Jump to content
Search In
  • More options...
Find results that contain...
Find results in...
Dirkoff

Is Crafting too much RNG?

Recommended Posts

 

This is just as bad as the current year argument.   Lets try something fun with this statement!

 

Sorry, but scientific research and study that I'm not going to reference when I say this disagree with your assessment of people not having an alien creature from another planet attached to their spinal column controlling their thoughts!

 

Unless of course you have counter scientific studies and principles you would care to point out that explain how they don't!

 

Show me yours I'll show you mine.

Sigh.  

 

Since you can't be bothered to go back a couple of pages where I first brought in operant conditioning, I'll go over it again with a bit more detail.

 

Link to point in the video where they describe it as a slot game "but deeper".

 

https://youtu.be/og_DJoG08T4?t=149

 

Since I used to build slot games for a living, I think I know a few things about how they work.  (Whats your game building experience?)

 

Anyway, Slots are built primarily on the premise of Operant conditioning. Basically a Skinner box. Here is an overview of the concepts from a paper published Saul McLeod published 2007, updated 2015

 

 

Of note is the effect of punishment described on page 3.

 

Punishment is defined as the opposite of reinforcement since it is designed to weaken or eliminate a response rather than increase it. It is an aversive event that decreases the behavior that it follows.

 

Like reinforcement, punishment can work either by directly applying an unpleasant stimulus like a shock after a response or by removing a potentially rewarding stimulus, for instance, deducting someone’s pocket money to punish undesirable behavior.

 

Note: It is not always easy to distinguish between punishment and negative reinforcement.

There are many problems with using punishment, such as:

  • Punished behavior is not forgotten, it's suppressed - behavior returns when punishment is no longer present.
  • Causes increased aggression - shows that aggression is a way to cope with problems.
  • Creates fear that can generalize to undesirable behaviors, e.g., fear of school.
  • Does not necessarily guide toward desired behavior - reinforcement tells you what to do, punishment only tells you what not to do.

 

Losing items for using the create button is direct punishment for behavior, the behavior you are supposed to find pleasure in, and has the exact opposite effect a game should have. Punishment weakens the positive response, so if there are more failures than successes, you will eliminate any positive reinforcement or pleasure responses you were trying to develop.  Crafting will create fear, even before the success happens.  The experience will be more negative than positive, and the net effect will make it not fun.

 

Just listen to the 5 minutes of the conversation from where I linked, with Blair grinning like a Cheshire cat and almost completely oblivious or even reveling in the obvious discomfort that Markee dragon has to the idea. "You may have made a substandard item". No, your RNG made the substandard item, even though I did everything I was supposed to.  That is text book punishment, taking my stuff and making me feel like I did something wrong for even playing the game.

 

The psycological impact is obvious, with just demonstration resources, Markee twingding and squirming and calling the abort/loss button "Pixel scream" (5:55)  from this model.  The exact guy this is supposed to appeal to, has to fight through discomfort to look for what he hopes to find.

 

If I ever saw this reaction in a blind user group study for a game I was in charge of giving approval on, I would send it back for such a heavy review that the wallets of the investors would be gasping in pain.

 

A quick google will show you to hundreds of studies on the effect as it relates to gambling. (I didn't make the gambling reference, TBlair has been peppering his videos with gambling examples for both harvesting and crafting practically every time he gets in front of the camera and talks about them.)

 

You might like to read "Organizational Behavior and Human Performance" Volume 4, Issue 4, November 1969, Pages 375–401 the article "Beyond the teaching machine: The neglected area of operant conditioning in the theory and practice of management", where the topics of "The power of positive reinforcement, the unanticipated consequences of punishment, and the value of partial reinforcement are stressed."

 

That article, and a plethora of articles like it, fully validate the opinions I have expressed in this and other threads.

 

I've shown you mine, now you show me yours.

 

Afterthought.

 

Really the majority of the crafting game is pretty solid.  With the experimentation points (free spins), and the way that math works, all they need to really do is change the odds tables to give a real choice given the influence BP's are going to have on the system.  Also making substandard is not terrible, any more than little wins in slot games are terrible.  Those little wins reinforce that a big win could be coming right around the corner.

