These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

EVE General Discussion

 
  • Topic is locked indefinitely.
12Next page
 

Stealth|Unintended Nerf: NPC Commander/Officer Rate

Author
Vraygan
#1 - 2017-06-15 13:35:35 UTC  |  Edited by: Vraygan
Since the patch, there has been a drastic difference in the number of Rogue Drone Sentient spawns when doing combat sites. My corpmates and I are getting these spawns at least 1/5th to 1/10th as often, compared to before the patch. I don't make this claim lightly. Combined my corp has accumulated over 100 hours of data since the patch. Last month we ran over 1,000 Drone Patrols.

Is this intended?

When each ratter finishes a site, they make a corp BM (with their name-Time) on any Sentient wreck so that it can be looted while they move to the next site. While warping to the next site they add the BM name to our shared spreadsheet, with change tracking, so that us looters can't "forget" to log their BMs. They have been in the habit of making BMs for Sentients for over a month now. (And after the patch they are REALLY motivated to remember to BM them because they've noticed the lack of spawns.) Us looter/salvagers (work at home with personal monitor on Eve) collect the loot, verifying that the ratter isn't padding their numbers with false BMs, and make a portion of the profits.

Same system, same maxed out military index, similar enough ship setups that all make 40-60 mil per hour.

I have access to the API of the people ratting and have looked at their ratting ticks. Sentients give an extra 8-11mil in a single tick, and are pretty easy to find. Their ticks show that they are doing the same amount of work, but getting 1/5th (or worse) as many Sentient spawns.

Yesterday there was 3-4 ratters on at a time for 10 hours, and we didn't get a single Sentient. Last month our lowest Sentient count under the same circumstances is 4, going as high as 10, and we hit that scenario more than half of the days last month...

CCP_Fozzie wrote:
In June we are starting to address the supply of pirate battleship BPCs, with moderate changes to the chance of escalations spawning from the most popular high-end anomalies and larger changes to the drop rates of the BPCs themselves from quite a few NPCs.

I would wager 5Bil that Rogue Drone Sentient spawns are considered escalations, or are some how affected by the patch changes (but CCP_Fozzie didn't know that they were connected in such a way). Seems that moderate means 20%, and the developer inverted the intended chance.
Lan Wang
Knights of the Posing Meat
Snuffed Out
#2 - 2017-06-15 14:12:15 UTC
stealth? they said they were nerfing the spawn rate to combat the pirate battleship bpc drops

Domination Nephilim - Angel Cartel

Calm down miner. As you pointed out, people think they can get away with stuff they would not in rl... Like for example illegal mining... - Ima Wreckyou*

Jenn aSide
Shinigami Miners
Already Replaced.
#3 - 2017-06-15 14:22:41 UTC
Lan Wang wrote:
stealth? they said they were nerfing the spawn rate to combat the pirate battleship bpc drops


They said drop rate of bpcs, not of Commander (faction) spawns themselves.

To the OP, 100 hours is not long enough to tell anything. Random really is random and it's impossible to quantify 'chances' even though that's what people always try to do. Every week in the missions and complexes forum there is a post about how "something must have gotten nerfed because I'm not getting the same amount of X as I'm used to.

I had a corp mate say the same thing last night ie "they must have nerfed more than they said because I'm not getting faction spawns anymore. We live in the north (Guristas space) and i've had 3 Dread Guristas in the same time period (2 with just ammo and a tag, 1 with a valuable mod).

I recently went 2 weeks (even before this past patch) without a single faction spawn and only 2 escalations. This last monday night I got a triple (Faction spawn + dreadnaught spawn + escalation) and 2 doubles (faction spawn + escalation) in the same 2 hour span.

Give it a couple weeks and see how it goes, it's not impossible that something got messed up, but again you haven't had enough time to tell even if it has.
Spurty
#4 - 2017-06-15 14:32:25 UTC
I observed someone in Corp asking if the spawn rates had been increased.

Seems he was getting them [True Sansha] 400% more often.

