” ??

2500 is a large sample size and statistically significant. Whether or not you can flip heads 900 times irrelevant. That’s the purpose of a large sample size, to reduce statistical variability. If you don’t agree that 2500 is enough or you’re stuck on the idea that, for whatever reason, he just can’t be right, please suggest a sample size that would satisfy you. Using market trends as proof that he’s wrong is sketchy at best.

To clarify, I don’t agree or disagree (or care really). As often as I see this same sort of back and forth about statistics go on, rarely does someone provide the number that they think would satisfactorily provide statistical validity for a conclusion.

Black Frog.9274:

2500 is a large sample size and statistically significant. Whether or not you can flip heads 900 times irrelevant. That’s the purpose of a large sample size, to reduce statistical variability. If you don’t agree that 2500 is enough or you’re stuck on the idea that, for whatever reason, he just can’t be right, please suggest a sample size that would satisfy you. Using market trends as proof that he’s wrong is sketchy at best.

To clarify, I don’t agree or disagree (or care really). As often as I see this same sort of back and forth about statistics go on, rarely does someone provide the number that they think would satisfactorily provide statistical validity for a conclusion.

Well, in a situation like flipping a coin, 2500 flips is only enough to say that with ~68% confidence that there is no more than a 0.01 margin of error.

You would need 10,000 flips to be 95% confident. And 27,225 flips to be 99.9% confident.

Given that ecto salvage isn’t a direct you get 1 or 0, the numbers would likely change, but I doubt it’s enough for a 99.9% confidence level.

Definitely enough to warrant a post asking if it got changed. The OP shouldn’t have worded it as a certainty, not with his sample size, but the sample size is high enough to warrant asking.

Seera’s got the right idea.

Though one day I was doing SE P1 and all three of the golems dropped me exotic gear outside of their champ bags.

It all feeds into my illogical superstition that ArenaNet programmed a second line of RNG.

RNG Line 1 dictates drop rates, salvage rates and forge rates.

RNG Line 2 dictates individuals’ (or shards’) RNG Line 1 rates per day/hour.

It’s an RNGception!

Seera wrote what I tried to meant but with reasonable examples

For the number of possibilities from ecto salvage, in a pool of 2500 tries, your results can be inside the error margin.

Seera.5916:

Black Frog.9274:

2500 is a large sample size and statistically significant. Whether or not you can flip heads 900 times irrelevant. That’s the purpose of a large sample size, to reduce statistical variability. If you don’t agree that 2500 is enough or you’re stuck on the idea that, for whatever reason, he just can’t be right, please suggest a sample size that would satisfy you. Using market trends as proof that he’s wrong is sketchy at best.

To clarify, I don’t agree or disagree (or care really). As often as I see this same sort of back and forth about statistics go on, rarely does someone provide the number that they think would satisfactorily provide statistical validity for a conclusion.

Well, in a situation like flipping a coin, 2500 flips is only enough to say that with ~68% confidence that there is no more than a 0.01 margin of error.

You would need 10,000 flips to be 95% confident. And 27,225 flips to be 99.9% confident.

Given that ecto salvage isn’t a direct you get 1 or 0, the numbers would likely change, but I doubt it’s enough for a 99.9% confidence level.

Definitely enough to warrant a post asking if it got changed. The OP shouldn’t have worded it as a certainty, not with his sample size, but the sample size is high enough to warrant asking.

Sounds good! Though now I’m wondering why polling is generally considered solid with around 1500 people responding.

Black Frog.9274:

Seera.5916:

Black Frog.9274:

2500 is a large sample size and statistically significant. Whether or not you can flip heads 900 times irrelevant. That’s the purpose of a large sample size, to reduce statistical variability. If you don’t agree that 2500 is enough or you’re stuck on the idea that, for whatever reason, he just can’t be right, please suggest a sample size that would satisfy you. Using market trends as proof that he’s wrong is sketchy at best.

To clarify, I don’t agree or disagree (or care really). As often as I see this same sort of back and forth about statistics go on, rarely does someone provide the number that they think would satisfactorily provide statistical validity for a conclusion.

Well, in a situation like flipping a coin, 2500 flips is only enough to say that with ~68% confidence that there is no more than a 0.01 margin of error.

You would need 10,000 flips to be 95% confident. And 27,225 flips to be 99.9% confident.

Given that ecto salvage isn’t a direct you get 1 or 0, the numbers would likely change, but I doubt it’s enough for a 99.9% confidence level.

Definitely enough to warrant a post asking if it got changed. The OP shouldn’t have worded it as a certainty, not with his sample size, but the sample size is high enough to warrant asking.

