Datamined Mission Rewards Are Bad. If True, Please Change Them.

[sambarpowder]

Wow, I'm late to this. Not much left to say, other than that I'm glad I never paid for this game. I've a huge amount of time logged in the game and on the forums, and this just reassures me when I start giving in and think about reaching for the wallet. Still, I've said it before and I'll say it again, I'm sticking around. When DE finally decides to man up and release the game,  I'll judge it.

It's fun enough to be worth it.

Edited June 3, 2013 by Crossflip

[Tent27]

Dear Devs,

 

I left WoW because I hated having to grind an impossible amount of hours to achieve F*** all.

 

Sincerely,

 

Tent27

 

P.S. Having restricted amounts of weapons and Warframes is dumb.

[[DE]Steve]

Open Kimono: we're doing it they way you guys are suggesting. Maybe we're not doing it right, have bugs in some areas (and yes, we're looking at them). Maybe we're math idiots, maybe we need to explore other systems (tokens, going out of business). There is not a hotfix coming that sets "drop rates for everything fixed" however. We'll tweak and tune but I don't agree we're on outlier on these rates.

 

Rare started at 0.5%, blah blah bell-curve distribution but we upped it radically through the months of playing this game for ourselves.

 

Binning results, 10k rolls.

 

BEFORE:
Nanospores = 2051 (20.5%)
Ferrite = 7704 (77%)
ControlModule = 245 (2.5%)

WITH ENERGYFRAGMENT:
Nanospores = 2051 (20.5%)
Ferrite = 3870 (38.7%)
ControlModule = 249 (2.5%)
EnergyFragment = 3830 (38.3%)

 

Here is the copy/pasted C++ code into Lua, and fixed for table formating and array-indexing by 1.

 

[edited for whitespace]

local COMMON = 1local UNCOMMON = 2local RARE = 3local RESERVED = 4local FLT_MIN = 1.175494351e-38local PROBABILITIES = { 0.76, 0.215, 0.025, 0 } -- normalizedlocal ItemRarity = { COMMON, UNCOMMON, RARE, RESERVED }local ItemRarityNames = { "COMMON", "UNCOMMON", "RARE", "RESERVED"}local ResourcesA = {    {        mItemType="Nanospores",        mRarity=UNCOMMON    },    {        mItemType="Ferrite",        mRarity=COMMON    },    {        mItemType="ControlModule",        mRarity=RARE    }}local ResourcesB = {    {        mItemType="Nanospores",        mRarity=UNCOMMON    },    {        mItemType="Ferrite",        mRarity=COMMON    },    {        mItemType="ControlModule",        mRarity=RARE    },    {        mItemType="EnergyFragment",        mRarity=COMMON    }}      local function Lerp(a, b, t)    return a * (1 - t) + b * t;endlocal function NormalizeProbability(t)    local countsByProb = { 0, 0, 0, 0 }    for i=1,#t do        countsByProb[t[i].mRarity] = countsByProb[t[i].mRarity] + 1.0    end    local totalProb = 0.0    for i=1,#PROBABILITIES do        if (countsByProb[i] > 0.0) then            totalProb = totalProb  + PROBABILITIES[i]        end    end    if (totalProb > 0.0) then        local normalizedProb = { 0, 0, 0, 0 }        for i=1,#PROBABILITIES do            if (countsByProb[i] > 0.0) then                normalizedProb[i] = PROBABILITIES[i] / (totalProb * countsByProb[i]);            end        end        for i=1,#t do            t[i].mProbability = normalizedProb[t[i].mRarity]        end    endendlocal function PickRandomBiasAtten(t, bias, atten, perElemAtten)    -- normalization pass based on atten/bias    local totalP = 0.0       for i=1,#t do        if perElemAtten[i] == nil then            perElemAtten[i] = 0.0        end           local biasProb = math.max(FLT_MIN, t[i].mProbability - bias)        local peA = perElemAtten[i]        if peA == nil then            peA = 0.0        end        t[i].mBiasedProb = Lerp(biasProb, 1.0 - biasProb, math.min(1.0, atten + perElemAtten[i]));        totalP = totalP + t[i].mBiasedProb;    end       if totalP > 0.0 then        for i=1,#t do            t[i].mBiasedProb = t[i].mBiasedProb / totalP        end           -- pick random        local fr = math.random()        for i=1,#t do            local biasProb = t[i].mBiasedProb            if(fr <= biasProb) then                return i            end            fr = fr - biasProb        end    end    return 1endlocal function RunTest(t, numRolls)    NormalizeProbability(t)    local resultBins = {}    for i=1,#t do        resultBins[i] = 0.0    end       for i=1,numRolls do        local idx = PickRandomBiasAtten(t, 0.0, 0.0, {})        resultBins[idx] = resultBins[idx] + 1           end       for i=1,#t do        print("" .. t[i].mItemType .. " = " .. resultBins[i] .. " (" .. math.floor(0.5 + (resultBins[i] / numRolls) * 1000.0) / 10 .. "%)")    endendlocal seed = 75732453 -- os.time()print("BEFORE:")math.randomseed(seed)RunTest(ResourcesA, 10000)math.randomseed(seed)print("WITH ENERGYFRAGMENT:")RunTest(ResourcesB, 10000)

