Notes on new team difficulty settings
By
Abigail Frost,
Posted on: 2009-08-26 15:02 in The Market and Inventions
Quote:
I can compare my spines/fire on one account with my warshade on the other running the same arc in Ouroboros. I'll set them both to the horribly boring +0/1 hero setting. The only drawback is both are on East Coast servers (Pinnacle and Triumph) if that would have any impact on it.
|
Originally Posted by Canis_Lupus
This has been tested a few times, and the drop rates that are being seen a generally MUCH higher than this theory would allow for (getting in the 20-70% range of expected drops, if run at /8 equivalent and this was the case, it would be stuck in the 16.7% range).
One other question, given that the same players/characters seem to have problems consistently. Is the Character name and/or Global name used as part of the RNG seed through some sort of hash function? Similarly for the client IP address or computer serial number, client specific checksum or something? This is something that only Synapse and friends can answer, but given the behavior that we're seeing, I would think it's something to be verified. |
total kick to the gut
This is like having Ra's Al Ghul show up at your birthday party.
Quote:
|
Originally Posted by EarthWyrm
Same. On the off chance that something had changed on test, I hopped back on and used the same Abyss mission I'd given Synapse to test. He ran it 15 times and got 14-19 recipe drops each time. I've run it over 20 times on a single character and never gotten above 6. My average is three.
In 344 kills, I again got 3 recipes. I'm sorry he can't reproduce it. It's real and it's consistent, but not across characters/accounts in a way that we've been able to isolate the common factor. |
Quote:
|
Originally Posted by SwellGuy
I can compare my spines/fire on one account with my warshade on the other running the same arc in Ouroboros. I'll set them both to the horribly boring +0/1 hero setting. The only drawback is both are on East Coast servers (Pinnacle and Triumph) if that would have any impact on it.
|
By the way I got 2 recipes during that run, both level 50 Fly
Who do I have to *&^% around here to get more Targeted AoE recipes added?
Arc Name: Tsoo In Love
Arc ID: 413575
Quote:
Your theory interests me. Over the weekend I ran some sweeps on The Wall and in RWZ and experienced the same pathetic drop rate I'd been getting solo and indoors. I'm on Victory, so it's an East Coast server. (Unfortunately, I'd accidentally shut down Hero Stats, so I didn't receive hard numbers to do the math with.) During that span I received about 50 salvage and only 5 recipes. Based on the math of the rates of drops for both, this is a heavy disparity. (It should be about 3 salvage for every 1 recipe, not 10 to 1.)|
Originally Posted by Starcloud
Conditions on Live are different, the random number generator is hit far more often, and there might be a contributing factor in the lag and occasional loss of sync on Live that are not possible on your internal test server.
|
I'll see if I can do some testing on a West Coast server this week.
Arc# 92382 -- "The S.P.I.D.E.R. and the Tyrant" -- Ninjas! Robots! Praetorians! It's totally epic! Play it now!
Arc # 316340 -- "Husk" -- Azuria loses something, a young woman harbors a dark secret, and the fate of the world is in your hands.
I don't really want to do any work myself but I've got lots of fillers and I'll be semi around the computer all day. If you want to test #s with real players on your team send me a tell @Smurphy
I may be slow to respond but CoH is up and my computer is on.
Sacrificing a little sleep, I ran the first Abyss mish during the 6am eastern hour this morning just to test the "lag equals crappy drops theory".
I didn't perform a complete run, but being somewhat familiar with the length of the mission, I estimated that I completed about 40-60% of a run.
Set for -1/x8, my 50 Brute received 2 common recipe drops.
I didn't not account for salvage drops.
Like I said upthread, I'll test this map out plenty more this week.
Edit: This is NOT intended to be data. Maybe just a little piece of the same ol' nugget.
