Notes on new team difficulty settings
By
Abigail Frost,
Posted on: 2009-08-26 15:02 in The Market and Inventions
Quote:
I agree with your inner developer. Introducing a feature like this is likely to touch a lot of unforeseen bits of code. Hopefully the data being gathered can help them isolate the issue.
|
Originally Posted by _23X_
I agree,
The conspiracy theorist in me now wonders if virtual teams is a ploy to hamstring farmers. The general population will be less willing to anchor/fill/pad teams because that isn't needed any more. even tho the drop rate is much lower with virtual teaming. The developer in me thinks this is a bug no different that ticket drop bug when AE shipped (the one where not only were ticket drops split with the team but the number of tickets where split too ( ie: solo you would get say 10 tickets with 2 you would get 5, but only half the time)) |
My postings to this forum are not to be used as data in any research study without my express written consent.
Quote:
|
Originally Posted by EarthWyrm
I agree with your inner developer.
|
He is often the voice of reason, but not near as fun at parties.
Card Carrying DeFulmenstrator--Member Crazy 88s
We burn more Influence before 8am than you make all day.
OK, another 1040 kills over 20 runs at x1, 10 runs at x2.
Inspirations:
x1: 401 total, 0.3856 per kill, 2.59 kills per drop
x2: 336 total, 0.3231 per kill, 3.10 kills per drop
Common Salvage:
x1: 64 total, 0.0615 per kill, 16.25 kills per drop
x2: 64 total, 0.0615 per kill, 16.25 kills per drop
Uncommon Salvage:
x1: 17 total, 0.016 per kill, 61.18 kills per drop
x2: 22 total, 0.021 per kill, 47.27 kills per drop
Rare Salvage:
x1: 2 total, 0.002 per kill, 520 kills per drop
x2: 4 total, 0.004 per kill, 260 kills per drop
Common Recipes:
x1: 19 total, 0.018 per kill, 54.74 kills per drop
x2: 21 total, 0.020 per kill, 49.52 kills per drop
Uncommon recipes:
x1: 8 total, 0.008 per kill, 130 kills per drop
x2: 8 total, 0.008 per kill, 130 kills per drop
Rare Pool A:
x1: 0 total
x2: 1 total, 0.001 per kill, 1040 kills per drop
Enhancements:
x1: 6 total, 0.006 per kill, 173.33 kills per drop
x2: 6 total, 0.006 per kill, 173.33 kills per drop
0 purple drops, 0 costume drops. That makes 4160 kills on this map without a single purple.
Cumulatives, from previous post and this one: 2080 kills in each sample.
Inspirations:
x1: 740 total, 0.356 per kill, 2.81 kills per drop
x2: 701 total, 0.337 per kill, 2.97 kills per drop
Common Salvage:
x1: 125 total, 0.060 per kill, 16.64 kills per drop
x2: 124 total, 0.060 per kill, 16.77 kills per drop
Uncommon Salvage:
x1: 33 total, 0.016 per kill, 63.03 kills per drop
x2: 41 total, 0.020 per kill, 50.73 kills per drop
Rare Salvage:
x1: 5 total, 0.002 per kill, 416.00 kills per drop
x2: 8 total, 0.004 per kill, 260.00 kills per drop
Common Recipes:
x1: 53 total, 0.025 per kill, 39.25 kills per drop
x2: 40 total, 0.019 per kill, 52.00 kills per drop
Uncommon Pool A:
x1: 11 total, 0.005 per kill, 189.09 kills per drop
x2: 13 total, 0.006 per kill, 160.00 kills per drop
Rare Pool A:
x1: 2 total, 0.001 per kill, 1040 kills per drop
x2: 2 total, 0.001 per kill, 1040 kills per drop
Enhancements:
x1: 17 total, 0.008 per kill, 122.35 kills per drop
x2: 11 total, 0.005 per kill, 189.09 kills per drop
Looks like my hunch about virtual teammates didn't pan out. I don't know the math to find standard deviation and confidence intervals, but first glance says the numbers between x1 and x2, especially with the larger sample, are pretty close to each other.
119088 - Outcasts Overcharged. Heroic.
Just an amusing aside, after I finished logging my runs, I set it to -1/x6 for a run. minions and Lt's at level 48, probably close to 300 kills.
2 purples in one run... Murphy's Law strikes again.
119088 - Outcasts Overcharged. Heroic.
Quote:
It's late and I'm flying without a net (doing formulas from memory), so if I make a mistake, forgive me.|
Originally Posted by Lakanna
Looks like my hunch about virtual teammates didn't pan out. I don't know the math to find standard deviation and confidence intervals, but first glance says the numbers between x1 and x2, especially with the larger sample, are pretty close to each other.
|
It looks like an overall drop rate of around 2.9%. If these are all minions (you're on the cargo ship, yes?) with 2080 observations your margin of error is around .7%. So the x1 drop rate (3.17%?) and the x2 drop rate (2.64%?) both look like they're within margin of error for what we would expect of minions, and when compared to each other.
