Have you noticed less surveys?

rdhdkc · October 2020

I run dailies on 10 different characters. Over the last month I've noticed significantly less surveys. Prior to this, if I didn't do surveys every week, my bank space became very limited. Now I only wind up doing them once a week due to not having as many. I've spoken to several others that also say they have noticed far less surveys. Anyone else experiencing this or am I just getting horrid RNG lately?

rdhdkc · October 2020

rdhdkc wrote: »

I run dailies on 10 different characters. Over the last month I've noticed significantly less surveys. Prior to this, if I didn't do surveys every week, my bank space became very limited. Now I only wind up doing them once a week due to not having as many. I've spoken to several others that also say they have noticed far less surveys. Anyone else experiencing this or am I just getting horrid RNG lately?

What I meant to say was I only wind up doing them every 3 weeks now instead of every week due to less surveys.

tmbrinks · October 2020

I only track on a per-patch basis, but updating on a weekly-ish basis. Have noticed no significant changes in expected drop rate from the 36 characters I do writs on most days.

Data:

https://docs.google.com/spreadsheets/d/1yQDUjYNQVsIFl0ktkbkSlfYPkzP6pgCOTMQ-qCuzfaI/edit#gid=1039171516

Stinkyremy · October 2020

Despite ^^^ spreadsheets, I think you misunderstand what RNG is.
If you flip a coin 100 times, as it has 2 sides that doesn't mean it will land 50 on each side despise being a 50% chance of being either side.

You could do 7000 writs and never get a survey, while another could do the same and get 7000 surveys.
Extreme example of course.

tmbrinks · October 2020

Stinkyremy wrote: »

Despite ^^^ spreadsheets, I think you misunderstand what RNG is.
If you flip a coin 100 times, as it has 2 sides that doesn't mean it will land 50 on each side despise being a 50% chance of being either side.

You could do 7000 writs and never get a survey, while another could do the same and get 7000 surveys.
Extreme example of course.

is that directed at me? about how RNG works? appealing to the extremes does not make your argument a strong one

I am fully aware that with a 50% chance, that it's not going to hit 50/50 every single time you do 100 trials. But that is still the most likely outcome. some will be 49/51, rarely even 40/60... but the long term statistical average will be 50/50.

Yes, we have to make some basic assumptions that the drops follow a standard normal distribution. (I even account for this in my refining data) But as data sets get larger and larger (I think the over 150,000 writs I've done (and tracked) qualifies as "large") The odds will start to coalesce around the true average (Law of Large Numbers)

katanagirl1 · October 2020

I only do daily crafting writs on four toons and I see a lot of variability, some days multiple surveys and then a stretch of two or three days where I get none. It seems highly cyclical for me. Same for master writs.

Doing writs on more toons probably evens that out more.

I sometimes get the feeling that I am getting fewer but I don’t bother to track them. Writing them down to do them later is enough work for me.

Stinkyremy · October 2020

Despite ^^^s spreadsheets, I think you misunderstand what RNG is.
If you flip a coin and it has 2 sides that doesn't mean it is a 50 chan

tmbrinks wrote: »

Stinkyremy wrote: »

Despite ^^^ spreadsheets, I think you misunderstand what RNG is.
If you flip a coin 100 times, as it has 2 sides that doesn't mean it will land 50 on each side despise being a 50% chance of being either side.

You could do 7000 writs and never get a survey, while another could do the same and get 7000 surveys.
Extreme example of course.

is that directed at me? about how RNG works? appealing to the extremes does not make your argument a strong one

I am fully aware that with a 50% chance, that it's not going to hit 50/50 every single time you do 100 trials. But that is still the most likely outcome. some will be 49/51, rarely even 40/60... but the long term statistical average will be 50/50.

Yes, we have to make some basic assumptions that the drops follow a standard normal distribution. (I even account for this in my refining data) But as data sets get larger and larger (I think the over 150,000 writs I've done (and tracked) qualifies as "large") The odds will start to coalesce around the true average (Law of Large Numbers)

Not really because you again seem to not understand what RNG means.
While my comment wasn't aimed at you it was OP, you could easily have a human counterpart do the exact same amount of writs or any action based on RNG and the counterpart could come out with a vastly different outcome.

