Dismiss Notice
This Section is READ ONLY - All Posts Are Archived

Crafting: Why do you bother?

Discussion in 'Release 34 Feedback Forum' started by Duke Death-Knell, Oct 8, 2016.

Thread Status:
Not open for further replies.
  1. Arlin

    Arlin Avatar

    Messages:
    379
    Likes Received:
    603
    Trophy Points:
    43
    That's not how probability works.
     
    Preachyr, NRaas and Olthadir like this.
  2. Aklys

    Aklys Avatar

    Messages:
    39
    Likes Received:
    62
    Trophy Points:
    8
    Gender:
    Male
    Location:
    Australia
    Your success rate on the 12 items over all is 33% I believe, Your success per items is a different story. If you have a 83% chance of creating an item each item is irrelevant to the others' success or failure. You have a 17% chance of failure on each time you try to make an item. Now the odds of you getting a failure for 8 items in 12 would likely be a pretty phenominal odds (this is the extent of my calculations at 6 in the morning) which is why I think the system is definitely broken. But we need to be accurate with where it's broken for it to be hopefully fixed.

    But someone who knows statistics even better could probably give the exact calculations.
     
  3. Arlin

    Arlin Avatar

    Messages:
    379
    Likes Received:
    603
    Trophy Points:
    43
    The exact calculations are: you don't understand probability either. Even if I started believing his numbers, he has an extremely small sample set, we don't know the distribution of the algorithm, and while 8 out of 12 failures at 83% is unlikely(1 in 10,000) it's far more likely to be a cherry picked result with an algorithm subject to clustering than the developers being unable to call a random function correctly.


    EDIT: Let me give an example. Assume the following is masterwork attempts at 83% success rate. An X indicates failure and an O indicates success:

    XXXXXXXXXXXXXXXXXOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

    That is a success rate of exactly 83%, but due to clustering you can cherry-pick your 12 trials to make it look like 0%:

    [XXXXXXXXXXXX]XXXXXOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

    33%:
    XXXXXXXXX[XXXXXXXXOOOO]OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

    or 100%:
    XXXXXXXXXXXXXXXXX[OOOOOOOOOOOO]OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
     
    Last edited: Oct 11, 2016
    Olthadir, Aklys and Lord Barugon like this.
  4. Duke Death-Knell

    Duke Death-Knell Avatar

    Messages:
    1,751
    Likes Received:
    1,825
    Trophy Points:
    125
    Gender:
    Male
    Location:
    Philadelphia PA area
    I know how probability works. But unless your willing to give me 500 gold ingots I can only gauge on what I have. But y'all know that so the standard defense is to fall back on "small sample size". But the fact is even 100% fails. There is an issue with the RNG. I see it with my rapidly disappearing reagents which only have a 7% chance of being used, I see it harvesting.
     
    uhop likes this.
  5. Aklys

    Aklys Avatar

    Messages:
    39
    Likes Received:
    62
    Trophy Points:
    8
    Gender:
    Male
    Location:
    Australia
    You are probably right I don't have the best understanding. I was just trying to demonstrate the difference between the statistical reference to a bulk of outcomes vs a single action being different. You can have an 83% success rate on an action that doesn't mean you'll have 83% success rate represented in the number of successes over a number of attempts (if not poorly).

    Such as if you pull a card out of a deck and expect to get a club it's 25% chance. And if you return that card to the deck and attempt again you have the same odds of success but to do it two times in a row the odds will be considered different and go down to being 6.25% chance (I think that's right).

    It depends on the calculation utilised if it has knowledge of the prior attempts. But I don't know the answer to whether it does or not. That's all i was trying to say may be why we see multiple failures in a row and that component itself isn't broken just misunderstood.

    Hope I'm making sense but maybe I'm also just crazy :)

    EDIT:

    The example was nice by the way but that's basing that the success rate is based off knowledge of prior attempts. I agree if it is based on that then clustering can misrepresent the output of an algorithm. But if it's based off each attempt then you could have 90% of those fail even if it is pretty extreme odds for that scenario to happen.
     