[Miaradz]

Yeah i go away for the weekend and I'm late to the bash fest too. People also iterated on some points that i wanted to make too. :(

I am of the opinion that DE have indeed done goofed - Wave/Mission reward tables should feel (if not actually are) better than simply grinding defense missions for mods. The datamined information (if it is indeed correct) reflects this awful state of affairs. I have had but 1 Void Key drop from a looot of defense missions since U8 dropped, and high wave Defense missions are an exercise in futility.

 

While I don't believe this was a malicious act of $&*^ery on behalf of DE, I do believe they could have done more to prevent this state of affairs from happening. As Ced23Ric has mentioned - their inspiration coming from Diablo, is not a crime - but they forget that aRPG loot table tiers are scaled in such a way that you are rewarded for your progression, and once in awhile, rewarded with something awesome. As it stands I can get Split Chamber off M Prime (and I have) and have gotten absolute piss from Xini with regards to mob drops. Mission rewards are obviously a total crapshoot. 

 

I think there should be even more specific and granular tiering of reward tables, going from planet to planet. Emphasise on the factions unique drop rates to make people go from planet to planet to actually get the mod they want. It's going to be more work for your math wizards, but I feel it would be a much better improvement over this mess of a single pool.

Nevermind just read Wutever's post. I don't agree with warframe specific skewing though, there should still be a sense of "damn that would have been good for something else" randmoness in there.

 

Much like my previous assertions, too much RNG hurts you. Players still want agency - hence my advocacy of a form of token or trade-in system ever since I started posting here. Providing players just that nugget of info on loot availability gives them that carrot to chase without making them begrudge you for the perception that you are forcing them to grind to proceed. Nobody likes coercion, even if it is the illusion thereof. For example, in my line of work, a theatre performance that relies on audience participation to progress the story (not just to break the fourth wall, but to actually ransom for progress to the next line) causes immense resentment in the audience and they will value your performance less. This is exactly what you are doing. The scale of "effort" is not the same, but the perception of coercion is there.

 

I am not saying make your game hand it out on a platter, but make the effort we put into it be perceived as rewarding while balancing your own economic needs.

[[DE]Rebecca]

It was definitely interesting to have all this data appear without access to the code and coders! Someone in the other thread pointed out that rarely do other games share their RNG tables, and the same applies to us. After digging and showing links to the team about all these "datamining" threads, the fact that I've been told is that these numbers aren't far from the real ones in terms of content drops. See Steve's above post for how drops are tiered and coded. What matters to me is not only the affect this kind of data has on the well-being of our players, but the suggestions and outcomes that follow.  The "frustration with the lottery" system pre-dates the datamining topics, and our aptness to change has been (Under ECONOMY https://forums.warframe.com/index.php?/topic/55421-update-8-related-work/) is an incredibly important issue that needs to be improved. How?! Well...