For simple sake, you both flip a coin 100 times, yours is heads 83 times, his is heads 22 times.
Completely different outcomes because it in RANDOM
The only way you would be able to get close to making an informed assumption to the % rate of an RNG is if you had a large pool of statistics from different sources to measure against and observe a median.
This is common knowledge of statistical analysis.

Anyway, you do you, carry on and enjoy yourself.

tmbrinks · October 2020

Stinkyremy wrote: »

Despite ^^^s spreadsheets, I think you misunderstand what RNG is.
If you flip a coin and it has 2 sides that doesn't mean it is a 50 chan

tmbrinks wrote: »

Stinkyremy wrote: »

Despite ^^^ spreadsheets, I think you misunderstand what RNG is.
If you flip a coin 100 times, as it has 2 sides that doesn't mean it will land 50 on each side despise being a 50% chance of being either side.

You could do 7000 writs and never get a survey, while another could do the same and get 7000 surveys.
Extreme example of course.

is that directed at me? about how RNG works? appealing to the extremes does not make your argument a strong one

I am fully aware that with a 50% chance, that it's not going to hit 50/50 every single time you do 100 trials. But that is still the most likely outcome. some will be 49/51, rarely even 40/60... but the long term statistical average will be 50/50.

Yes, we have to make some basic assumptions that the drops follow a standard normal distribution. (I even account for this in my refining data) But as data sets get larger and larger (I think the over 150,000 writs I've done (and tracked) qualifies as "large") The odds will start to coalesce around the true average (Law of Large Numbers)

Not really because you again seem to not understand what RNG means.
While my comment wasn't aimed at you it was OP, you could easily have a human counterpart do the exact same amount of writs or any action based on RNG and the counterpart could come out with a vastly different outcome.

For simple sake, you both flip a coin 100 times, yours is heads 83 times, his is heads 22 times.
Completely different outcomes because it in RANDOM
The only way you would be able to get close to making an informed assumption to the % rate of an RNG is if you had a large pool of statistics from different sources to measure against and observe a median.
This is common knowledge of statistical analysis.

Anyway, you do you, carry on and enjoy yourself.

The 150,000 writs isn't a "large pool"?

BTW, I teach high school mathematics and statistics

Do you know the odds of getting 83 heads when flipping a coin 100 times? You are once again appealing to the extremes. Yes, it CAN happen. The odds are astronomically small of those values. (it's about 1 in 200 BILLION, just FYI). I'm not saying it CAN'T happen, but just very, very, very unlikely.

VaranisArano · October 2020

Stinkyremy wrote: »

Despite ^^^s spreadsheets, I think you misunderstand what RNG is.
If you flip a coin and it has 2 sides that doesn't mean it is a 50 chan

tmbrinks wrote: »

Stinkyremy wrote: »

Despite ^^^ spreadsheets, I think you misunderstand what RNG is.
If you flip a coin 100 times, as it has 2 sides that doesn't mean it will land 50 on each side despise being a 50% chance of being either side.

You could do 7000 writs and never get a survey, while another could do the same and get 7000 surveys.
Extreme example of course.

is that directed at me? about how RNG works? appealing to the extremes does not make your argument a strong one

I am fully aware that with a 50% chance, that it's not going to hit 50/50 every single time you do 100 trials. But that is still the most likely outcome. some will be 49/51, rarely even 40/60... but the long term statistical average will be 50/50.

Yes, we have to make some basic assumptions that the drops follow a standard normal distribution. (I even account for this in my refining data) But as data sets get larger and larger (I think the over 150,000 writs I've done (and tracked) qualifies as "large") The odds will start to coalesce around the true average (Law of Large Numbers)

Not really because you again seem to not understand what RNG means.
While my comment wasn't aimed at you it was OP, you could easily have a human counterpart do the exact same amount of writs or any action based on RNG and the counterpart could come out with a vastly different outcome.

For simple sake, you both flip a coin 100 times, yours is heads 83 times, his is heads 22 times.
Completely different outcomes because it in RANDOM
The only way you would be able to get close to making an informed assumption to the % rate of an RNG is if you had a large pool of statistics from different sources to measure against and observe a median.
This is common knowledge of statistical analysis.