    Last edited: Oct 11, 2016
    Preachyr likes this.
  6. Aklys

    Aklys Avatar

    Messages:
    39
    Likes Received:
    62
    Trophy Points:
    8
    Gender:
    Male
    Location:
    Australia
    Yeah this definitely seems like a bug or a misrepresentation of the algorithm IMO. I'm guessing there's like a critical failure decision where you fail no matter, but then we possibly shouldn't represent 100% success rate if that is the case. Not sure why it fails on 100%.
     
  7. majoria70

    majoria70 Avatar

    Messages:
    10,347
    Likes Received:
    24,869
    Trophy Points:
    153
    Gender:
    Female
    Location:
    United States
    And another thing missing with say a 83% rate and failure is why. Now if you randomly fail that doesn't make sense, but why. Nothing is added to make it engaging. If you are failing because of a broken tool or because you are starving or your finger slipped and got cut that is one thing and still you should get some of your ingredients back, not everything can be damaged in the process.

    Also I just have to mention, what is the reward for trying (besides the great fear of destroying all your hard earned resources), will you have a chance to discover a recipe (like in wow crafting you would get a reward of knowing how to make a better heal from making heal), sure you may get a skill gain but is that all? IMO the risk and rewards are not there in sufficient amount to make it as fun as it could be or for some people fun at all. In real life crafting, more glue could help or stronger wire could help, and I have messed up things by being hungry and shaky. So I take a break and nourish myself ;)

    There has to be more than this, say it isn't done. Most of me knows crafting is not done, but part of me wonders if they are going to add more features and events that would help to make it more fun and a better system. The times when it says add an extra coal is on the right track but it fails to do what it should.
     
  8. Arlin

    Arlin Avatar

    Messages:
    379
    Likes Received:
    603
    Trophy Points:
    43
    I don't have to give you 500 gold ingots because I've used significantly more than 500 silver ingots leveling masterwork blacksmithing, and have seen zero indication that the success rates are wrong.
     
    Preachyr likes this.
  9. Aklys

    Aklys Avatar

    Messages:
    39
    Likes Received:
    62
    Trophy Points:
    8
    Gender:
    Male
    Location:
    Australia
    Agreed.
     
    majoria70 and Lord Baldrith like this.
  10. Waxillium

    Waxillium Avatar

    Messages:
    3,311
    Likes Received:
    9,043
    Trophy Points:
    165
    Gender:
    Male
    Location:
    Rift's End
    If you're not power grinding there is a time when crafting is the best way to increase survivability.
     
    Bluefire and Dwalin Bombardin like this.
  11. Duke Death-Knell

    Duke Death-Knell Avatar

    Messages:
    1,751
    Likes Received:
    1,825
    Trophy Points:
    125
    Gender:
    Male
    Location:
    Philadelphia PA area
    Well yay for you. But there are quite a few of us who seem to get all your bad rolls.
     
  12. uhop

    uhop Avatar

    Messages:
    237
    Likes Received:
    400
    Trophy Points:
    18
    That's a very good example. Imagine that you masterworked 500 items in the row, while the screen says that the probability of it to succeed is 83%, and all of them failed. Is it even possible? Sure thing. "But if you try it a million times, you will see that the ratio is exactly 83%". It is called "distribution", and this is precisely how it manifests itself with a binary process.

    (Personally, I am on the "bad streak" side of the fence --- practically all my attempts at the 1st MW fails, and I get exceptional items rarely. Given that I suspect that the proverbial 83% success rate is hardcoded, and the real rate is much lower. I don't mind: I am a pretty low-level crafter. I do mind the incorrect number.)

    The idea is to provide a distribution, which is correct, (pseudo) random, and produces small streaks. To illustrate that, let's imagine a binary process, where both outcomes are equally possible (50%) --- sounds simple enough. The infinite sequence 101010101010(10) fits the bill, it is totally fair, no unbelievable streaks, yet it is not really random, isn't it? Being non-random opens stuff up for various exploits. In this particular case, RNG should generate pseudo-random strings longer than 2. Yet, imagine 10 zeros in a row --- how probable is it? Given our assumption of equally possible outcomes, the probability of that is 1 / 1,024, or ~0.1%. 20 times in a row? 1 / 1,048,576 --- one millionth! The problem as I see it, we have a lot of streaks, which are humanly improbable, making players justifiably upset.