 

There are two really terrible things in the model. The total, abrupt, and punishing loss of final combine failure, that players who did everything right up to that point, and did nothing wrong by pushing the button, will definitely feel, and the incentive to trigger the 50%+  "risk" bonus that makes all other options substandard.  

 

Dark souls was harsh as hell, but you almost always felt like if you had done something differently you would have had a chance.  This crafting model does not feel that way.  

 

Without a mini game you can influence with "skill", there is nothing you could have done better, and yet you are punished just the same.

 

It would actually be more interesting if they based the game more on a fruit machine (UK style bar game) models.  Nudge/Hold and Hi/low games really make you feel like you can use skill to influence the outcome.

Edited by KrakkenSmacken

Share this post


Link to post
Share on other sites

On mobile and that's a monster of a post so now gonna quote it, but @ that last post:

 

100%. The end of your afterthought actually made me think of EverQuest 2 crafting with their little mini game. It took way too long and was a little too simplistic once you figured it out, but I liked what they attempted to do with that system.

 

That said, I'd still rather see the outright fail chance on final combines changed. Salvaging and higher chance of making poor quality items would fix it properly. You can even add a "Critical Failure" at a very very low rate (higher on higher quality items) that actually is a real failure, I think even having that differentiation that it's not a "normal" occurance would go a long way.

Share this post


Link to post
Share on other sites

On mobile and that's a monster of a post so now gonna quote it, but @ that last post:

 

100%. The end of your afterthought actually made me think of EverQuest 2 crafting with their little mini game. It took way too long and was a little too simplistic once you figured it out, but I liked what they attempted to do with that system.

 

That said, I'd still rather see the outright fail chance on final combines changed. Salvaging and higher chance of making poor quality items would fix it properly. You can even add a "Critical Failure" at a very very low rate (higher on higher quality items) that actually is a real failure, I think even having that differentiation that it's not a "normal" occurance would go a long way.

 

I gave advice, make it more like a fruit machine, so right now I am rolling around an eventual suggestion that uses all the existing components and leavers, most of the existing graphics, but adds a more significant choice factor to the game.

 

Frankly, these types of games are literally the hardest types of games to make.  The pure RNG we have now is easy, but ones that add a feeling of having a skill component, has a feeling of being fun, is still statistically an RNG, AND takes into account a fictional numerical skill associated with your avatar that can improve your odds.  Yea that's a tough game to make.

 

That said, I think I have just the small addition necessary, 90% fleshed out in my head.  

 

It really wouldn't be fair to say "make it more like X", without offering reasonable suggestions on how to do that.  My last effort had way too many needed changes to make it work. This one will be a rather "easy" addition.

 

I expect I will polish it a bit in the next day or two, and then post in suggestions.

 

UPDATE:

 

Decided to make it my first post in the Development partners forum.

 

Sorry to those who can't see it, but I wanted to be sure ACE at least read it.

Edited by KrakkenSmacken

Share this post


Link to post
Share on other sites

Share the nuts and bolts of your post for those of us that cannot see it?

 

I don't want to be accused of double posting, but since you asked.

 

 

So I have been rather vocal in the main forums about the user experience related to "failure" taking all resources and "flushing" them as thomasblair put it in this video.

 

I also suggested in that comment that making the game a bit more like a UK "Fruit Machine", so as to give the impression of skill, would greatly improve the experience.

 

So given the above, here is my suggestion as to how that can be done, with the current crafting mechanics as I understand them, not having a direct knowledge of the behind the scenes algorithms.

 

Replace the pip "Success - Amazing Success" results with a modified mini game I call High/Low & Range.

 

The game is based on a D20 check, (cause who doesn't love the D20?), made every time you perform either an experiment, or combine, in the crafting window.

 

The first "game" High/Low, is really simple, in that one is guessing if the next roll will be higher or lower than the last roll. This is a continuous series of events, so that if your first roll is a 5, and you guess higher and roll a 12, you now have to guess higher or lower on the 12.

 

If you get a high/low correct, your pip value increments by 1 + (0.1% per value on the die) (3.0% cap with a 20).

 

If your second roll exactly matches your first roll, the values are double of the value that would normally be assigned regardless of which choice you picked. (6.0% cap)

 

The second part is where I placed trained skill, that part is called range.  