Anyway, you do you, carry on and enjoy yourself.

I'd be more inclined to believe that the amount of writs surveys had deviated substantially over the last month for certain players enough to indicate that the RNG had changed for them or in general...if they ever actually brought the data.

See, we've got a huge dataset of what "normal" writ RNG is over time, thanks to tmbrinks - you know, that "large pool of statistics from different sources to measure against and observe a median", you mentioned. We know what people "should" be getting over time, and thus we can tell if they've had normal or abnormal deviations...or we could, if they published their data. And a month's worth of data ought to be enough to spot significant deviations.

Instead, its usually like "I noticed I've been getting less of X."

Like, okay, you've noticed that you've gotten less? Cool, what are your numbers this month compared to last month?

Oh, you didn't keep track? Well, that makes it hard to figure out if there's really been a decrease or if there has, is it a significant decrease or within the bounds of normal RNG...

We humans are primed to remember bad outcomes.
What we "notice" isn't likely to be very useful when figuring out if we're getting less surveys.
The plural of anecdote is not data.

We really do need the data to evaluate the merit of the complaint.

Edited: it was pointed out that I said writs, when surveys are the topic of the thread.

Nestor · October 2020

RNG is a fickle mistress. She ebbs and she flows. She is only noticed when she is ebbing, and rarely appreciated when she is flowing.

No wonder she has a snitfit sometimes and does not cough up the shinys.

Mike0987 · October 2020

Same 10 crafters. Yes, about 30% less.

DaveMoeDee · October 2020

No data = nothing to see here.

Trinket61 · November 2020

Its been a while since i played seriously as i have spent nearly a year away on Destiny 2.

But yes , significantly less.

1 year ago with 15 chars i would get 1 or 2 each per day. it was very occasionally i would get nothing on 1 char.
This amounted to a loose average of 105 surveys a week.

Now i "might" get 1 survey every 5 chars so about 3 per day ....if i am lucky.
so thats about 21 per week.

I have gone from self sustaining + using a surplus to sell , to having to spend hours farming the mats.

This activity was something i used to enjoy , but now the growing conclusion is that it is simply not worth the effort anymore.

fioskal · November 2020

I don't keep pretty spreadsheets like Brinks does, but I haven't noticed a lower drop rate. I'm currently doing level 50 writs on 32 characters a day, on average.

SgtWinnie · November 2020

I do love tm's spreadsheets
I do more writs per day than most and I also haven't noticed a lower drop rate as of yet.
Hopefully it's just a bad month for you mate and it will pick up soon.

Trinket61 · November 2020

Folks , the clear title of the post is "SURVEYS"...............not writs.

I cannot fathom why some folks are confused.

idk · November 2020

tmbrinks wrote: »

I only track on a per-patch basis, but updating on a weekly-ish basis. Have noticed no significant changes in expected drop rate from the 36 characters I do writs on most days.

Data:

https://docs.google.com/spreadsheets/d/1yQDUjYNQVsIFl0ktkbkSlfYPkzP6pgCOTMQ-qCuzfaI/edit#gid=1039171516

If @tmbrinks says there is no indication of a change then there is no indication of a change. Afaik, they are the only person active in the forum that actually tracks this type of information and does it across many characters.

It is easy to think something has changed from either a small sample or just casual observance. RNG can paint a very skewed picture in both situations.

tmbrinks · November 2020

Trinket61 wrote: »

Folks , the clear title of the post is "SURVEYS"...............not writs.

I cannot fathom why some folks are confused.

the only way to get the surveys is from doing the daily crafting writs. 0.o They are intricately related.

SirAndy · November 2020

Y'all are missing the part where ESO is using weighted RNG. Always has been.

There's no pure, unweighted dice roll anywhere.

tmbrinks · November 2020

SirAndy wrote: »

Y'all are missing the part where ESO is using weighted RNG. Always has been.

There's no pure, unweighted dice roll anywhere.

You can't make a statement like this without proof. Please provide it.