The most striking thing, to me, is that this really does come out to almost exactly what we would expect. Archie's speculation about lieutenants being a big piece of the problem may be correct. Does the drop rate on this map skew downward when the team size is raised and there are lieutenants in the spawns?
My postings to this forum are not to be used as data in any research study without my express written consent.
Quote:
Real Life™ has kept me from working on this for the past few days, but working with this theory, my data from runs already gathered would be much closer to expected drop rates if my LTs had something like a 0.5% or 1.0% drop chance.|
Originally Posted by EarthWyrm
The most striking thing, to me, is that this really does come out to almost exactly what we would expect. Archie's speculation about lieutenants being a big piece of the problem may be correct. Does the drop rate on this map skew downward when the team size is raised and there are lieutenants in the spawns?
|
It dawns on me to wonder if LTs might have a zero drop rate, but I think then that my results end up high enough (over expected) to be suspect.
I've been avoiding trying to do this, but I might go back and try to parse my chat logs and try to determine the drop rates by rank.
Blue
American Steele: 50 BS/Inv
Nightfall: 50 DDD
Sable Slayer: 50 DM/Rgn
Fortune's Shadow: 50 Dark/Psi
WinterStrike: 47 Ice/Dev
Quantum Well: 43 Inv/EM
Twilit Destiny: 43 MA/DA
Red
Shadowslip: 50 DDC
Final Rest: 50 MA/Rgn
Abyssal Frost: 50 Ice/Dark
Golden Ember: 50 SM/FA
Quote:
Yeah. And the 2080 was for each one separately: 2080 kills at x1, 2080 at x2. I'm SO tired of that map right now.
|
Originally Posted by EarthWyrm
If these are all minions (you're on the cargo ship, yes?)
|
119088 - Outcasts Overcharged. Heroic.
Quote:
OK, I've gotten a parser working. Using it, I've validated that I have gotten recipe drops from LTs.|
Originally Posted by UberGuy
It dawns on me to wonder if LTs might have a zero drop rate, but I think then that my results end up high enough (over expected) to be suspect.
|
In case anyone can see anything wrong with this idea, here's how the parser works.
- It reads the chat log line by line.
- It starts looking for a "You have defeated" line.
- When it finds a defeat, it notes the critter and switches to looking for an inf reward line.
- When if finds an inf reward line, it notes that inf on the most recent defeat, then switches to looking for drop lines until it finds a new defeat line.
- When it finds drops before a new defeat, it appends them all to a list on the most recent defeat.
- When it finds a new defeat, it starts over at looking for an inf reward line.
This approach is only reasonable for parsing a log of solo play. On a team you can defeat mobs that give their drops to other people, and you can get drops from mobs that other people defeat, so it's basically impossible to correlate drops and defeats. It might be possible if you merged logs from every player on the team, but I would expect things like different latency for different players to be a problem, since message order is important. (I don't even always see a defeat line for all mobs on large teams.)
Even solo, I think there is a small chance for incorrect parsing, because things like DoTs can mean mobs are defeated in the middle of your drops being reported. However, this seems unlikely in practice, as I can't find any examples of another message being interwoven between the defeat+reward lines. Since it doesn't seem likely to happen and I don't know of a good way to deal with it, I'm basically ignoring it.
Blue
American Steele: 50 BS/Inv
Nightfall: 50 DDD
Sable Slayer: 50 DM/Rgn
Fortune's Shadow: 50 Dark/Psi
WinterStrike: 47 Ice/Dev
Quantum Well: 43 Inv/EM
Twilit Destiny: 43 MA/DA
Red
Shadowslip: 50 DDC
Final Rest: 50 MA/Rgn
Abyssal Frost: 50 Ice/Dark
Golden Ember: 50 SM/FA
Quote:
I have as well. I need to figure out where those files are of mine - I was just cutting/pasting my reward chat logs (I put my rewards in a tab by themselves) into a word document, doing a find-replace of every instance of minion-level inf with the word "minion" and every instance of lt-level inf with the word "lieutenant," and then using that to count total minions and lieutenants. What that means is, if I can locate the document, I can see whether there was any consistent pattern in who dropped what.|
Originally Posted by UberGuy
OK, I've gotten a parser working. Using it, I've validated that I have gotten recipe drops from LTs.
|
The problem is that my laptop, which is my computer most able to run City, is in the shop; it had a MB replaced and I got it back last week, only to have it start pixelating and crashing whenever I logged into City Saturday night (video card failing). And that's where my document with the tracking data was. I may have a copy of it on the wife's laptop as well, since I used that while the MB replacement was ongoing, so I'll go through this afternoon and see.