Seems "Randomness is spikey"

There are good ships,

And wood ships,

And ships that sail the sea

But the best ships are Spaceships

Built by CCP

Vraygan
#5 - 2017-06-15 14:43:23 UTC
There is a big difference between item drop rate and npc spawn rate...

100 hours of dedicated ratting time is enough to indicate there's probably something is probably different. Before the patch, 100 hours meant 80-120 Sentient spawns. After the patch, we've gotten 12 total. Since I'm the dedicated looter, and my income depends on sentient/elite spawns, I went so far as to combat probe our system for over 6 hours, looking for people not reporting spawns. They aren't spawning... 100 is valid starting sample size. And by this point the rate should at least be creeping towards the mean, which it isn't. This new average is stable...
Tuttomenui II
Aliastra
Gallente Federation
#6 - 2017-06-15 15:18:37 UTC
Vraygan wrote:
There is a big difference between item drop rate and npc spawn rate...

100 hours of dedicated ratting time is enough to indicate there's probably something is probably different. Before the patch, 100 hours meant 80-120 Sentient spawns. After the patch, we've gotten 12 total. Since I'm the dedicated looter, and my income depends on sentient/elite spawns, I went so far as to combat probe our system for over 6 hours, looking for people not reporting spawns. They aren't spawning... 100 is valid starting sample size. And by this point the rate should at least be creeping towards the mean, which it isn't. This new average is stable...


Maybe if you have 1,000 people with 100 hrs each. Don't you know how to Science? You need controls and multiple data points.
Vraygan
#7 - 2017-06-15 19:42:00 UTC  |  Edited by: Vraygan
Tuttomenui II wrote:
Maybe if you have 1,000 people with 100 hrs each. Don't you know how to Science? You need controls and multiple data points.

Sounds like you can't math. You want 11.4 years of data? LOL!

Since me and one other salvager track spawns and who gets them for our corp, so we can divvy it up fairly, i know we had 300 hours of data with an average of 4 people ratting 2 patrols or hordes per hour. The average Sentient per hour per site runner was around 0.2 with a 95% confidence.

The current average Sentient per hour per site runner is 0.075 and over 90% outside of the previous standard deviation. The new sample is 10% of data volume from last month.

The new sample meets the minimum standard for​ statistical significance.
Ebony Texas
Un4seen Development
Sev3rance
#8 - 2017-06-15 20:07:29 UTC
don't you guys understand when fozzie and ccp nerf something they don't and never control how hard the nerf is.. so of course it will impact everything nearby... including your precious officer spawns.


its working as intended.. blame those farmers that caused it.. then again its not their fault.. they're playing the fawking game the way we all should but ccp cried foul cause they were never needing to buy some stinking plex..


i'll be so damn happy when Karma comes back on CCP games.. so help me god!
Axhind
Eternity INC.
Goonswarm Federation
#9 - 2017-06-15 20:18:52 UTC
Vraygan wrote:
Tuttomenui II wrote:
Maybe if you have 1,000 people with 100 hrs each. Don't you know how to Science? You need controls and multiple data points.

Sounds like you can't math. You want 11.4 years of data? LOL!

Since me and one other salvager track spawns and who gets them for our corp, so we can divvy it up fairly, i know we had 300 hours of data with an average of 4 people ratting 2 patrols or hordes per hour. The average Sentient per hour per site runner was around 0.2 with a 95% confidence.

The current average Sentient per hour per site runner is 0.075 and over 90% outside of the previous standard deviation. The new sample is 10% of data volume from last month.

The new sample meets the minimum standard for​ statistical significance.


Minimum standard doesn't actually imply statistical significance and all those tests are based on certain assumptions. Even before the changes you could go for weeks without anything and then get 3 factions spawns and escalations in one evening. Your numbers are tiny and so is their significance especially since you are dealing with a competent RNG and not some kind of balanced distribution algorithm that gives you cake after x amount of effort.
Vraygan
#10 - 2017-06-15 20:44:08 UTC
Axhind wrote:
Vraygan wrote:
Tuttomenui II wrote:
Maybe if you have 1,000 people with 100 hrs each. Don't you know how to Science? You need controls and multiple data points.