 

A combination of suggested solutions and "what I'd like to see" are very helpful in forming a better drop system for Warframe's content.

So far, solutions that have caught my eye are:

 

Wutever explaining that the current problem and a lot of frustration comes from that fact that 'Your mission choices don't matter, having high level frames doesn't really make things better and things are counter-intuitive in general."

With this solution: "Loot tables need to be split up much more using level ranges, factions and locations to satisfied the diverse needs of different players."

Then,

 

ArsVampyre's suggestion of "Common, Uncommon, and Rare", with this kind of breakdown: "Give the reward RNG a percent change to pull a particular pool, say 75% for common pool, 20% for uncommon pool, and 5% for rare pool. After the RNG decides on the pool, each item in the pool has exactly the same chance as any other item in the pool". From what I have seen of our current design and can be viewed above, we indeed to have 3 areas (Common, Uncommon, Rare), but further tweaks need to be made to result in this proposed system.

 

Then, in other threads (https://forums.warframe.com/index.php?/topic/56949-void-drop-rates-void-keys-drop-rates-void-key-pack-drop-rates/),

Thelonious has suggested (https://forums.warframe.com/index.php?/topic/56949-void-drop-rates-void-keys-drop-rates-void-key-pack-drop-rates/?p=596722) that ability cards should be out of loot tables, factions should have consistency in their drops, and of course a whole overhaul of defense rewards (which I emphasize daily ;)).

The void key suggestions here are also interesting.

Then, all the 'mud slinging' aside, token systems were brought up which is something that is also being considered here.

Now, the important question that I know you're all asking: "WHEN and HOW ARE YOU GOING TO FIX/CHANGE IT?!" I don't know. Steve's post above depicts the results of adding new items into the rarity tears, and reading results after 10,000 missions. Thanks for all the suggestions so far!

[Miaradz]

By all means keep the new tables obfuscated. Make the community work for their conspiracy theories. But I think the key here is to be a smidgen, a pinch even, more transparent on the new mechanics (remember, players love carrots).

 

Also, I see what you did there, "rarity tears" indeed. :P

[shinydunsparce]

Conspiracy theories? It's not like the drop chances typed themselves.

[[DE]Steve]

By all means keep the new tables obfuscated. Make the community work for their conspiracy theories. But I think the key here is to be a smidgen, a pinch even, more transparent on the new mechanics (remember, players love carrots).

 

Also, I see what you did there, "rarity tears" indeed. :P

 

... I just posted the code. Unless I'm not understanding what obfuscation you are talking about? I don't think we need to paste the tables.

[Miaradz]

I apologise, the obfuscation comment was in reference to this:
 

Someone in the other thread pointed out that rarely do other games share their RNG tables, and the same applies to us.

And yes dunsparce, someone did type the tables out, but do not attribute to malice so easily when you have not gotten the full picture.

[ArsVampyre]

Rarity tears seems like a Freudian slip.

[Nillard]

So it looks like the real problem isn't the common/uncommon/rare system, but instead that anything in the mRarity=RARE set has its "drop rate" pushed down lower and lower as more stuff gets added to the game.

 

I like the idea of making some of it faction specific, adding higher tiers to defense rewards (so going past level 30 enemies/wave 5 later on is worth it), and possibly putting fieldtrons/mutagens/detonite and skill cards on separate "drop tables."

[Miaradz]

Rarity tears seems like a Freudian slip.

 

[[DE]Steve]

So it looks like the real problem isn't the common/uncommon/rare system, but instead that anything in the mRarity=RARE set has its "drop rate" pushed down lower and lower as more stuff gets added to the game.