 

I noticed that skill and difficulty values all appear to be between 0-75 (100?), and seem to have increments of 5 every "point". This also fits with a D20 model of 5% increments per value.

 

So when you select range you are guessing the range of numbers the roll will be in. This range you have available to select is determined by taking skill minus difficulty, with a minimum value of 3 (15)  and a maximum value of 10 (50) . (1/6.66 - 1/2 chance of being correct).

 

If you get that range correct, you add 0.5 to the pip for each point in your total range (5.0 cap). If your range includes your last roll, and is an exact match, this value is also doubled.

 

If you get either the High/Low or the range correct, you continue the series and carry forward half the previous Pips value. If you get both wrong, the pip is worth 0 “Failure”. If you get two in a row wrong it’s treated as critical failure and all value in any previous series on that row is lost.

 

For every miss, be it High/Low or Range, the item complexity increases.

 

A "Risk" bonus is applied if you leave open at least 50% of the spaces. This doubles the value for the Hi/Low check (6.0 cap) and range (10.0 cap)

 

In the most extreme circumstances, where there are 12 pips in a single line, and every roll was a 20, and every time you selected the range to include 20, the total bonus for that would be 704%.  The odds of that happening EVER is 1/4,095,999,999,999,990. (4,095 quadrillion). The odds exceed the average lottery at 7 in a row, with a 385% gain. (See Below Graphic for all Maximums)

 

 

Examples.

 

eAOD2B9.jpg

 

Skill range in the first example is 3 (minimum)

The starting roll is 5.  The best high/low bet is to go with higher, so you do that.  You want to eliminate as much risk as possible of getting zero from your test, so you pick a range 1-3 for that. Only 4 will hurt you.

 

You roll a 12, hit your hi/low, but miss on the range.  Pip value increases by 2.2 and your series starts.

 

You are close to 50/50 on the hi/low, so you want to take a big risk, so you guess higher. (55% Risk).  You also and put your range to 12-14.  This leaves the values 1-11 as a total loss if they come up. You could have selected low and 13-15 range, and only had 5 positions that you could have lost in, but you are after that big win.

 

You get lucky and hit 14.  Total gain is (2.4 + 1.5) X [2 (Risk)] = 7.8 added to 1.1 (half from the previous pip), and this pip is now worth 8.9

 

Since you got an 14, and are on a roll, the obvious choice is to bet low, and put your range 15-17, so you have very little chance of losing it all.  This does eliminate the possibility of a 2X multiplier for risk, guarantees an increase in complexity, but you get to carry half of 8.9, 85% of the time this way.

 

The roll is 8.  So the pip is worth 1.8 + 4.5  = 6.3.

 

Again you're near the middle, but this time you want to play it safe.  So you go higher, and select 5-7 as your range.  

 

You roll a 1, and bust out.  Pip is worth nothing, you add 2 complexity, and the series is broken.

 

You only had 5 pips to spend, and now have one that if you want to be high risk is a flat 5% chance, so if you get everything right on the low risk, you may get 10.4 for it if you roll another 1, but most likely simple 1.1 for hitting hi/low.

 

But there is another option, you can skip that last pip and enter the combine phase with a 1 as your start value, guaranteeing that it will not fail, but giving up the potential benefit of more experimentation.

 

Also for the combine phase, hitting the different values would have a different result.  Getting one (High/Low) or Range correct, you would get the lowest material quality used in the result, getting both would give you the higher. Hitting an exact match on the roll and that value being in the range, would give you one quality grade higher.

 

Failures that destroy your item could still happen, but most often because you chose not to move into the combine phase on a good roll, or take a risk and over lap Hi/Low and Range to try to get a higher quality result with lower quality materials mixed in.

 

 

Example 2. Full mastery of crafting

vSCsDGx.jpg

 

In this example I will show the same rolls, but how a crafter with a 10 range could be affected/played out.

 

From the 5, you want an early high risk, so choose low and range 1-10. This gives you a 25% chance of getting both matches, and the risk double up.

 

Bad luck, the 12 is out of range, so you get nothing.

 

Continuing with High risk, now you go High and 11-20

 

14 hits, so that's 2.4 for the Hi/Low 5.0 for the skill/range and doubled for high risk. 14.8 total and now you are higher than the safer player was, even though you missed a full pip.