Hotdog_23 · November 2020

All I can is thanks to tmbrinks for his/her time and attention to tracking this valuable information and then publishing if for everyone to see.

RNG can be a cruel mistress but over time even she succumbs to gods of math and time and must obey their law of averages.

Be safe and have fun

Trinket61 · November 2020

tmbrinks wrote: »

the only way to get the surveys is from doing the daily crafting writs. 0.o They are intricately related.

[snip]

VaranisArano wrote: »

I'd be more inclined to believe that the amount of writs had deviated substantially over the last month for certain players enough to indicate that the RNG had changed for them or in general...if they ever actually brought the data.
S.

Varanis is clearly talking about the drop rate of master writs as the number of daily crafting writs does not vary.
note: they are different!

[snip]

[Edited to remove Baiting]

tmbrinks · November 2020

nvm... shouldn't have responded to the baiting comments... just report and move on.

Reverb · November 2020

I do not have spreadsheets to back up my data, but over last 2(ish) years that my number of toons doing writs has been consistent, I’ve come to count on 8-12 surveys per day. That has not decreased of late. Saturday I got 10, today I got 9.

Sephyr · November 2020

Hotdog_23 wrote: »

All I can is thanks to tmbrinks for his/her time and attention to tracking this valuable information and then publishing if for everyone to see.

RNG can be a cruel mistress but over time even she succumbs to gods of math and time and must obey their law of averages.

Be safe and have fun

My thoughts exactly. Without their data, I wouldn't know how to begin checking on my own and while I haven't done nearly as many as they have - my short-term data is pretty on par with a few slight deviations (at most my monthly discrepancy is ranging around 3-7 since I've been keeping records which isn't all that long) and some of that can be attributed to maybe skipping a character and I don't realize to just the low luck of the draw. But I've not experienced any kind of significant drop unless, you know... I'm not doing them.

idk · November 2020

tmbrinks wrote: »

nvm... shouldn't have responded to the baiting comments... just report and move on.

I agree. More importantly, arguing about this significantly minor difference of opinion is very much detracting and derailing the thread.

tmbrinks · November 2020

Hotdog_23 wrote: »

All I can is thanks to tmbrinks for his/her time and attention to tracking this valuable information and then publishing if for everyone to see.

RNG can be a cruel mistress but over time even she succumbs to gods of math and time and must obey their law of averages.

Be safe and have fun

idk wrote: »

tmbrinks wrote: »

I only track on a per-patch basis, but updating on a weekly-ish basis. Have noticed no significant changes in expected drop rate from the 36 characters I do writs on most days.

Data:

https://docs.google.com/spreadsheets/d/1yQDUjYNQVsIFl0ktkbkSlfYPkzP6pgCOTMQ-qCuzfaI/edit#gid=1039171516

If @tmbrinks says there is no indication of a change then there is no indication of a change. Afaik, they are the only person active in the forum that actually tracks this type of information and does it across many characters.

It is easy to think something has changed from either a small sample or just casual observance. RNG can paint a very skewed picture in both situations.

SgtWinnie wrote: »

I do love tm's spreadsheets
I do more writs per day than most and I also haven't noticed a lower drop rate as of yet.
Hopefully it's just a bad month for you mate and it will pick up soon.

Sephyr wrote: »

Hotdog_23 wrote: »

All I can is thanks to tmbrinks for his/her time and attention to tracking this valuable information and then publishing if for everyone to see.

RNG can be a cruel mistress but over time even she succumbs to gods of math and time and must obey their law of averages.

Be safe and have fun

My thoughts exactly. Without their data, I wouldn't know how to begin checking on my own and while I haven't done nearly as many as they have - my short-term data is pretty on par with a few slight deviations (at most my monthly discrepancy is ranging around 3-7 since I've been keeping records which isn't all that long) and some of that can be attributed to maybe skipping a character and I don't realize to just the low luck of the draw. But I've not experienced any kind of significant drop unless, you know... I'm not doing them.

Thank you all. That's why I keep the data. I like knowing what they should be, and being a math and science teacher myself I teach all my students to use data and information to prove their hypothesis. I know I do enough writs and have enough data that it would be statistically relevant. And I just want to share it with everybody, so they can make their own conclusions.