Sounds good! Though now I’m wondering why polling is generally considered solid with around 1500 people responding.

2500 is needed for 68% confidence that the error is not more than 0.01.

Polling typically has a margin of error of 2 or 3. A much larger margin of error. And for polling a margin of error of 2-3 is enough to work with in many cases where the public sees the polls (political polls being one of them).

Some people playing the market players?

Wolfheart.7483:

In regard to examples like the coin flipping:

Think of it this way: each individual time you flip the coin, it’s still a 50/50 chance it lands on tails. So if you flipped the coin 999 times and got tails 899 times, then on that 1000th flip it is still a 50/50 chance it lands on tails.

Having it land on tails 899/999 times previously has no bearing whatsoever on the 1000th flip. None at all.

Thats not how it works…

Wolfheart.7483:

In regard to examples like the coin flipping:

Think of it this way: each individual time you flip the coin, it’s still a 50/50 chance it lands on tails. So if you flipped the coin 999 times and got tails 899 times, then on that 1000th flip it is still a 50/50 chance it lands on tails.

Having it land on tails 899/999 times previously has no bearing whatsoever on the 1000th flip. None at all.

But if a coin is Tails side heavy, it will more than likely land with Heads face up more than 50% of the time just due to gravity.

That’s what this post is asking. Is the coin a fair coin?

Ecto salvage rate should be 1.77 dust/ecto according to the OP. He’s gotten a decent sample size for determining a hypothesis. And his hypothesis is the ecto salvage rate has been altered somehow. AKA: the coin isn’t fair.

The larger the sample size, the closer the results will be to the true probability.

The more times you flip a coin, the more likely you’ll hit 50/50. Flip a coin twice and you could get 100% heads or 100% tails, even with a fair coin. But flip it 10 times and the chances are much lower. 100 times, even lower. Due to the independent nature of coin flipping.

That’s why other posters have called him out his sample size. It’s not large enough to say that it’s definitely changed. But enough to warrant a check to make sure it hasn’t been affected by something else unintentionally.

Wolfheart.7483:

In regard to examples like the coin flipping:

Think of it this way: each individual time you flip the coin, it’s still a 50/50 chance it lands on tails. So if you flipped the coin 999 times and got tails 899 times, then on that 1000th flip it is still a 50/50 chance it lands on tails.

Having it land on tails 899/999 times previously has no bearing whatsoever on the 1000th flip. None at all.

Expanding on this^

There is an exact same chance for EVERY combination of results from 1000 tosses to occur. The only reason 999/1 seems less likely is because you are looking for a SPECIFIC result in a sea of possible results. As there are a LOT more SPECIFIC results with a more even mix of heads and tails, the average will tend towards a 50/50 split and less towards the outlier results (1/999 and 999/1). Just be aware that you have the exact same chance of getting an outlier as you do of getting any other combination.

Granted, most people don’t care about the individual combination of toss results and care more about the overall average, which is why outliers are seen as “good” or “bad” luck.

Malediktus.9250:

Wolfheart.7483:

In regard to examples like the coin flipping:

Think of it this way: each individual time you flip the coin, it’s still a 50/50 chance it lands on tails. So if you flipped the coin 999 times and got tails 899 times, then on that 1000th flip it is still a 50/50 chance it lands on tails.

Having it land on tails 899/999 times previously has no bearing whatsoever on the 1000th flip. None at all.

Thats not how it works…

It is, actually. Any results from previous flips of the coin don’t tally up in a magical scale to influence what the next outcome will be. When you physically flip that coin, it’s a 50/50 chance. Every single time.

John Smith.4610:

There’s been no intentional nerf to ecto salvage rate, but I’ll pull the numbers and check it out.

John, you have the resources in game, I am sure, to do an actual test on this. Not just a database number reference check. You may look at the secret file that says what drop rates are supposed to be, but just because those variables are right doesn’t mean something else is wrong.

It’s time consuming, but the only way to verify this would be to do a test on 5000 ectos with the exact same salvage kit that the OP used, and look at the results. Even then some will complain the sample size isn’t big enough, but this sort of hard testing needs to be done in several areas if you ask me.
I do not just buy it when we are told the records show nothing. We’ve been through this before, and it was only after months that Anet finally admitted the problem was a bug, and not “RNG is RNG”

250 salvaged, 434 dust (1.73 average)

Slightly lower (but in line with expectations) than 1.844 average I’ve recorded the past 2-3 months (~18k ectos salvaged)

Neural.1824:

John Smith.4610:

There’s been no intentional nerf to ecto salvage rate, but I’ll pull the numbers and check it out.

