In-game Data Collection Question

pheregas

With another DDQ 4* challenge upon us, I had a question...

In the new era on MPQ where we are, more and more, "supposed to lose," what is the best way to give in-game performance feedback on difficulty?

For example, in today's 4* challenge of Blade vs. Elektra, my 2/1/1 lvl 78 version stands little to no chance. I can try over and over again every 35 minutes to prove how far out of my league this battle is (even though I know how I'm supposed to play it, I just know it will be so difficult as to not get my hopes up). Or is it better to play once, get wiped, then not try again? Or is it better to not even attempt it once, knowing that the data is there that I do have a rostered Elektra?

I'd like to leave the debate out of this question regarding who these new DDQ challenges are meant for. I think the best way to convince the devs that something is, or isn't too difficult, is to give concrete data. If 80% of everyone who tries, dies, and that is what was intended by the devs, then kudos to them for achieving their goal. If they want more people than that to win, then I am truly interested in knowing what type of in-game data collection feedback would be most beneficial to the data miners in how to progress.

Thanks!

Polares

Not only your Elektra doesnt stand a chance my 130 2/3/3 doesnt stand a chance either.

Blade gets black? 3000 damage and I cant get purple anymore after that (if I am still alive of course)
More than 9 red? 2 93 strike tiles
Blade gets purple? 1200+ (easily 2000 with a couple of strike tiles) attack tiles

7300 hit points against 14000
64 match damage against 85+

Yes Elektra can use her black power, but my traps jump just with match damage so in a couple of turns my 3 traps were gone (I cant delay the traps until I see he has 8 black, but he can kill me just with match strike damage easily). Purple can steal a couple of tiles, but thats not enough when he can produce more for free and I can't.

After seeing that Squirrel Girl was level 222, I thought that being Elektra one of the worst 4 there is, they would use a level 200 or less enemy, not 270 (wrooooong devs are massoquist, and so I am for playing this game). And Blade is quite good, so at 5/5/5 and level 270 he is very good.

I have tried 4 times, and 4 times I got wiped. I doubt anybody with maxed or almost maxed Elektras (10+ covers) in the 180 lvl range stand a chance. This is too hard for DDQ.

This is by far the most difficult DDQ4 fight.

moogles85

Dino VS SG at lvl 222 was only for the 2nd year anniversary - hence all 2s.

Don't expect anything other than 270s for any other 4* DPD.

Someone in my alliance beat him with a 1/3/2 after 10 tries so it's possible.

pheregas

I'd like to reiterate that this particular question in the forum is not meant for discussion regarding the feasibility of winning today's challenge. This is a real question about optimal in-game data collection and how those results are tabulated and the best way for the devs to see things.

If people barely squeak by after 15 attempts, is that better, or worse, for having difficulty levels reduced? I mean, if people are completing it, it isn't too difficult, right?

I've yet to attempt today. I'm busy clearing DP vs. Marvel (Heroes) again.

GrumpySmurf1002

Polares wrote:

I doubt anybody with maxed or almost maxed Elektras (10+ covers) in the 180 lvl range stand a chance.

A couple of guys in my alliance have cleared it at level 125ish, but they have her max or near max covered.

tanis3303

I feel like I (and anyone else that had some really nice luck) skewed the numbers on this one today. My Elektra is fully covered and lvl 200, so that definitely helps, but the board I got...oh man. Started with a T-shaped match 5 on purple on turn 1 that exploded into so many cascades that I couldn't even believe my eyes. By the time Blade got to make his first move, I already had enough purple for a double-double cross (not that i could use it, i took ALL the red) and already had a Ballet of Death trap out on the only red tile left on the board. There wasn't even any black for him to take, my cascades got most of that too. Eventually I double-cast her purple and got enough reds for a 2nd red trap...man, this thing was over for me in a matter of like 5 turns. Insane luck.

I guess my point is, for everyone that wipes continuously on these nodes, there's guy or gal that gets the luckiest board ever and clears it easily, essentially skewing the numbers and making everything seem peachy on paper, when in reality, not too many people are actually happy. As quoted in the WoM/Black Vortex thread (and don't even get me started...) "you can't fix feelings with math"

BlackSheep101

tanis3303 wrote:

I guess my point is, for everyone that wipes continuously on these nodes, there's guy or gal that gets the luckiest board ever and clears it easily, essentially skewing the numbers and making everything seem peachy on paper, when in reality, not too many people are actually happy. As quoted in the WoM/Black Vortex thread (and don't even get me started...) "you can't fix feelings with math"

This is flawed logic. People that get the luckiest board ever win and are done. People that wipe can lose a couple dozen times in a day.

Unknown

pheregas wrote:

This is a real question about optimal in-game data collection and how those results are tabulated and the best way for the devs to see things.

I suspect nobody is answering your actual question because the devs can no-doubt already track how often people are winning, with which teams, and by which margins. In terms of just math, there really is nothing we can give them that would be better than what they already have.

In terms of "this sucks to play" ... that's a different story and I imagine the forums are the right, uh, forum. See how it worked during Galactus? They knew how many alliances won (and I'm sure also can track how many force-closes happened), but everyone hated it so much (rightly so, imo) that they made it MUCH easier and actually play-able the second time around even though their stats showed more completion than they expected the 1st time around.

Unknown

Sundance2112 wrote:

pheregas wrote:

This is a real question about optimal in-game data collection and how those results are tabulated and the best way for the devs to see things.

I suspect nobody is answering your actual question because the devs can no-doubt already track how often people are winning, with which teams, and by which margins. In terms of just math, there really is nothing we can give them that would be better than what they already have.