Sephyr · November 2020

tmbrinks wrote: »

Hotdog_23 wrote: »

All I can is thanks to tmbrinks for his/her time and attention to tracking this valuable information and then publishing if for everyone to see.

RNG can be a cruel mistress but over time even she succumbs to gods of math and time and must obey their law of averages.

Be safe and have fun

idk wrote: »

tmbrinks wrote: »

I only track on a per-patch basis, but updating on a weekly-ish basis. Have noticed no significant changes in expected drop rate from the 36 characters I do writs on most days.

Data:

https://docs.google.com/spreadsheets/d/1yQDUjYNQVsIFl0ktkbkSlfYPkzP6pgCOTMQ-qCuzfaI/edit#gid=1039171516

If @tmbrinks says there is no indication of a change then there is no indication of a change. Afaik, they are the only person active in the forum that actually tracks this type of information and does it across many characters.

It is easy to think something has changed from either a small sample or just casual observance. RNG can paint a very skewed picture in both situations.

SgtWinnie wrote: »

I do love tm's spreadsheets
I do more writs per day than most and I also haven't noticed a lower drop rate as of yet.
Hopefully it's just a bad month for you mate and it will pick up soon.

Sephyr wrote: »

Hotdog_23 wrote: »

All I can is thanks to tmbrinks for his/her time and attention to tracking this valuable information and then publishing if for everyone to see.

RNG can be a cruel mistress but over time even she succumbs to gods of math and time and must obey their law of averages.

Be safe and have fun

My thoughts exactly. Without their data, I wouldn't know how to begin checking on my own and while I haven't done nearly as many as they have - my short-term data is pretty on par with a few slight deviations (at most my monthly discrepancy is ranging around 3-7 since I've been keeping records which isn't all that long) and some of that can be attributed to maybe skipping a character and I don't realize to just the low luck of the draw. But I've not experienced any kind of significant drop unless, you know... I'm not doing them.

Thank you all. That's why I keep the data. I like knowing what they should be, and being a math and science teacher myself I teach all my students to use data and information to prove their hypothesis. I know I do enough writs and have enough data that it would be statistically relevant. And I just want to share it with everybody, so they can make their own conclusions.

That and you just put so much information in your posts too. I'd never have known about the alchemy survey thing, running back and forth to get the materials I want. For someone who's kept records for so long, I know you're not arguing opinions here. You're arguing the data that you've kept for literal years on several accounts. I learn a lot even with something as so simple as just record keeping. So seriously, thanks for all of this!

Trinket61 · November 2020

tmbrinks wrote: »

Being a math and science teacher myself

tmbrinks wrote: »

BTW, I teach high school mathematics and statistics

To be absolutely clear it is not my intent to bait or mock you or violate the ZoS Terms. I am merely seeking clarification and guidance.

using the universal formula [z2 * p(1-p)] / e2 / 1 + [z2 * p(1-p)] / e2 * N] what would the standard sample size of your data be?

I would like to look through your data, but 781+ days of data from 32 characters x 6 writs or a sample size of 150,000 is overwhelming. I can barely squeeze in the time doing the writs , i cant imagine the diligence it takes to enter 192 sets of data every day for 2+ years!!

tmbrinks wrote: »

They are intricately related.

Is this a mathematical principle?

Thanks in advance for your answers

tmbrinks · November 2020

Trinket61 wrote: »

tmbrinks wrote: »

Being a math and science teacher myself

tmbrinks wrote: »

BTW, I teach high school mathematics and statistics

To be absolutely clear it is not my intent to bait or mock you or violate the ZoS Terms. I am merely seeking clarification and guidance.

using the universal formula [z2 * p(1-p)] / e2 / 1 + [z2 * p(1-p)] / e2 * N] what would the standard sample size of your data be?

I would like to look through your data, but 781+ days of data from 32 characters x 6 writs or a sample size of 150,000 is overwhelming. I can barely squeeze in the time doing the writs , i cant imagine the diligence it takes to enter 192 sets of data every day for 2+ years!!

tmbrinks wrote: »

They are intricately related.

Is this a mathematical principle?