    I can't say I saw 20 misfires in a row, but I bet I saw 10 or more, while fighting, and trying to use spells (I use more spells on a daily basis, than I craft items, both use RNG). I see this problem every single day, if I go adventuring (or mining these days). And usually I use spells with probabilities (as indicated) >50%. Another thing I have noticed that misfires usually come in batches. It is pretty rare that a spell has fizzled, then worked the very next time (as I would expect). Typically there is a misfire streak, and if I spam as fast as I can usually all spells fizzle for several seconds. Possible? IMHO, yes. Probable? IMHO, no.

    I suspect that another side of it manifests itself like that: time to time in a middle of a fight I have no spells in 1-5 slots for a long time (these slots are unlocked, but a bunch of glyphs are assigned to them). Again, it looks highly improbable.

    PS: There is a simple explanation for that: all probabilities we see are bogus. I suspect that people complaining about "unfair" crafting has lower level than people, who see no problems. Another point is: the probability itself tells just a part of a story, its distribution is important, and it looks like it is far from optimal. I posted about that in RNG follies, but it went unnoticed.

    So if anybody wants to see, if RNG works fine, it is not enough to count 0s and 1s. Count streaks, and verify their probabilities to see if they correspond to the common sense (players tend to use it), or any other reasonable criteria.
     
    Bluefire, Olthadir, Preachyr and 4 others like this.
  13. Arlin

    Arlin Avatar

    Messages:
    379
    Likes Received:
    603
    Trophy Points:
    43
    This is the post I don't have the patience to write. To expand on it, I seriously doubt anything is hardcoded(except caps to success rates - exceptional doesn't seem to go above 25% for example) and it isn't quite as bad as 500 failures in a row. I've had bad streaks just like everyone else, but as I've said, every time I sit down and measure it with a large sample it comes out more or less correct. The actual code probably looks something like this:

    Code:
    x = (calculated threshold based on skills)
    if(rand(seed)< x)
        Success!
    else
        Failure!
     
    Preachyr likes this.
  14. Barugon

    Barugon Avatar

    Messages:
    15,679
    Likes Received:
    24,294
    Trophy Points:
    153
    Gender:
    Male
    It was the same complaints in UO. Nothing changes.
     
    Olthadir likes this.
  15. Duke Death-Knell

    Duke Death-Knell Avatar

    Messages:
    1,751
    Likes Received:
    1,825
    Trophy Points:
    125
    Gender:
    Male
    Location:
    Philadelphia PA area
    After a few more tries I'm now at 14 attempts 9 failures. That's a lot of wasted effort, this game is suppose to be fun but I guess fun isn't high on portalariums list of things this game should be.
     
    Bluefire and uhop like this.
  16. Dwalin Bombardin

    Dwalin Bombardin Avatar

    Messages:
    78
    Likes Received:
    232
    Trophy Points:
    8
    Gender:
    Male
    Location:
    Las Vegas, NV
    OK all I know is, I'll never look at 95% the same way as before I got into SOTA :oops:
     
  17. Weins201

    Weins201 Avatar

    Messages:
    7,121
    Likes Received:
    10,958
    Trophy Points:
    153
    This is how the whole game should have started SLow, long, drawn out for the gathering / making of this stuff.

    To much achieved in too little time has led to the "I WANT IT NOW" mentality and a good lasting game has basically been put in the crapper with the hand on the handle.

    this is what is wrong not what is being given , it is "it should be easier" Is the problem.

    They should have mage everything like this sadly to much to fast and now. . . . . .

    we shold barley be able to have a gm or two in the first few months - LMAO.


    Yepper Items wil never break nor are they lost - and if they every try and implement that type of system the player base will go nutz like they did over some skill loss upon death and the developers will cave in and give the player the mindless play they have nor.
     