Sounds like you can't math. You want 11.4 years of data? LOL!

Since me and one other salvager track spawns and who gets them for our corp, so we can divvy it up fairly, i know we had 300 hours of data with an average of 4 people ratting 2 patrols or hordes per hour. The average Sentient per hour per site runner was around 0.2 with a 95% confidence.

The current average Sentient per hour per site runner is 0.075 and over 90% outside of the previous standard deviation. The new sample is 10% of data volume from last month.

The new sample meets the minimum standard for​ statistical significance.


Minimum standard doesn't actually imply statistical significance and all those tests are based on certain assumptions. Even before the changes you could go for weeks without anything and then get 3 factions spawns and escalations in one evening. Your numbers are tiny and so is their significance especially since you are dealing with a competent RNG and not some kind of balanced distribution algorithm that gives you cake after x amount of effort.

Incorrect. I'm consolidating data from many people. We're talking about 4 characters ratting each hour, each running 2 of the same type of sites per hour. That is 264 sites run since patch. I'm not talking about one person's experience of RNG. This is from 20 or so people. The sample size is sufficient.
Axhind
Eternity INC.
Goonswarm Federation
#11 - 2017-06-15 20:48:50 UTC
Vraygan wrote:
Axhind wrote:
Vraygan wrote:
Tuttomenui II wrote:
Maybe if you have 1,000 people with 100 hrs each. Don't you know how to Science? You need controls and multiple data points.

Sounds like you can't math. You want 11.4 years of data? LOL!

Since me and one other salvager track spawns and who gets them for our corp, so we can divvy it up fairly, i know we had 300 hours of data with an average of 4 people ratting 2 patrols or hordes per hour. The average Sentient per hour per site runner was around 0.2 with a 95% confidence.

The current average Sentient per hour per site runner is 0.075 and over 90% outside of the previous standard deviation. The new sample is 10% of data volume from last month.

The new sample meets the minimum standard for​ statistical significance.


Minimum standard doesn't actually imply statistical significance and all those tests are based on certain assumptions. Even before the changes you could go for weeks without anything and then get 3 factions spawns and escalations in one evening. Your numbers are tiny and so is their significance especially since you are dealing with a competent RNG and not some kind of balanced distribution algorithm that gives you cake after x amount of effort.

Incorrect. I'm consolidating data from many people. We're talking about 4 characters ratting each hour, each running 2 of the same type of sites per hour. That is 264 sites run since patch. I'm not talking about one person's experience of RNG. This is from 20 or so people. The sample size is sufficient.


No it's not as you are assuming a certain distribution of results which is wrong thing to assume with an RNG that has results 0 and 1 (essentially at least) with very heavy skew towards 0. Thus likelihood of seeing just zeros is very high since it is a proper RNG and not balanced distribution algorithm. The more random your output the more data you need. In this case you will need a lot more to be able to detect changes in the drop rates as those are very low to begin with and you need huge numbers in order to get reasonable statistical resolution. Even then you can just talk about certain level of confidence as you are dealing with an RNG.
Vraygan
#12 - 2017-06-15 22:12:21 UTC  |  Edited by: Vraygan
Now about 90% of all data points are zero. Before, over 70% were 1+ (up to 5).

If you want to be taken seriously, why don't you show how to calculate the minimum data points to achieve statistical​ significance. Original sample was 300 hours with 4 people doing the same 2 types of sites as before pdr Hour All together that is 2400 data points (300*4*2). Here's a tip, the efficacy of the RNG is the same, so given the sample size you calculate, the RNG factors cancel out.
Gimme Sake
State War Academy
Caldari State
#13 - 2017-06-15 22:20:16 UTC
RNG is R. Pretty sure any change, besides bpc drops, would have been announced. No reason for a "stealth nerf".

