More evidence that Forgehammer doesn't work in Mastercraft tasks

ontheleftcoast

For several weeks I've noticed poor rates of success with my Forgehammer. However, I was told to chalk this up to the fact most of what I was using it for was the new Tier II tasks with low percentage chances of success. But a few weeks ago I had completed a few of the Tier II missions and finally decided to start mass producing items for sale in the AH and I began keeping running totals of my success rates with, and without, the Forgehammer.

On the toon Magrath Kinslayer I unlocked Weaponsmithing Tier II and create Adamant Blooms. Out of 28 attempts 17 yielded blooms, 17/28 = 60.7% pretty much what you'd expect.

On the toon Varis Trollbane I've unlocked Alchemy, Artificing, Jewelcrafting, Mailsmithing, and Weaponsmithing. Out of 64 attempts without the Forgehammer 37 succeeded, 37/64 = 57.8% a bit low but totally inline with the sample size. But with the 64 attemps WITH the Forgehammer only 38 succeeded. That's right, only 1 more success in 64 tries. The success rate WITH the Mythic Forgehammer was 38/64 = 59.4% basically the same as the two other rates without it. Despite telling me the success rate will be 75% it is not even close to it in any way, shape, or form.

Furthermore, when combined with the nearly 100 samples I took trying to create gold nuggets I now have over 160 documented attempts of crafting with the Mything Forgehammer of Gond and it has *NEVER* yielded an improved result over not using it.

Cryptic, this is very badly broken. Will you please take a look at this? If not, why not?

tripsofthrymr

Most people that play slot machines lose. Some win big. That's the long tail of the normal probability distribution curve. While the majority of players will end up near the expectation value of 75%, some will succeed far more often and some will succeed far less often.

Sometimes, you'll see people mention that they blew through a stack of preservation wards for a 10% upgrade chance. They're less likely to advertise the times it took a handful of wards for the same attempt. Same thing.

I have paid attention and recorded samples of a few hundred rolls on multiple occasions since open beta, and am convinced that the RNG is fair in the long run. Some software random number generators are known to be streaky. I don't have evidence one way or the other whether that's the case in Neverwinter.

It sucks to be on the wrong side of the probability distribution. In the long run, it will even out.

ontheleftcoast

tripsofthrymr said:

Most people that play slot machines lose. Some win big. That's the long tail of the normal probability distribution curve. While the majority of players will end up near the expectation value of 75%, some will succeed far more often and some will succeed far less often.

Sometimes, you'll see people mention that they blew through a stack of preservation wards for a 10% upgrade chance. They're less likely to advertise the times it took a handful of wards for the same attempt. Same thing.

I have paid attention and recorded samples of a few hundred rolls on multiple occasions since open beta, and am convinced that the RNG is fair in the long run. Some software random number generators are known to be streaky. I don't have evidence one way or the other whether that's the case in Neverwinter.

It sucks to be on the wrong side of the probability distribution. In the long run, it will even out.

You are clearly out of your field of knowledge on this. After 160+ attempts the chance this is anything but a bug is vanishingly small. You'd have better odds at getting hit by a meteor. The code is broken. I could collect another 160, or a thousand samples and it would be the same. It is BROKEN. But apologists and fools apparently don't realize it.

bronto111

RNG is and has always been broken,the proof is very simple :doing the same tasks on a map with fewer players consistantly yields higher success.
which is why you see so many high geared toons popping into helms hold etc where its easy to find an instance that has a low population.Master crafters especially as well as people refining.
Likewise you get a better success when you first log in ,and immediately after the server resets.
After years of playing my sample size is HUGE and shows the same thing RNG is broken,
!!!forgehammer and legendary assets with 95% success still fails 25% of the time!!!

plasticbat

bronto111 said:

RNG is and has always been broken,the proof is very simple :doing the same tasks on a map with fewer players consistantly yields higher success.
which is why you see so many high geared toons popping into helms hold etc where its easy to find an instance that has a low population.Master crafters especially as well as people refining.
Likewise you get a better success when you first log in ,and immediately after the server resets.
After years of playing my sample size is HUGE and shows the same thing RNG is broken,
!!!forgehammer and legendary assets with 95% success still fails 25% of the time!!!

I personally do not go to an instance that has a low population. I only go to the instance that has no population to do RNG stuff. However, profession RNG does not apply.

hustin1

I've been wondering for a long time if there is an issue with how often the RNG is seeded. Each roll should give it a different seed value, but I wonder if that's not happening every time in an instance with lots of traffic going on. When I want to do some serious refining I always go somewhere with little to no traffic. It would be nice if we could see the actual generated value each time, even if it only showed up in the chat log via a chat command to toggle the feature.

pterias

hustin1 said:

It would be nice if we could see the actual generated value each time, even if it only showed up in the chat log via a chat command to toggle the feature.

That will never happen because then people would be able to prove how screwy the RNG is. We might even be able to work out consistent strategies for better results. The horror.

minotaur2857

hustin1 said:

I've been wondering for a long time if there is an issue with how often the RNG is seeded. Each roll should give it a different seed value, but I wonder if that's not happening every time in an instance with lots of traffic going on. When I want to do some serious refining I always go somewhere with little to no traffic. It would be nice if we could see the actual generated value each time, even if it only showed up in the chat log via a chat command to toggle the feature.

QFT this is my opinion, the NW RNG is streaky as hell, and I'm sure it's a seeding issue. I've seen once in a lifetime events 3 times in a week often enough to know there are issues.