Repeat Offenders
Quote:
This is something definitely worth looking at, I would think. I'm going to try a slight variation, using the same character running the same missions over and over, but doing this on different computers running different OSes, even. Unfortunately, I won't be able to properly test any potential IP related issues, but if there's something about the HW/OS causing it, maybe that can help track it down.
|
Originally Posted by SwellGuy
I can compare my spines/fire on one account with my warshade on the other running the same arc in Ouroboros. I'll set them both to the horribly boring +0/1 hero setting. The only drawback is both are on East Coast servers (Pinnacle and Triumph) if that would have any impact on it.
|
Quote:
If this keys off our HW/OS then things could be really strange. I had always assumed it would be server side to prevent any manipulating by a player.
|
Originally Posted by Canis_Lupus
This is something definitely worth looking at, I would think. I'm going to try a slight variation, using the same character running the same missions over and over, but doing this on different computers running different OSes, even. Unfortunately, I won't be able to properly test any potential IP related issues, but if there's something about the HW/OS causing it, maybe that can help track it down.
|
total kick to the gut
This is like having Ra's Al Ghul show up at your birthday party.
Quote:
There's zero chance any drop determination is done by the client.
|
Originally Posted by SwellGuy
If this keys off our HW/OS then things could be really strange. I had always assumed it would be server side to prevent any manipulating by a player.
|
Freedom: Blazing Larb, Fiery Fulcrum, Sardan Reborn, Arctic-Frenzy, Wasabi Sam, Mr Smashtastic.
Quote:
If someone could wander over to Test, do a few runs, then change their global name and do the exact same few runs, that would at least rule out some sort of weird global-name seed as part of the bug.|
Originally Posted by SwellGuy
If this keys off our HW/OS then things could be really strange. I had always assumed it would be server side to prevent any manipulating by a player.
|
Although the account-based testing above will probably have the same type of information, unless the globals on those two accounts are highly similar.
My postings to this forum are not to be used as data in any research study without my express written consent.
This morning I ran 1st mish from Abyss (game contact, no Ouro) 3 times. +0/6x/yes/no (as all my testing has been). It isn't quite enough kills as I would like for more reliable numbers, but 735 min equivalents is pretty good. The results on recipe drops were uniformly abysmal (Archie's report at the bottom).
Just some hunches and guesses, I don't think it is map related. I do think it is going to be tricky to track down. I believe the "drop code" hasn't changed, but obviously (to me, and many here any way), the amount of recipes Toons are getting certainly has changed.
Other code has changed, it is perhaps interfering with the rewards related delivery, recipe only, in an unanticipated way. There also may be some interaction between the live database, the server environment there, and a host of other factors that are not the drop code itself. But the end result is toons aren't getting rewards.
It'll take some dedicated research to find and fix. I do hope some progress happens soon-ish.
/w
Code:
Type DropRate Drops Expected Mobs -------------------------------------------------------- Pool A 1.22% 9 12-29 735 minion equivalents Salvage 8.78% 57 39-66 649 minion equivalents -------------------------------------------------------- (Table excludes mobs of unknown rank and their drops.) (1 in 20 runs will be outside the expected range.) 547 rewarding mobs defeated 29 bosses and elite bosses 134 lieutenants 380 minions 4 underlings (XP only) 0 pets and 7 greys
I rarely farm and have not been doing any scientific studies, but one striking trend that I have noticed since i16 is an extreme predominance of common recipe drops over uncommons and rares. I've been on two ITF's where I failed to get any recipes other than commons... something that previously was a very rare (and unpleasant) occurrence. On the second run my entire (6-man) team received a grand total of three non-common recipes.
Last night's session, however, really got my attention. I played 12 missions in the RWZ with teams of 6-8 players (real, not virtual)... the entire Serpent Drummer arc and a number of Borea's repeatable missions. That, along with half an hour of random street sweeping, netted me ZERO non-common recipes...
Since results seem to vary so much between different players (and devs), has anyone speculated whether the issue is character based? Could the RNG be seeded somehow with data specific to an individual character? The most notorious example of such a glitch is the "Wi Flag" from Asheron's Call, although that involved the RNG for NPC aggro, not item drops. ( For more details see: http://asheron.info/~asheron/index.php/Wi_Flag )
You'd think the CoH devs would have avoided a "rookie" mistake from a 1st-gen game... and that if not, that type of error would have already caused obvious issues in the past... but we seem to be running out of other explanations for this problem...
