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

"Quality Bootstrapping"

Recommended Posts

When Crafting in Crowfall, you don't always have a full set of same-quality high-quality materials to work with. It is possible however to 'bootstrap' the quality of a component or item up, by using some pieces of the higher quality and the rest of lower quality. I recorded an example of this recently:

 

 

There is some anecdotal evidence that suggests certain components are more heavily weighted in this regard than others, but no testing-based evidence as of yet...

Edited by Anthrage

Share this post


Link to post
Share on other sites
15 minutes ago, Vectious said:

This is why i feel that quality levels should have a quality 'number'. So when you add two different qualities you average the result. 

So a green and purple would net a blue.

I've seen that behavior, once.

Put some blue mats with a white coal and ended up with a green.

 

Share this post


Link to post
Share on other sites
13 minutes ago, KrakkenSmacken said:

I've seen that behavior, once.

Put some blue mats with a white coal and ended up with a green.

 

White brings quality value down significantly, I would guess less than 20% of items using blue and white mats end up green.   One grade down for 1/3 of mats is my rule of thumb...


6FUI4Mk.jpg

                                                        Sugoi - Senpai

Share this post


Link to post
Share on other sites
14 minutes ago, KrakkenSmacken said:

I've seen that behavior, once.

Put some blue mats with a white coal and ended up with a green.

 

In crafting vessels, I always get blue no matter the components color, so long as one item is blue. Sample size of 2. :P

But it could be that not all crafting professions are treated the same way. I have definitely observed what Anthrage is describing and I have an empirical "feel" for the results.

Share this post


Link to post
Share on other sites
5 hours ago, Anthrage said:

There is some anecdotal evidence that suggest certain components are more heavily weighted in this regard than others, but no testing-based evidence as of yet...

Obviously anecdotal, but: I made three swords with mixed qualities, each time with the grips (6x leather/wood) a grade higher than all the ore components (90x ore), and two out of three got bootstrapped up to the leather/wood's quality. This is a tiny sample pool, but the unusual luck stood out to me regardless.

Consider that it has to succeed two rolls in a row for this to happen, and each time your quality components have to bootstrap 2x lower quality ones: the two wood components have to beat 4 metal components to make a higher quality hilt, and then the hilt has to beat two more metal components to make the higher quality finished sword. On the one sword that didn't work out, it actually did succeed the first roll to make a higher quality hilt and only failed on the last roll; so I succeeded that roll against the odds 5/6 times.

If combining one higher quality and two lower quality components has a 33% chance to get a higher quality result, as one might naively expect, I calculate my odds of getting at least 5 successes out of 6 rolls at 2.9% 1.78%. That's not unthinkably unlikely, but it's enough to make me suspicious. [Edit: math. Been a long time since I took statistics]

Edited by Avloren

Share this post


Link to post
Share on other sites
2 minutes ago, Avloren said:

Obviously anecdotal, but: I made three swords with mixed qualities, each time with the grips (6x leather/wood) a grade higher than all the ore components (90x ore), and two out of three got bootstrapped up to the leather/wood's quality. This is a tiny sample pool, but the unusual luck stood out to me regardless.

Consider that it has to succeed two rolls in a row for this to happen, and each time your quality components have to bootstrap 2x lower quality ones: the two wood components have to beat 4 metal components to make a higher quality hilt, and then the hilt has to beat two more metal components to make the higher quality finished sword. On the one sword that didn't work out, it actually did succeed the first roll to make a higher quality hilt and only failed on the last roll; so I succeeded that roll against the odds 5/6 times.

If combining one higher quality and two lower quality components has a 33% chance to get a higher quality result, as one might naively expect, I calculate my odds of getting at least 5 successes out of 6 rolls at 2.9%. That's not unthinkably unlikely, but it's enough to make me suspicious.

Agreed. In my above example, bootstrapping occurred where the higher quality materials consisted of 50%, 40% and 33% of the whole - 3 in a row, all on the same item, with successful bootstrapping. Criminal small sample, highly anecdotal evidence, but there it is.

Share this post


Link to post
Share on other sites
1 hour ago, Anthrage said:

Agreed. In my above example, bootstrapping occurred where the higher quality materials consisted of 50%, 40% and 33% of the whole - 3 in a row, all on the same item, with successful bootstrapping. Criminal small sample, highly anecdotal evidence, but there it is.

Further note: say I'm wrong about 1 purple + 2 blue components having a 33% chance to go purple. Say for example the quality results are split even, 50% blue/purple, for an item that combines blue and purple in any ratio (so no advantage to using 2 purple/1 blue vs. 1 purple/2 blue). That would put the odds of the results I saw at 10.9% instead of 1.78%, which is a lot easier to excuse as mere luck.

Small sample size, can't draw any conclusions yet, etc., just saying there are plenty of ways the math could work that we may be able to narrow down with further testing. I'm going to start writing all results down next test, if I can stand to craft more. I gotta admit, potion timers and durability of picks are trying my patience.

Edited by Avloren

Share this post


Link to post
Share on other sites
3 hours ago, Avloren said:

Further note: say I'm wrong about 1 purple + 2 blue components having a 33% chance to go purple. Say for example the quality results are split even, 50% blue/purple, for an item that combines blue and purple in any ratio (so no advantage to using 2 purple/1 blue vs. 1 purple/2 blue). That would put the odds of the results I saw at 10.9% instead of 1.78%, which is a lot easier to excuse as mere luck.

Small sample size, can't draw any conclusions yet, etc., just saying there are plenty of ways the math could work that we may be able to narrow down with further testing. I'm going to start writing all results down next test, if I can stand to craft more. I gotta admit, potion timers and durability of picks are trying my patience.