Thanks in advance for your answers

1. sample size is used in getting information from a random sample. Thus, your standard sample size would be adequate. ~500 "trials" for about a 5% error, ~1000 "trials" for about ~3% error. This is of EACH INDIVIDUAL TRIAL. So you'd have to do 500 alchemy writs, to have an idea within about 3% of the drop rate. You aren't getting that in a few weeks

(yes, there are some more "fine" adjustments we can make to this, but I'm not going to argue any more complex statistical theories about it. It doesn't need to be overthought). Long story short, you need sample sizes in the high hundreds to make any sort of relevant conclusions about if drop rates have changed... and even then a rate of between 9.5%-15.5% (3% margin of error on either side of the expected drop rate of 12.5% is a plausible outcome)

2. Add-ons track the data. Dolgubon's does this nicely. I'm just compiling it, analyzing it, tracking it from patch to patch to look for trends. Once I have the spreadsheet set up, I spend literally about 5 minutes, once a week, to update it. So it's not a huge time sink.

3. One begets the other. You can't get surveys without doing daily writs. That wasn't meant to be some "mathematical principle" in that statement. It was merely a mention that they can be (and need to be) talked about together. Varanis and I have both be long time contributors to the crafting forums... there's an understanding of what they mean when statements are made, and the particular language that is used when describing things, so there was no confusion in what he "meant" in my eyes.

VaranisArano · November 2020

tmbrinks wrote: »

Trinket61 wrote: »

tmbrinks wrote: »

Being a math and science teacher myself

tmbrinks wrote: »

BTW, I teach high school mathematics and statistics

To be absolutely clear it is not my intent to bait or mock you or violate the ZoS Terms. I am merely seeking clarification and guidance.

using the universal formula [z2 * p(1-p)] / e2 / 1 + [z2 * p(1-p)] / e2 * N] what would the standard sample size of your data be?

I would like to look through your data, but 781+ days of data from 32 characters x 6 writs or a sample size of 150,000 is overwhelming. I can barely squeeze in the time doing the writs , i cant imagine the diligence it takes to enter 192 sets of data every day for 2+ years!!

tmbrinks wrote: »

They are intricately related.

Is this a mathematical principle?

Thanks in advance for your answers

1. sample size is used in getting information from a random sample. Thus, your standard sample size would be adequate. ~500 "trials" for about a 5% error, ~1000 "trials" for about ~3% error. This is of EACH INDIVIDUAL TRIAL. So you'd have to do 500 alchemy writs, to have an idea within about 3% of the drop rate. You aren't getting that in a few weeks (yes, there are some more "fine" adjustments we can make to this, but I'm not going to argue any more complex statistical theories about it. It doesn't need to be overthought). Long story short, you need sample sizes in the high hundreds to make any sort of relevant conclusions about if drop rates have changed... and even then a rate of between 9.5%-15.5% (3% margin of error on either side of the expected drop rate of 12.5% is a plausible outcome)

2. Add-ons track the data. Dolgubon's does this nicely. I'm just compiling it, analyzing it, tracking it from patch to patch to look for trends. Once I have the spreadsheet set up, I spend literally about 5 minutes, once a week, to update it. So it's not a huge time sink.

3. One begets the other. You can't get surveys without doing daily writs. That wasn't meant to be some "mathematical principle" in that statement. It was merely a mention that they can be (and need to be) talked about together. Varanis and I have both be long time contributors to the crafting forums... there's an understanding of what they mean when statements are made, and the particular language that is used when describing things, so there was no confusion in what he "meant" in my eyes.

For what its worth, I've edited the post to specifically address surveys, since they are correct that surveys are the topic, as opposed to any other crafting writ reward.

But you are entirely correct, my post was more of a general gripe about the way people talk about crafting writ rewards, whether surveys or master writs.

Whether we're talking about the drop chance of surveys, master writs, improvement mats, etc. people post threads saying "I've noticed I've been getting less of X" but posting little to no data.

As I said in that post:

"We humans are primed to remember bad outcomes.
What we "notice" isn't likely to be very useful when figuring out if we're getting less surveys.
The plural of anecdote is not data.

We really do need the data to evaluate the merit of the complaint."