I know that at some point I was going through the logs and noticing that the drops seemed to be sorting by rank, at least somewhat, I just don't remember how that played out or whether it was more than what my brain perceived as an idle pattern.
My postings to this forum are not to be used as data in any research study without my express written consent.
Hm, OK, I definitely need to add some data to my collection. With what I have I'm coming up short of Wiki percentages on both ranks. I need about four times as many kills as I have currently to feel good about the numbers I'm getting though.
Blue
American Steele: 50 BS/Inv
Nightfall: 50 DDD
Sable Slayer: 50 DM/Rgn
Fortune's Shadow: 50 Dark/Psi
WinterStrike: 47 Ice/Dev
Quantum Well: 43 Inv/EM
Twilit Destiny: 43 MA/DA
Red
Shadowslip: 50 DDC
Final Rest: 50 MA/Rgn
Abyssal Frost: 50 Ice/Dark
Golden Ember: 50 SM/FA
Random question, but how many of us are using "first-generation" characters as opposed to multi-numbered copies to test this? I doubt it has any effect, but I've gotten utterly miserable drop rates with Colorblind Goon5, but pretty close to standard with Lord Erik Villon1.
Just tossing out ideas. I think it's much more likely to have something to do with the mix of different ranks (maybe they both get diminished if two types of enemies are there).
Or does turning off bosses matter? I only ran one run, ever, while testing with bosses enabled. Wouldn't explain the cargo ship map, though, unless Lakanna's one of the few among us who hasn't told the game not to spawn bosses...
My postings to this forum are not to be used as data in any research study without my express written consent.
I have had bosses off, but I could easily turn them back on, since I've been skipping their spawns anyway.
Blue
American Steele: 50 BS/Inv
Nightfall: 50 DDD
Sable Slayer: 50 DM/Rgn
Fortune's Shadow: 50 Dark/Psi
WinterStrike: 47 Ice/Dev
Quantum Well: 43 Inv/EM
Twilit Destiny: 43 MA/DA
Red
Shadowslip: 50 DDC
Final Rest: 50 MA/Rgn
Abyssal Frost: 50 Ice/Dark
Golden Ember: 50 SM/FA
I've also built myself a log parser. This one uses the file that's saved to logs/game when you turn on chat logging.
I've been running the Demon cargo ship on difficulty 0, team size 6. I'm also _really_ sick of it now. 
The results are 1,629 defeats including 544 lieutenants and no bosses. I had 43 recipes, 140 salvage and 16 enhancements.
This gives total drop rates of 1.98 for recipes and 7.73% for salvage. The salvage rate is totally consistent with the old drop rate. The recipes are harder to say. They're low, but the 95% confidence interval is exactly 2.66%. In other words, there's about a 5% chance that this is an unlucky run with the old drop rate. We really need another set of runs to be sure. Any volunteers?
I counted the minions' and lieutenants' drops. The minions dropped 24 recipes and 87 salvage and the lieutenants dropped 21 recipes and 46 salvage. This is about the right ratio.
I've had bosses off for all my tests.
This is the first time I've been on test and the first time I've copied this toon.
I've done all tests with lvl 50 brute.
I just want to /threadjack a little bit here to say "You people are very dedicated and have a lot more patience than I do."
Thanks.
Where to now?
Check out all my guides and fiction pieces on my blog.
The MFing Warshade | The Last Rule of Tanking | The Got Dam Mastermind
Everything Dark Armor | The Softcap
don'T attempt to read tHis mEssaGe, And believe Me, it is not a codE.
Since there's been a patch since I started this, and I switched to bosses enabled, I started a new data set.
My goal is to get to around 2000 minions, with however many LTs that works out to. I'm running -1 Nemesis set to a team size of 6.
My data so far. (There were bosses on the map, but I skipped them, just defeating their buddies.) I'll just keep updating this data set with additional runs.
Defeats:
Rank: Minion, Defeats: 530
Rank: LT, Defeats: 176
Recipes:
Rank: Minion, Drops: 11, Rate: 2.075% (77.830% of expected)
Rank: LT, Drops: 7, Rate: 3.977% (74.574% of expected)
Salvage:
Rank: Minion, Drops: 43, Rate: 8.113% (101.415% of expected)
Rank: LT, Drops: 15, Rate: 8.523% (80.101% of expected)
As an aside, a new change with I16 is that Nemesis spawn snipers on outdoor maps. I always wondered why they didn't before. What's interesting about it is that they seem to replace minions, not other LTs. Because of this, I've gotten spawns with as many as 5 LT+snipers. That's a lot of Vengeance. Even if you defeat them in a smart order, the last couple of guys hit hard. And these are -1s.