 

I like the idea of making some of it faction specific, adding higher tiers to defense rewards (so going past level 30 enemies/wave 5 later on is worth it), and possibly putting fieldtrons/mutagens/detonite and skill cards on separate "drop tables."

 

The point of me giving out the code and example run is to show this doesn't seem to be the case for the data sets we have...

[Nillard]

Ah, sorry. I misunderstood. When something gets added to a rarity of rewards, the drop rate goes down for that set. The dataset shows adding a common item to COMMON makes the rate go down. Same would apply for the other rarity sets.

 

Sooooo.... rare stuff is just rare.

[Lumireaver]

I got caught up posting this and I missed you guys signing on. Would it be out of line with your vision of the game/business practice to somehow (but not totally) restrict loot based on hours played on a daily/weekly system? Within reasonable limits, naturally, and that's if certain rare rarities weren't as...rare.

[Thelonious]

Ah, sorry. I misunderstood. When something gets added to a rarity of rewards, the drop rate goes down for that set. The dataset shows adding a common item to COMMON makes the rate go down. Same would apply for the other rarity sets.

 

Sooooo.... rare stuff is just rare.

err...you were right the first time.

In steve example, mRarity pools have the same drop rate with or without Energy Fragments. As expected.

In the second case, with energy fragment, the common pool drop rate is equally divided between its two parts (Ferrite and energy, 38.7+38.3=77)

But what I don't get, Steve, is the purpose of showing us that.

Is it what you want to do, or is it already how it's working? I'm a bit confused here (and the page of gibberish code didn't help >_< )

Edited June 3, 2013 by Thelonious

[GottFaust]

err...you were right the first time.In steve example, mRarity pools have the same drop rate with or without Energy Fragments. As expected.In the second case, with energy fragment, the common pool drop rate is equally divided between its two parts (Ferrite and energy, 38.7+38.3=77)But what I don't get, Steve, is the purpose of showing us that.Is it what you want to do, or is it already how it's working? I'm a bit confused here (and the page of gibberish code didn't help >_< )

That's not gibberish, it's LUA.

[Kynian]

This is a re-post of a post I made on another, similar thread (https://forums.warframe.com/index.php?/topic/56949-void-drop-rates-void-keys-drop-rates-void-key-pack-drop-rates/page-10). I believe it has a bit more visibility here.

Just my two cents on all this:

you are correct, Waframe is a third-person shooter that uses RPG elements to distribute gear and deal with stats.

Wich is the argument that has been going on in here. That it needs to be looked at urgently. Several pages ago there was some comparisons to Diablo (wich DE said they were basing on, among other similar games) that seemed very valid.

In there there is A LOT more variety, any "trash" item you get you can trade/sell/make something useful out of it, and they have conditions where you get more chances of getting rare loot.

Hence all the whining. I believe rare things, such as mod, should be kept as rare. That said, I want rare to be rare, and not nearly impossible. I want it to scale with mission level and difficulty - it's only fair. I want to be able to keep previous rewards on each 5 defenses (say, when I exit a defense on 15th wave, I also get the 10th and 5th wave rewards) simply because it increases the risk as well as reward (one little mistake and you lose EVERYTHING you stacked so far. but do it right and you leave with a mountain of goodies.) I want to have quality game time, not a korean grindfest where new core content is nearly unreachable and/or locked behind paywalls.

And although that is my own personal take on it (notice I said a lot of "I want" in there), I believe most players will agree with me. Hopefully you will too.

I do not think I am entitled to stuff by saying all that, and I am against instant gratification. I really am. Things do need to change however.

I am a hardcore gamer on this game atm. I gotta check again, but pretty soon I'll be into 300's game time. And I'm not getting rewarded for that. I would if I had bought tons of keys.

With all that time, I'm lacking stalker visits (no hate or dread for me yet, and I'm ectively searching for him for over 80 games now, no visits)

I've yet to get any of the new rare mods introduced a while ago, such as Retribution, Thunderbolt and Handspring (and I've done a fair bit of grinding for Thunderbolt. Yet again, no luck). No Master Thief either.