John, you have the resources in game, I am sure, to do an actual test on this. Not just a database number reference check. You may look at the secret file that says what drop rates are supposed to be, but just because those variables are right doesn’t mean something else is wrong.

It’s time consuming, but the only way to verify this would be to do a test on 5000 ectos with the exact same salvage kit that the OP used, and look at the results. Even then some will complain the sample size isn’t big enough, but this sort of hard testing needs to be done in several areas if you ask me.
I do not just buy it when we are told the records show nothing. We’ve been through this before, and it was only after months that Anet finally admitted the problem was a bug, and not “RNG is RNG”

They may track what people actually salvage from ectos so he can just run a trend to see if that changed at all recently.

Neural.1824:

John Smith.4610:

There’s been no intentional nerf to ecto salvage rate, but I’ll pull the numbers and check it out.

John, you have the resources in game, I am sure, to do an actual test on this. Not just a database number reference check. You may look at the secret file that says what drop rates are supposed to be, but just because those variables are right doesn’t mean something else is wrong.

It’s time consuming, but the only way to verify this would be to do a test on 5000 ectos with the exact same salvage kit that the OP used, and look at the results. Even then some will complain the sample size isn’t big enough, but this sort of hard testing needs to be done in several areas if you ask me.
I do not just buy it when we are told the records show nothing. We’ve been through this before, and it was only after months that Anet finally admitted the problem was a bug, and not “RNG is RNG”

Currently, ectos can salvage into 1-5 dust. That means the current odds are potentially at a 20% chance each to get 1, 2, 3, 4, or 5 dust. Alternatively, they could be using a more complex method; 50% of 1 dust, 25% of 2, 12.5% of 3, 6.25% of 4, and 6.25% of 5 dust. The more complex method may be the more likely scenario accounting for the average being just under 2 dust an ecto. That being said, regardless of your sample size, if karma decided to make you her kitten for the day, you’re going to possibly end up with a 1:1 gain of dust, which is entirely within the realm of possibility. If that the case for everyone, the the prices of DUST would increase and be nearly on par with ectos themselves.

mtpelion.4562:

Wolfheart.7483:

In regard to examples like the coin flipping:

Think of it this way: each individual time you flip the coin, it’s still a 50/50 chance it lands on tails. So if you flipped the coin 999 times and got tails 899 times, then on that 1000th flip it is still a 50/50 chance it lands on tails.

Having it land on tails 899/999 times previously has no bearing whatsoever on the 1000th flip. None at all.

Expanding on this^

There is an exact same chance for EVERY combination of results from 1000 tosses to occur. The only reason 999/1 seems less likely is because you are looking for a SPECIFIC result in a sea of possible results. As there are a LOT more SPECIFIC results with a more even mix of heads and tails, the average will tend towards a 50/50 split and less towards the outlier results (1/999 and 999/1). Just be aware that you have the exact same chance of getting an outlier as you do of getting any other combination.

Granted, most people don’t care about the individual combination of toss results and care more about the overall average, which is why outliers are seen as “good” or “bad” luck.

False, flipping a coin follows a Bernoulli distribution
500-500 is more likely than 999-1
the probabilities looke like a gaussian curve
http://en.wikipedia.org/wiki/Gaussian_function

Malediktus.9250:

mtpelion.4562:

Wolfheart.7483:

In regard to examples like the coin flipping:

Think of it this way: each individual time you flip the coin, it’s still a 50/50 chance it lands on tails. So if you flipped the coin 999 times and got tails 899 times, then on that 1000th flip it is still a 50/50 chance it lands on tails.

Having it land on tails 899/999 times previously has no bearing whatsoever on the 1000th flip. None at all.

Expanding on this^

There is an exact same chance for EVERY combination of results from 1000 tosses to occur. The only reason 999/1 seems less likely is because you are looking for a SPECIFIC result in a sea of possible results. As there are a LOT more SPECIFIC results with a more even mix of heads and tails, the average will tend towards a 50/50 split and less towards the outlier results (1/999 and 999/1). Just be aware that you have the exact same chance of getting an outlier as you do of getting any other combination.

Granted, most people don’t care about the individual combination of toss results and care more about the overall average, which is why outliers are seen as “good” or “bad” luck.

False, flipping a coin follows a Bernoulli distribution
500-500 is more likely than 999-1

I’d re-read what he said.

mystic salvage kit has always had low ecto returns, you got lucky before, not now. black lion salvage kit is what has over 1.8+ returns