"Never not blob!" ~ Plato

Nevyn Auscent
Broke Sauce
#14 - 2017-06-15 22:20:54 UTC
Right now, approx 25,000 person hours are being done in EVE, this very hour.
Your 300 is utterly insignificant, especially since you have a sample size of 4 out of that 25,000. Or 0.016%.
You can not claim that is representative of EVE.
Vraygan
#15 - 2017-06-15 22:38:32 UTC
Imbicil, when a not so random number (1-100) is multiplied by another number (spawn chance was around 15%), the results are not random. They are quite predictable, especially over a 2 day period window with over 300 points.

I thought eve players were supposed to be smarter than you dunces.
Ebony Texas
Un4seen Development
Sev3rance
#16 - 2017-06-15 22:47:32 UTC
Vraygan wrote:
Imbicil, when a not so random number (1-100) is multiplied by another number (spawn chance was around 15%), the results are not random. They are quite predictable, especially over a 2 day period window with over 300 points.

I thought eve players were supposed to be smarter than you dunces.



they're just trolling you with their nerf-jizzism due to fact you mentioned officer spawns, which not many are able to get to systems which spawn those..

you're part of the 1% ccp targeted since they're so delusional about null sec..

officer spawns should be opened to all of null-sec.. screw this tru-sec bullshit.
Sabriz Adoudel
Move along there is nothing here
#17 - 2017-06-15 23:22:37 UTC
Math degree here.

This is statistically significant, but not proof beyond reasonable doubt.


Have you investigated for any systemic biases in your data gathering? Anything you might be doing differently post patch?

I support the New Order and CODE. alliance. www.minerbumping.com

Vraygan
#18 - 2017-06-16 00:28:27 UTC  |  Edited by: Vraygan
Sabriz Adoudel wrote:
Math degree here.

This is statistically significant, but not proof beyond reasonable doubt.


Have you investigated for any systemic biases in your data gathering? Anything you might be doing differently post patch?

I totally agree that it isn't undeniable proof. Here's the scenario.
Let me know if there are any biases that you can see.

When each ratter finishes a site, they make a corp BM (with their name-Time) on any Sentient wreck so that it can be looted while they move to the next site. While warping to the next site they add the BM name to our shared spreadsheet, with change tracking, so that us looters can't "forget" to log their BMs. They have been in the habit of making BMs for Sentients for over a month now. (And after the patch they are REALLY motivated to remember to BM them because they've noticed the lack of spawns.) Us looter/salvagers (work at home with personal monitor on Eve) collect the loot, verifying that the ratter isn't padding their numbers with false BMs, and make a portion of the profits.

Same system, same maxed out military index, similar enough ship setups that all make 40-60 mil per hour.

I have access to the API of the people ratting and have looked at their ratting ticks. Sentients give an extra 8-11mil in a single tick, and are pretty easy to find. Their ticks show that they are doing the same amount of work, but getting 1/5th (or worse) as many Sentient spawns.

Yesterday there was 3-4 ratters on at a time for 10 hours, and we didn't get a single Sentient. Last month our lowest Sentient count under the same circumstances is 4, going as high as 10, and we hit that scenario more than half of the days last month...
Vraygan
#19 - 2017-06-16 01:44:51 UTC  |  Edited by: Vraygan
CCP_Fozzie wrote:
In June we are starting to address the supply of pirate battleship BPCs, with moderate changes to the chance of escalations spawning from the most popular high-end anomalies and larger changes to the drop rates of the BPCs themselves from quite a few NPCs.

I would wager 5Bil that Rogue Drone Sentient spawns are considered escalations, or are some how affected by the patch changes (but CCP_Fozzie didn't know that they were connected in such a way). Seems that moderate means 20%, and the developer inverted the intended chance.
Rain6637
NulzSec
#20 - 2017-06-16 07:04:12 UTC
Vraygan wrote:
Imbicil, when a not so random number (1-100) is multiplied by another number (spawn chance was around 15%), the results are not random. They are quite predictable, especially over a 2 day period window with over 300 points.

I thought eve players were supposed to be smarter than you dunces.

I read this in the voice of Ren from Ren & Stimpy "YOU EEEEMBACILE YOU BLOATED STUPID FATHEAD YOU"
12Next page