Picture Jack Emmert in front of a mirror on patch day, holding a tinfoil Goa'uld
hand-device and shouting, "You dare question ME?! Kneel before your God!"
-An anonymous GDN victim
Quote:
I will definitely admit that any RNG that relies on client-side information as a seed would be hideously broken, and I truly do not expect to find that here. |
Originally Posted by SwellGuy
If this keys off our HW/OS then things could be really strange. I had always assumed it would be server side to prevent any manipulating by a player.
|
However, a basic step in troubleshooting any sort of issue is "What's different?" between the tests where it worked and the tests where it doesn't. This may be a change if the case is a "it used to work, now it doesn't" or other differences in the case of "it works for me, but not for my neighbor"
For us, we have both issues. The first part is relatively simple to explain. What's different is Issue 16 was added. Of course, there is a much finer degree of detail that would be required on the programming side, but for our purposes, that's enough.
Moving past that, we have a scenario where one group of people (those posting in this thread, for example) are experiencing a problem and another (Synapse and company) are seeing different behavior. For this, we need to examine what the difference between his tests and our tests is. In addition to the obvious Map instance, enemies fought, Character AT, etc (but I believe the existing testing seems to have ruled these out), the options I've come up with so far (and I'm willing to be informed there are more that need to be tested for) are:
Global chat handle
Character (edited to be character, not character name, to include behind the scenes flags/etc)
Time of day
Server
Hardware (CPU SN, etc)
OS
Original Install date/time of product
Client IP address
Again, I truly don't expect that this will be traced to any of the client-side options listed here, but they should be tested for.
Given the veritable mountain of hard data that's been compiled along the way, I must say I'm incredibly disappointed in this being dismissed as a "perception" issue. While I was willing to admit early on that it may have been perception, the raw data that people here have collected (and I'll be doing some major data collection myself starting Thursday night) indicates that this is NOT a perception issue, but rather a very odd behavior of the system, likely one that is a major benefit to some players/characters and a detriment to others.
I definitely hope one of more of the Devs will take up the offers to join in on teams on live servers with people here that have indicated problems, so that things can be observed directly as they occur, both client-side and (presumably) server-side. I would think that running this as Duos would be best, making it much easier to track, yet still have that direct involvement. I think that most everyone that's been posting here is much more interested in helping them fix a problem than simply placing blame, and I know I would definitely be willing to assist on that myself to whatever extent I can.
First, thanks Synapse for posting in here.
But as you said, that's not what we wanted to hear, because that's not what we're seeing. If I gather all my data, here's what I have:
10436 kills: 7727 minions, 2709 lieuts. It represent 27 runs of the mission Close Dimensionnal Ruptures from Unai Kemen (Council Empire). I tracked runs randomly, I had some decent ones, some are even over my expectations, but if you look at the data:
Expected recipes drop: 346,5
Effective recipes drop: 247 = 71.3% of expected
Again and again and again: 70%. ALWAYS 70%.
Can you call this a "perception issue"? Most of this tests were done on test server, when I16 was still in beta. I had no other interest than finding if they was, or not, a bug. As I said, this are random runs, but when it comes to 10 thousands + kills, I don't see how it could be a "perception issue".
We tried almost everything we could. We killed mobs, we saw a problem. Then we asked "Where could it come from?". Many great people came in here and said "It could be that... or that", and others made tests... and at the end, we have no theory of "Where could it come from?", but we sure know one thing at this point: there is a bug.
I can't imagine devs doing a "stealth nerf", when you "nerf" something, you always accept it. I also understand that you whole job can't be "finding the droprate bug that those people in this thread are talking about all day long", but please, you read the topic, you saw our data... I mean, we don't do those reports just for fun, if droprates were correct, we'd be happy farming and/or playing as usual, but we all saw one thing: droprates aren't correct.
So please, don't tell us that it's just a "perception issue". What do we have to do, now, to prove you that there's a bug? Find the pattern of the bug? We just spent weeks trying to do that, with nothing conlusive. I know that you don't have just that to do, but saying "I tried, I had good result, PM me if you find something more relevant than what you've already got" is kind of the easy way out. I'm sure you have plenty other thing on your desk that are waiting to be done, but... please, don't let us down now (well, if you're still reading what's going on around here).