In terms of "this sucks to play" ... that's a different story and I imagine the forums are the right, uh, forum. See how it worked during Galactus? They knew how many alliances won (and I'm sure also can track how many force-closes happened), but everyone hated it so much (rightly so, imo) that they made it MUCH easier and actually play-able the second time around even though their stats showed more completion than they expected the 1st time around.

The hard part of course is "How can we make something that only a fraction of the playerbase can beat ALSO feel fun?"

Which I wager is part of the challenge of working on a game like this.

How can you make something that only has a .05% clear rate also fun?

Not to mention, how can you do it without making it super easy for those with firm and strong top tier rosters WITHOUT changing that .05% clear rate?

Good luck.

grunth13

Well, I finished it with a level 120 2/2/5. I used 5 health packs. Total of 12 r/y, 12 g/b, 6 all ap boost, 10 10% boost on p/b. The only reason I finished it was that I started with a match 5 that gave me an insane cascade (I guess it happens for us to sometimes) and was able to cast her black and red immediately. Got some purple for her double double cross when he finally starting putting strike tiles down. I still lost 50% of my health and he didn't even cast a power. The other 5 times I tried, I barely made a dent as he started spawning strike tiles immediately. So...after spending all these resources, I get....ANTMAN yellow. Didn't really want it since I don't have enough covers for him to make him worthwhile anytime soon. I feel that the whole 4* dpq was meant to make us level our **** 4* to a reasonable level so that we have to be even more ISO starved. What happened to "you aren't supposed to level all your characters, you are supposed to pick and chose". Well, then don't keep making essentials or required characters so that I can chose my best 30-40 that I want to keep and get rid of the ones I never play with even when they are boosted. BTW, I had to level my Electra from 104-120 right before playing, so thanks for letting me use my minimal ISO on her instead of JG who I have been trying to get worth using.

MarvelMan

Polares wrote:

Not only your Elektra doesnt stand a chance my 130 2/3/3 doesnt stand a chance either.

Blade gets black? 3000 damage and I cant get purple anymore after that (if I am still alive of course)
More than 9 red? 2 93 strike tiles
Blade gets purple? 1200+ (easily 2000 with a couple of strike tiles) attack tiles

7300 hit points against 14000
64 match damage against 85+

Yes Elektra can use her black power, but my traps jump just with match damage so in a couple of turns my 3 traps were gone (I cant delay the traps until I see he has 8 black, but he can kill me just with match strike damage easily). Purple can steal a couple of tiles, but thats not enough when he can produce more for free and I can't.

After seeing that Squirrel Girl was level 222, I thought that being Elektra one of the worst 4 there is, they would use a level 200 or less enemy, not 270 (wrooooong devs are massoquist, and so I am for playing this game). And Blade is quite good, so at 5/5/5 and level 270 he is very good.

I have tried 4 times, and 4 times I got wiped. I doubt anybody with maxed or almost maxed Elektras (10+ covers) in the 180 lvl range stand a chance. This is too hard for DDQ.

This is by far the most difficult DDQ4 fight.

I tried it twice. First was a lvl 70 3/5/4 TinyDancer and got Blade to ~60%, second was after bumping her to lvl 99 (same covers), bossed b/p and b/g....won first time moderately easily and it wasnt a great board. No match 5s the entire time, only one match 4.




But to address the actual original question.....Play it more. The more you wipe (as long as it doesnt prevent you from playing other aspects) the more data it gives the devs that its a difficult fight.

smatterchoo

pheregas wrote:

In the new era on MPQ where we are, more and more, "supposed to lose," what is the best way to give in-game performance feedback on difficulty?

As a past mentor of mine liked to say, "Figures lie and liars figure."

At a blind guess, I'd say the more times players throw themselves at the wall, the better. It will decrease the numerator in the ratio of completions, yielding a lower completion ratio.
However, the point of bringing up the old saw is to say "they can spin the data however they want".

If it were me running the numbers, I'd be actively filtering out players with less than a certain % chance of winning to get more 'real' stats. Or creating an actuarial table with #covers & level into wins. Or both, cuz I'm a numbers nerd like that. Come to think of it, I'd expect the data to be in some sort of OLAP or other sort of data cube, then pivot it out to highlight the interesting bits.

I did manage to win on my fourth try with a 1/3/3 lvl 15whatever. Lucky board.

Good luck, both with your question & your matches.

Boommike

I have a feeling that this will be another thread void of red names... but this is such a valid point. Im towards the end of the 3* transition (I know, i know, DDQ4 isnt for me), but i'd like to help provide feedback to the devs on how someone at my level is dealing with DDQ4 (namely I'm not).
I attempted against cyclops a couple times and havent tried even once since. So is the lack of data input from me (and others in my boat) skewing the overall results?

Unknown

(moved: wrong place)

Unknown

Boommike wrote:

I have a feeling that this will be another thread void of red names... but this is such a valid point. Im towards the end of the 3* transition (I know, i know, DDQ4 isnt for me), but i'd like to help provide feedback to the devs on how someone at my level is dealing with DDQ4 (namely I'm not).
I attempted against cyclops a couple times and havent tried even once since. So is the lack of data input from me (and others in my boat) skewing the overall results?

There is also this, well, intentional skewing of data if you know exactly how it's collected.

Like the people who heard "TBE is usually more for people in 3star transistion then those with 2star rosters" and was like "Well then screw that I'm never using a 3star in that fight again to try and change said data"

If somebody like you heard "We measure success rate based on number of attempts" you'd be more likely to toss Electras at the node over and over again intentionally trying to boost said failure rate, while it's "We measure success by the number of attempts" would lead to you intentionally skipping it, even if you do have a slim chance to win.

Both of which would kinda skew the data and make it less reliable.