 

Now because you know that getting either one correct, there is a bonus of 7.4 automatically assigned, and you decide to take a zero risk chance. Betting low with range 11-20.  Literally no chance of failure, but no chance of risk bonus, and will certainly increase complexity.

 

8 is rolled, so hit’s the lower but not the range.  You get 1.8 for the pip, plus the 7.4 from the last pip, for 9.2.

 

4.6 is kinda low for a carry forward,so now you want take a high risk again, low and 1-10 range.

 

Jack pot, hits a 1. That’s another ((2.4 + 5.0) X 2) 14.8 plus the 4.6 for a total pip value of 19.4  

 

EDIT (Got this wrong, should be (1.1 + 5.0) X 2 for 12.2 + 4.6 for total 16.8)

 

Going into the 5th pip you have a harder choice.  Give up the 9.7 (8.4) automatic carry forward, plus the guaranteed win on the high/low and risk entering combine with a poor roll, or take the 1 into the combine and guarantee the item at 43.4% improvement, or even go for broke with a 1/20 exact match by guessing the impossible higher counting on an exact match and range 1-10. (What the image is showing.)

 

You will notice that both scenarios played out very differently with the exact same random rolls. Every roll could be good or bad, depending on how the player decided to guess.

 

Now that feels like a real choice.

 

By increasing passive trained skill, you increase range, increase the value of all range matches, and improve not only the average outcomes (more range points added to pips), but also the number of options you have in regards to how much risk you want to take with any given roll.

 

Another wrinkle that I have not mentioned yet, would be being able to select different rows and start/continue the series they have, without changing the last roll.  

 

So if you were pushing for example maximum damage, and you really wanted every pip to hit, when you see a start roll you didn’t like, use that start on something you are not really concerned about, say durability, and hope that roll gives you a better starting value to move back into maximum damage with. In this way you can somewhat control the roll you start with on the things that matter the most.

 

Again, real choice, and not that much of a change to the underlying systems. (I think).

 

Best part about testing this, you can just roll dice without writing a single line of code.  Cut out a few strips of paper to indicate range, and a scrap of paper to do math on, and give it a spin before anyone spends programming time before seeing how it feels.

Share this post


Link to post
Share on other sites

Yeah never been a fan of RNG in crafting. Think some have suggested and I agree with at worst a "fail" should result in a lower quality item being made not outright lose of the item and materials. At least then you can still use the item and maybe sell it to recoup for the mats.

Share this post


Link to post
Share on other sites

Thanks. I'll take the blame for the double posting if anyone asks. :)

 

 

That said, I generally don't enjoy mini games as a substitute for RNG but this merits a second and third read because I'm tired at this very moment. X_x

Edited by Rhast

Share this post


Link to post
Share on other sites

Yeah never been a fan of RNG in crafting. Think some have suggested and I agree with at worst a "fail" should result in a lower quality item being made not outright lose of the item and materials. At least then you can still use the item and maybe sell it to recoup for the mats.

 

Or a variation on this, where the resources spit back out (rather than the desired item), but each is lowered in quality.


IhhQKY6.gif

Share this post


Link to post
Share on other sites

Thanks. I'll take the blame for the double posting if anyone asks. :)

 

 

That said, I generally don't enjoy mini games as a substitute for RNG (or at all)

 

I'd prefer to have the RNG skewed in favor of the player based on vessel/account skills and remove the item loss on craft fail.

 

I saw elsewhere that someone suggested a quality downgrade. Go further, if crafting a blue/epic/legendary and your final craft just happens to hit that tiny percentage that says critical fail.... have it downgrade to white or gray.

 

Or maybe, if we really, really want the risk to be present.. have a catastrophic fail where the item isn't made, but you're left with something that can be salvaged for a partial refund of the mats that went into it.  (around 60%?)

May have been me, or others that suggested that option. 

 

I am not married to any particular solution, so I did suggest downgrade earlier, but I very strongly believe something needs to be done to remove the sense of "Punishment" in the crafting experience, simply because there is zero sense of control at that point. Push a button, maybe get a candy, maybe a punch in the face, does not a good game make.

 

If it can be made a funner game along the way, all the better.  Go ahead, try the game with just a D20, and tell me how you feel about it.  