Quote:
Swami: Very good point. I was thinking of this example previously, but I couldn't recall which game it was in, so I never mentioned it. Thanks for dropping this back in.|
Originally Posted by Swami_Monsoon
Since results seem to vary so much between different players (and devs), has anyone speculated whether the issue is character based? Could the RNG be seeded somehow with data specific to an individual character? The most notorious example of such a glitch is the "Wi Flag" from Asheron's Call, although that involved the RNG for NPC aggro, not item drops. ( For more details see: http://asheron.info/~asheron/index.php/Wi_Flag )
You'd think the CoH devs would have avoided a "rookie" mistake from a 1st-gen game... and that if not, that type of error would have already caused obvious issues in the past... but we seem to be running out of other explanations for this problem... |
And perhaps my list needs to be updated to "Character" as opposed to character name.... some behind the scenes flag like the Wi fiag.
Current cumulative stats below from 4 runs between 9/20 and now. Note that the Expedient Reinforcement is unknown because it doesn't end with (Recipe), so I actually got 8 Recipe drops. But both that and the Mako's Bite are Pool C/Gold drops that I got from bosses, so they shouldn't count. My Pool A results are therefore 6 recipes in 1708 minion equivalents, or 0.35%. That's about 1/8 the expected rate of 2.66%.
Test server, "Defeat all Clockwork at Factory" (Board Train), the regular mission from Unai Kemen, no Flashback, no teammates, Difficulty set at +1/8/yes/no, done by Clocker, a level 50 Stone/Stone Tanker.
Code:
----------------------------------
Archie Gremlin's Drop Stats v0.3.1
http://www.glasspaw.com/dropstats/
----------------------------------
Type DropRate Drops Expected Mobs
--------------------------------------------------------
Pool A 0.41% 7 34-60 1708 minion equivalents
Salvage 8.30% 127 103-145 1530 minion equivalents
--------------------------------------------------------
(Table excludes mobs of unknown rank and their drops.)
(1 in 20 runs will be outside the expected range.)
1364 rewarding mobs defeated
129 bosses and elite bosses
289 lieutenants
743 minions
203 underlings (XP only)
1364 inf/debt rewards
8203655 inf
44000 debt
7 Pool A recipes (Mob Defeats)
3 uncommon and rare recipes
Blood Mandate: Dam/End (Recipe)
Enfeebled Operation: Acc/End (Recipe)
Mako's Bite: Acc/End/Rech (Recipe)
4 common recipes
127 salvage
2 rare
22 uncommon
103 common
11 single origin enhancements
414 inspirations (11 tier 3, 133 tier 2, 270 tier 1, 0 special)
1 other drops
Expedient Reinforcement: Resist Bonus Aura for Pets
Goodbye and thanks for all the fish.
I've moved on to Diablo 3, TopDoc-1304
Quote:
Fixt. |
Originally Posted by Synapse
Thought I would chime in here and let you know we didn't nerf accuracy again; we just nerfed to hit
|
Thanks, though, Synapse for what I think is the 4th redname post here* (and shortly after a RickRoll no less).
Good questions from the testers, too (are the PW drop rates correct, did you actually defeat stuff or use the I Win button, and what is your reaction to the data already collected). I hope we get to see you stop back with the answers.
*let's see if I remember them: Ex's attempt to smooth things over post, Posi's villains don't have as much money post, Mod8's ebil 'crapping' post, and now Synapse's I haz some data post.
President of the Arbiter Sands fan club. We will never forget.
An Etruscan Snood will nevermore be free
Give it a little time, Me thinks he is still testing stuff...
>.>
Ignoring anyone is a mistake. You might miss something viral to your cause.
I haven't personally tested the problem and I'll admit I haven't bothered look into it in too much detail but...
... if you had the difficulty set to +0/x0 and 5 real players on your team standing outside (or 3 real players if you want a 4 man spawn) would, in theory, the drop rate act as normal?
I ask because I can provide whoever wants the test the fillers necessary to do this test.
I know this is subjective and, thus, not worth much, but I did want to mention it.