  18. The Hendoman

    The Hendoman Avatar

    Messages:
    574
    Likes Received:
    894
    Trophy Points:
    63
    Gender:
    Male
    Location:
    Asheville, NC
    STRIKE!

    Lets strike. Nobody craft a fakken thing. like, just make the handles and straps and hardened mats but hoard the mats and not craft the big stuff.

    STRIKE!

    THE HENDOMAN
     
    niteowl57 and Aklys like this.
  19. Aklys

    Aklys Avatar

    Messages:
    39
    Likes Received:
    62
    Trophy Points:
    8
    Gender:
    Male
    Location:
    Australia
    That's virtually what I've been doing :p
     
  20. Blightlord Knightmare

    Blightlord Knightmare Avatar

    Messages:
    53
    Likes Received:
    87
    Trophy Points:
    8
    Location:
    Storm's Reach
    It looks like you are in the minority with that thinking, but I wholeheartedly couldn't agree with you more than I do.

    If anything, the system is far too easy as it is, absolutely waaaay too easy. You shouldn't be able to hop in, and just within a month or two be pumping out a mastercrafted GM weapon or armor piece every single week. (In some cases people are actually crafting multiple +7,6,7,8,9,X items in a week).

    Outrageous really. When I go to a vendor and see a bunch of +10 copper two-handed or an entire meteoric armor set and its from the same person and they obviously re-stocked in less than a week, I get so depressed. It should take months or years to be a real master-crafter, not a couple of lousy weeks. If everyone is a mastercrafter, then no-one is really.

    I like that some higher-level trainers require quests to unlock slots/glyphs, but I think there's more that can be done.

    There's a way to get rid of the massive grind for crafting. There's also a way to make it so that you also don't have to craft 500 chairs/staves/string/etc to level up your expertise. You can do this simply by modifying a couple of things:

    • - The higher crafter level you are, the less "xp" you should gain from lower-level items
    • - Have the items take far more time. A chair should take a couple real-world hours to create. A table should take a real-world day.
    • - When you begin construction on an item, have it go into a background queue so you're not stuck at the station for those 6 hours or whatever, you can even log off and when you come back, boom there it is, nice and completed, just like when sending companions on SWTOR missions. It could even be something you have planned to do after adventuring during the day and before you logoff.
    • - Gain A MASSIVE TON more XP than regular from crafting something. E.g., if its your first time making a gem-socketed +3 axe, that took 2 whole days to make, then it should bump you up a massive amount of XP. Then you're not flooding the market trying to off load hundreds of axe's you're grinding out just to make exp.
    • - Items should be able to be made more unique by adding certain materials to them. Every single 1H Copper Sword or Bear Hat +2 shouldn't be the same. You should be able to add an optional reagent/ore/whatever to any item. Have the graphics team pump out 20 different color shades to apply to weapons/armor/milled goods so that light green, sky blue, obsidian black, etc. tables and chests can be crafted. A table made from Pine wood should have a different hue than one made from Maple and so on.
    • - Have actual failure rates go down far more. It should be outrageously difficult to fail, very low chance. This will be fine, because it would still take a week for like a +11 Polearm to be crafted, so the timesink is there, and the danger is still there but the frustration is taken away.

    So we'd see less item dumps at every vendor, and more unique items. Everyone would get a real sense of accomplishment when something is crafted and crafters could increase what they charge for their items.

    We shouldn't see such a minuscule difference between a +2 and a+3 (or even a +3 and +8) in terms of cost that only a couple of sad extra 1000 gold means that people will ignore the lower + items. It should be a major undertaking for someone to save the gold required to purchase a new item going from say a +1 to a +2 item. That should be a major purchase, forget about being good enough to just outright buy a +3 or +4.

    Perhaps the level of a weapon should also be dependent on your skill to lock/force lower + items to be considered. For example if your Blades is only at level 64, then you should only be able to use a +6 sword, if your Bludgeon is at 20, then only a +2 shall you use.
     
    ThunderSAJ and Rampage202 like this.
Thread Status:
Not open for further replies.