Blue
American Steele: 50 BS/Inv
Nightfall: 50 DDD
Sable Slayer: 50 DM/Rgn
Fortune's Shadow: 50 Dark/Psi
WinterStrike: 47 Ice/Dev
Quantum Well: 43 Inv/EM
Twilit Destiny: 43 MA/DA
Red
Shadowslip: 50 DDC
Final Rest: 50 MA/Rgn
Abyssal Frost: 50 Ice/Dark
Golden Ember: 50 SM/FA
Quote:
Is that just a tactful way of saying "Hey! Are you guys crazy?".|
Originally Posted by Dechs Kaison
I just want to /threadjack a little bit here to say "You people are very dedicated and have a lot more patience than I do."
|
(The answer's "yes", by the way.)
Hmm. Those drop rates for Minions and LTs are almost exactly the same as mine. I got:
Recipes:
Rank: Minion, Drops: 24, Rate: 2.21% (82.85% of expected)
Rank: LT, Drops: 21, Rate: 3.86% (72.43% of expected)
Salvage:
Rank: Minion, Drops: 87, Rate: 8.02% (100.23% of expected)
Rank: LT, Drops: 46, Rate: 8.46% (79.47% of expected)
I went back to my 2500 council datapoints, which spread over at least 2-3 patches.
For lieutenants, I had 9 Pool A drops out of 488 defeats (1.84% drop rate)
For minions, I had 16 Pool A drops out of 1996 defeats (0.8% drop rate)
I'll try a run with Bosses on, then turn them off and try a run with AVs on, then turn both on and see if turning them both off somehow triggered a, "You're a wimp, no drops for you!" setting the devs had forgotten about. 
My postings to this forum are not to be used as data in any research study without my express written consent.
vacation?
Quote:
Bosses, no AVs:|
Originally Posted by EarthWyrm
I'll try a run with Bosses on, then turn them off and try a run with AVs on, then turn both on
|
5 recipes (one Pool C drop from a boss - stoopid Glimpse of the Abyss proc...)
45 salvage
268 minions, 68 lt's, 8 bosses
Bosses and AVs both enabled:
1 recipe
32 salvage
268 minions, 68 lt's, 8 bosses
AVs, no bosses:
2 recipes
31 salvage
268 minions, 76 lt's
Tentative conclusion: Doesn't look like it matters, but n=344 per map, so I could be wrong
My postings to this forum are not to be used as data in any research study without my express written consent.
Continued testing on same settings: Bosses, -1 foes, 6-person effective team size. Edit: AVs are off (none on the map)
Code:
Defeats:
Minion Defeats: 1099
LT Defeats: 354
Recipes:
Minion Drops: 23, Rate: 2.093% (78.480% of expected)
LT Drops: 10, Rate: 2.825% (52.966% of expected)
Salvage:
Minion Drops: 76, Rate: 6.915% (86.442% of expected)
LT Drops: 29, Rate: 8.192% (76.993% of expected)
Blue
American Steele: 50 BS/Inv
Nightfall: 50 DDD
Sable Slayer: 50 DM/Rgn
Fortune's Shadow: 50 Dark/Psi
WinterStrike: 47 Ice/Dev
Quantum Well: 43 Inv/EM
Twilit Destiny: 43 MA/DA
Red
Shadowslip: 50 DDC
Final Rest: 50 MA/Rgn
Abyssal Frost: 50 Ice/Dark
Golden Ember: 50 SM/FA
1134 enemies (918/216)
Expected recipes drop: 36 (28/7/1)
Effective recipes drop: 33 (27/6/0) = 92% of expected
Expected salvages drop: 97 (76/18/3)
Effective salvages drop: 106 (86/19/1) = 109% of expected
It was made in 3 separated runs of a mission set for 6. First at -1, then +2, then 0. Details:
First run: 13 recipes, 34 salvages
Second run: 9 recipes, 33 salvages
Third run: 11 recipes, 39 salvages
Seems pretty ok. Last time I did 4 runs on the same mission and I got 75% of expected recipes. This time, 3 runs, 92%.
I agree,
The conspiracy theorist in me now wonders if virtual teams is a ploy to hamstring farmers. The general population will be less willing to anchor/fill/pad teams because that isn't needed any more. even tho the drop rate is much lower with virtual teaming.
The developer in me thinks this is a bug no different that ticket drop bug when AE shipped (the one where not only were ticket drops split with the team but the number of tickets where split too ( ie: solo you would get say 10 tickets with 2 you would get 5, but only half the time))
Card Carrying DeFulmenstrator--Member Crazy 88s
We burn more Influence before 8am than you make all day.