 

Looking for some experiential feedback, even if they never come close to implementing something like this.  

Edited by KrakkenSmacken

Share this post


Link to post
Share on other sites

Or a variation on this, where the resources spit back out (rather than the desired item), but each is lowered in quality.

and/or quantity as well. But yeah something at least. Not really even into crafting myself and I've always supported ACEs Risk/Reward philosophy but yeah total loss for a fail is a bit much in any game, IMO.

Edited by pang

Share this post


Link to post
Share on other sites

It's very cool idea. Crafting big items really would become much more interesting with such system. However, it can become a little bit annoying for small items craft like metal bars. Maybe I'm wrong. But I totally would vote for this anyway.

Share this post


Link to post
Share on other sites

It's very cool idea. Crafting big items really would become much more interesting with such system. However, it can become a little bit annoying for small items craft like metal bars. Maybe I'm wrong. But I totally would vote for this anyway.

Well I didn't want to overload it more than I already had, but it would be fairly easy to add a couple of quality of life features that would use the same system, but do it automatically if you were so inclined.

 

A simple setting option that was "High risk", or "Low Risk".

 

High risk would pick a setting with a high enough risk to trigger the bonus, while low risk would try to cover as many positions as it could, so you could simply click through like you do now. 

 

That way it would be easy to go quickly, or take your time and agonize over every choice.

Share this post


Link to post
Share on other sites

I think part of the reason I am not a fan of the current crafting system is My crow was not invested in the skill trees for crafting, or gathering for that matter. but even when I experimented with the crafting, it is extremely frustrating to have fails that give nothing. I think that even if you are an unskilled woodworker, you can make a chest of drawers that is ugly, but barely functional. It likely won't have the longevity nor the structural capacity as one that was made by a trained and skilled woodworker.

Now I know that we have access to many things that we wouldn't ordinarily be able to craft, because... testing.  But I am in favor of the idea of a substandard work or a barely useable work instead of an outright fail.  Think of the knife-makers that failed to harden their blades correctly. they would work until stressed, then fail in some capacity. or they have a wicked warp, which always causes a turn when wielded. 


56a6b2d2-1882-42f8-92d5-98014502b931.jpg

Share this post


Link to post
Share on other sites

Or a variation on this, where the resources spit back out (rather than the desired item), but each is lowered in quality.

 

An interesting thought, but I prefer the former option. Crafting and use of items should be a commitment. The items should be used specifically for that piece of gear and if it fails, you didn't put together the piece of armor/weapon/etc together in quite the right way, so you have a lower quality item.

Share this post


Link to post
Share on other sites

testing RNG in an Alpha with low level items with say 5 or 6 continuous fails on a 3 piece item that has a 67% chance. After a couple hours I logged out frustrated and done.

 

Ridiculous at this stage of testing to be spinning your wheels. 


3529195.jpg

Ellowyn of Prothoe • Death Server • Free Corps • Amazon

Share this post


Link to post
Share on other sites

testing RNG in an Alpha with low level items with say 5 or 6 continuous fails on a 3 piece item that has a 67% chance. After a couple hours I logged out frustrated and done.

 

Ridiculous at this stage of testing to be spinning your wheels. 

6 continuous fails in a row on a 67% chance of success should only happen 1/774 times (.04% of the time).

 

Tester observation indicates there is something dreadfully wrong with the RNG and it has a high propensity for streaks of failures.  

 

It would be very helpful in testing if in addition to the actual results, if we could see the RNG number used, and the range of effects checked for.

Edited by KrakkenSmacken

Share this post


Link to post
Share on other sites

Sorry if i havent read everything, as this is a lot of catch up to do, but has anyone actually suggested an increased cost to remove RNG system?

 

Basically all you'd be doing would be adding an item, or adding something to the recipe to increase the cost to lower/remove the RNG factor. Though this would have to be expensive, it would allow people who dont want to gamble their chance, or have extremely bad luck (myself included) to skip it. There are a lot of ways to modify this to suit the design, adding skills that allow this to change from reducing chance of failing, to removing it altogether.

 

This is actually done in a lot of f2p games via a cash shop, but obviously we're all against that. But such an item doent need a cash shop to exist and be balanced. 

Edited by Darknesse

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...