I would back this request to see the actual RNG roll, an older Cryptic game CoH allowed you to see this if you wanted to. It would not surprise me at all if they used the same RNG and that one was broken as hell (75% hit chance missing 20+ times in a row 6 times in a week where it wasn't forced to hit by the streakbreaker which the logs told you about for example).

greywynd

ontheleftcoast#2086 said:

You are clearly out of your field of knowledge on this. After 160+ attempts the chance this is anything but a bug is vanishingly small. You'd have better odds at getting hit by a meteor. The code is broken. I could collect another 160, or a thousand samples and it would be the same. It is BROKEN. But apologists and fools apparently don't realize it.

Actually, it isn't. The numbers you need to generate would be in the tens of thousands just to begin with.

minotaur2857

greywynd said:

ontheleftcoast#2086 said:

You are clearly out of your field of knowledge on this. After 160+ attempts the chance this is anything but a bug is vanishingly small. You'd have better odds at getting hit by a meteor. The code is broken. I could collect another 160, or a thousand samples and it would be the same. It is BROKEN. But apologists and fools apparently don't realize it.

Actually, it isn't. The numbers you need to generate would be in the tens of thousands just to begin with.

For this the number required is actually only about 40 or so (stats graduate talking here), 160 is plenty.

Beyond about 40 you can use the normal distribution to approximate to the binomial and you can say that the result lies outside the 95% confidence interval for the mean very easily. Once this happens repeatedly in the same direction something is not right.

Funnily enough the RNG had been behaving reasonably OK for me until last night on upgrading enchants when I did some mastercrafting with a forgehammer, and I only went 10 out of 16 at 75%. That is too small a sample on which to draw conclusions, but the same behaviour as experienced above.

callumf

Does

minotaur2857 said:

greywynd said:

ontheleftcoast#2086 said:

You are clearly out of your field of knowledge on this. After 160+ attempts the chance this is anything but a bug is vanishingly small. You'd have better odds at getting hit by a meteor. The code is broken. I could collect another 160, or a thousand samples and it would be the same. It is BROKEN. But apologists and fools apparently don't realize it.

Actually, it isn't. The numbers you need to generate would be in the tens of thousands just to begin with.

For this the number required is actually only about 40 or so (stats graduate talking here), 160 is plenty.

Beyond about 40 you can use the normal distribution to approximate to the binomial and you can say that the result lies outside the 95% confidence interval for the mean very easily. Once this happens repeatedly in the same direction something is not right.

Funnily enough the RNG had been behaving reasonably OK for me until last night on upgrading enchants when I did some mastercrafting with a forgehammer, and I only went 10 out of 16 at 75%. That is too small a sample on which to draw conclusions, but the same behaviour as experienced above.

What does that 10 out of 16 at 75% bit mean please? You got 10 enchants upgraded "correctly" in 3 out of 4 tries [75%]?

minotaur2857

callumf#9018 said:

Does

minotaur2857 said:

greywynd said:

ontheleftcoast#2086 said:

You are clearly out of your field of knowledge on this. After 160+ attempts the chance this is anything but a bug is vanishingly small. You'd have better odds at getting hit by a meteor. The code is broken. I could collect another 160, or a thousand samples and it would be the same. It is BROKEN. But apologists and fools apparently don't realize it.

Actually, it isn't. The numbers you need to generate would be in the tens of thousands just to begin with.

For this the number required is actually only about 40 or so (stats graduate talking here), 160 is plenty.

Beyond about 40 you can use the normal distribution to approximate to the binomial and you can say that the result lies outside the 95% confidence interval for the mean very easily. Once this happens repeatedly in the same direction something is not right.

Funnily enough the RNG had been behaving reasonably OK for me until last night on upgrading enchants when I did some mastercrafting with a forgehammer, and I only went 10 out of 16 at 75%. That is too small a sample on which to draw conclusions, but the same behaviour as experienced above.

What does that 10 out of 16 at 75% bit mean please? You got 10 enchants upgraded "correctly" in 3 out of 4 tries [75%]?

No I said the 10/16 was with the forgehammer, hence it was mastercrafting, 75% chance, 10 successes, out of 16 , it took me 16 attempts to make 10 vellum, which is a 62.5% success chance when I would expect 75%, but 16 attempts is not enough to judge that the RNG is wrong.

callumf

Ahh ok right. Thank you

zanaspus1

minotaur2857 said:

greywynd said:

ontheleftcoast#2086 said:

You are clearly out of your field of knowledge on this. After 160+ attempts the chance this is anything but a bug is vanishingly small. You'd have better odds at getting hit by a meteor. The code is broken. I could collect another 160, or a thousand samples and it would be the same. It is BROKEN. But apologists and fools apparently don't realize it.

Actually, it isn't. The numbers you need to generate would be in the tens of thousands just to begin with.

For this the number required is actually only about 40 or so (stats graduate talking here), 160 is plenty.

Beyond about 40 you can use the normal distribution to approximate to the binomial and you can say that the result lies outside the 95% confidence interval for the mean very easily. Once this happens repeatedly in the same direction something is not right.

Funnily enough the RNG had been behaving reasonably OK for me until last night on upgrading enchants when I did some mastercrafting with a forgehammer, and I only went 10 out of 16 at 75%. That is too small a sample on which to draw conclusions, but the same behaviour as experienced above.

Quoted for truth. Most stat people even change from t-distribution to normal distribution at n=30