I fired up a toon that hadn't been used since I16 came out, and she had two old missions.
One was from Gaussian's Red & Black arc, and I think the other was from On the Run.
I set her to +1/x3 to finish the arcs quickly.
On the Gaussian mission, I was surprised to see lots of recipes dropping. By now, I was sorta used to the diminished drops on several other toons, so the multiple recipe drops were very noticeable. It felt like what I would have expected from x3, definitely better than old-style solo drops.
I don't really remember drops from the On the Run mish I finished up. I did it before the Gaussian mish, and didn't notice anything unusual.
To the point, is there any chance that an 'old' mish, acquired before I16 could somehow have different properties? It doesn't sound likely to me, but thought I'd mention it.
Maybe I got a mish on that 'high end' of the wonky drop rates that was mentioned.
Quote:
Possibly, however some of us have questions as to how he is testing. I'd feel more comfortable if he said he was running tests on the live servers with normal toons. Rather than the internal and test servers.
|
Originally Posted by MrHassenpheffer
Give it a little time, Me thinks he is still testing stuff...
>.> |
Who do I have to *&^% around here to get more Targeted AoE recipes added?
Arc Name: Tsoo In Love
Arc ID: 413575
Quote:
Nope. I'm getting diminished drops on maps I acquired prior to issue 16. In fact, I acquired some of these maps with the express purpose of being able to farm them in issue 16 after the increased difficulty settings were announced.
|
Originally Posted by EbonyWall
To the point, is there any chance that an 'old' mish, acquired before I16 could somehow have different properties? It doesn't sound likely to me, but thought I'd mention it.
Maybe I got a mish on that 'high end' of the wonky drop rates that was mentioned. |
Quote:
Based on a PM he sent me yesterday, his 5 runs on the internal server all involved him going through the maps and manually defeating all the mobs. I don't know whether or not he used the [I Win Button] on his test server runs, but he definitely used the old-fashioned method of beating stuff up in at least 5 of his 15 runs.
|
Originally Posted by EnigmaBlack
Possibly, however some of us have questions as to how he is testing. I'd feel more comfortable if he said he was running tests on the live servers with normal toons. Rather than the internal and test servers.
|
My postings to this forum are not to be used as data in any research study without my express written consent.
Quote:
Thanks. |
Originally Posted by EarthWyrm
Based on a PM he sent me yesterday, his 5 runs on the internal server all involved him going through the maps and manually defeating all the mobs. I don't know whether or not he used the [I Win Button] on his test server runs, but he definitely used the old-fashioned method of beating stuff up in at least 5 of his 15 runs.
|
I take it the internal server is not the same as the ones we are experiencing these issues with is it? Such as Victory where I'm at (and the lag is killing me too!)
Who do I have to *&^% around here to get more Targeted AoE recipes added?
Arc Name: Tsoo In Love
Arc ID: 413575
Quote:
OK.... so, Synapse's testing was all done on either Training room or an internal only server, is this correct?
|
Originally Posted by EarthWyrm
Based on a PM he sent me yesterday, his 5 runs on the internal server all involved him going through the maps and manually defeating all the mobs. I don't know whether or not he used the [I Win Button] on his test server runs, but he definitely used the old-fashioned method of beating stuff up in at least 5 of his 15 runs.
|
Is it possible that the new difficulty settings add "virtual teammates", which are then assigned drops randomly, just like normal teammates? And since those "virtual teammates" don't actually exist, the drops are simply lost? That would explain the lower than expected and previously recorded averages for drop rates, as well as the variability from player to player.
Simple way to test would be for several people to team up at standard difficulty (+0/x0) and see what the drop rates average out to for each player on the team, then compare those results to what solo players are seeing for difficulty settings matching the team size and critter level. Example: team of 8 players facing +1 and +2 foes, then each team member tests solo with difficulty set at +1 or +2/x8.
One other question, given that the same players/characters seem to have problems consistently. Is the Character name and/or Global name used as part of the RNG seed through some sort of hash function? Similarly for the client IP address or computer serial number, client specific checksum or something? This is something that only Synapse and friends can answer, but given the behavior that we're seeing, I would think it's something to be verified.