And although I'm a great supporter of the game and will most likely continue playing for a long time, it is getting tiring. I can see myself soon only logging in and playing minimally as opposed to how addicted I am now, simply because of lack of interest. Maybe that could be a good thing, depending on point of view. Until, one day, the last drop finally comes and I abandon it completely. And I'm a founder, so I believe that is a very very bad thing (for them).

[Nerokerubina]

All dis beating around the bush.

[Doruben]

Wasn't this thread about mod/key drops as mission rewards and not resource drops from enemies and containers?

 

Besides that point, wouldn't 10000 rolls mean 10000 kills, rounding that down by 100, you would only get 2 control modules per 100 hyena runs, if he drops 3 random resources that's still 6/300. Even then there could still be 100 runs where no control modules drop at all.

 

"we don't want to make the game a grind"

"we don't like speedrunners"

[Miaradz]

Hence all the whining. I believe rare things, such as mod, should be kept as rare. That said, I want rare to be rare, and not nearly impossible. I want it to scale with mission level and difficulty - it's only fair. I want to be able to keep previous rewards on each 5 defenses (say, when I exit a defense on 15th wave, I also get the 10th and 5th wave rewards) simply because it increases the risk as well as reward (one little mistake and you lose EVERYTHING you stacked so far. but do it right and you leave with a mountain of goodies.) I want to have quality game time, not a korean grindfest where new core content is nearly unreachable and/or locked behind paywalls.

 

I disagree with your stacking defense rewards. It removes a layer of the risk management dynamic the current system enforces. Do you keep going in exchange for something better or do you exit now for a guaranteed pinch of fusion energy? Stacking rewards removes that question and has only the question of "How long can I keep the crypod alive?".

Also given that current mob kill drops are "better" than wave rewards, going to higher waves gets you a mountain of loot anyway. Why should DE let you double dip the loot lottery with a stacking reward?

Edited June 3, 2013 by Miaradz

[[DE]Steve]

err...you were right the first time.

In steve example, mRarity pools have the same drop rate with or without Energy Fragments. As expected.

In the second case, with energy fragment, the common pool drop rate is equally divided between its two parts (Ferrite and energy, 38.7+38.3=77)

But what I don't get, Steve, is the purpose of showing us that.

Is it what you want to do, or is it already how it's working? I'm a bit confused here (and the page of gibberish code didn't help >_< )

I'm showing adding the research didn't screw over the rare drop, which some people were suggesting we should fix in our code.

[Kynian]

I disagree with your stacking defense rewards. It removes a layer of the risk management dynamic the current system enforces. Do you keep going in exchange for something better or do you exit now for a guaranteed pinch of fusion energy? Stacking rewards removes that question and has only the question of "How long can I keep the crypod alive?".

Indeed, it becomes an endurance test. If you win on wave 70, as I've seen some players do, you walk out with a mountain of goods. If they fail to protect it, they lose EVERYTHING they stacked. Mats, dropped mods, stacked rewards.

So it's a Huge Risk vs. Huge Reward. Seems only fair to me.

[Webly]

I'm showing adding the research didn't screw over the rare drop, which some people were suggesting we should fix in our code.

Maybe they are suggesting it was always bad.

[[DE]Steve]

Wasn't this thread about mod/key drops as mission rewards and not resource drops from enemies and containers?

 

Besides that point, wouldn't 10000 rolls mean 10000 kills, rounding that down by 100, you would only get 2 control modules per 100 hyena runs, if he drops 3 random resources that's still 6/300. Even then there could still be 100 runs where no control modules drop at all.

 

"we don't want to make the game a grind"

"we don't like speedrunners"

Except that it is only rare for *every single enemy kill* and Hyena has radically improved chances. [edit: I believe its 25%]