Quite frankly it doesn't matter if it's 1/3 or 1/10, the cost savings with BP's once you build a bootstrapped one is going to far out pace any risk of loss for lesser materials.

Imagine the difference for 100 perfect blue bars that only cost 3 blue metal and 6 green in the BP.  300 - 900 just does not even compare.

Share this post


Link to post
Share on other sites
19 hours ago, Vectious said:

This is why i feel that quality levels should have a quality 'number'. So when you add two different qualities you average the result. 

So a green and purple would net a blue.

I like how it is. It makes for some interesting choices. I, for one, find it generally not worth it to use large amounts of ore and stone to make high quality tools, as the gain doesn't seem that significant. However, if I just use one blue or purple log, I can get 1-2 blue harvesting tools. It also takes materials from and puts output at all tiers of the market, creating a more healthy system. Crafters that can get better experimentation results reliably, though, would want to use a lot more blues, because ultimately whites wouldn't be taking advantage of their skills.

Still, it might call for changes when the market is fully implemented and we start seeing people trading for things.

Edited by ringhloth

Share this post


Link to post
Share on other sites

I really do feel like you aren't really tricking the system by doing this. I'd imagine theirs a good deal of information retained during the creation process in terms of the quality of materials used, and the grade of each item used. The end result doesn't really feel random to me.

The only possible way to know this for a fact, however, will only be available once blueprinting comes online. At that point we could experiment with the exact same materials and components at each stage to know how these interactions work.

Due to all the variables involved, mapping this out right _now_, is impossible. A single grade difference might tip the balance.

I think the person who used the word anecdotal sums up the conclusions in this thread quite nicely.

Share this post


Link to post
Share on other sites

I made 15 tool sigils today: 5 picks, 5 hammers, and 5 axes.  I used blue parchment on all. For 2 of 5 picks I used green ore for sigil but all others used white ore. I netted 2 blue sigils--one with green ores and another with white, so there does appear to be some opportunity to bring up white components.  

For the stone tablets, I had limited blue/green mats but made 3 blue tablets on 3 attempts using 2 blue/1 green stone.

 

Share this post


Link to post
Share on other sites

This has been a 'thing' since i started testing.  I did some more (incidental) testing of this this weekend.  Out of 22 combines (14 stitched leather, 3 padded leather, 5 metal bars) I intentionally used 3 varying qualities of leather or ore.  Out of 14 stitched leather ((blue or green) + white + grey) , I had 2 come up blue/green, 1 come up white, 11 come up grey. 2 padded came up grey and one white. 5 metal came 1 green, 4 grey.

I did more combines than this, but this was a specific experiment and had results mirroring past results.

The long & short of it is, you CAN do this, but I feel it will be more useful for blueprints, but the risk is very high, as one material at any stage can drop the entire quality level of your craft. 

 

It's not hard, nor complicated, it's just the system heavily favors the lowest quality.  This has been tested & re-tested & re-re-tested time and time again.

Share this post


Link to post
Share on other sites

I did some larger scale testing of this over the weekend. I might format the excel sheet nicely and share it all at some point, perhaps after I get enough data and draw some conclusions. For now I'll just type out the tests I had a decent quantity of (20+).

I tested using the unmarked stone runestone recipe (the one you use for gathering tools). It's a good recipe because it only uses 3 stone, 1 per stack, so they can all be of different quality. I ignored failures. Greys are cobblestone, white+ are various advanced stone (didn't record type), mostly slate. Results:

Format: Total# Input/Input/Input -> # result, # result

102x White/Grey/Grey -> 45 white, 57 grey

100x White/White/Grey -> 51 white, 49 grey

56x Green/White/White -> 13 green, 43 white

21x Blue/White/White -> 5 blue, 3 green, 13 white

Edited by Avloren

Share this post


Link to post
Share on other sites
On 3/13/2017 at 2:31 PM, Avloren said:

I did some larger scale testing of this over the weekend. I might format the excel sheet nicely and share it all at some point, perhaps after I get enough data and draw some conclusions. For now I'll just type out the tests I had a decent quantity of (20+).

I tested using the unmarked stone runestone recipe (the one you use for gathering tools). It's a good recipe because it only uses 3 stone, 1 per stack, so they can all be of different quality. I ignored failures. Greys are cobblestone, white+ are various advanced stone (didn't record type), mostly slate. Results:

Format: Total# Input/Input/Input -> # result, # result

102x White/Grey/Grey -> 45 white, 57 grey

100x White/White/Grey -> 51 white, 49 grey

56x Green/White/White -> 13 green, 43 white

21x Blue/White/White -> 5 blue, 3 green, 13 white

Hm. It's difficult to make judgements without more info. To me, that suggests that if an item has two different rarities, each rarity has a percentage chance based directly on the ratios of the rarity. If the item would be bootstrapped up by more than one rank, another roll is made, with an equal chance of each quality being picked (there was, or is, a bug where quality bootstrapping results in rarities outside the materials, so this might be where it comes from, where there's no floor for the quality picker). Still, that's not really an exhaustive set of data.

Share this post


Link to post
Share on other sites
On 3/13/2017 at 2:31 PM, Avloren said:

I did some larger scale testing of this over the weekend. I might format the excel sheet nicely and share it all at some point, perhaps after I get enough data and draw some conclusions. For now I'll just type out the tests I had a decent quantity of (20+).

I tested using the unmarked stone runestone recipe (the one you use for gathering tools). It's a good recipe because it only uses 3 stone, 1 per stack, so they can all be of different quality. I ignored failures. Greys are cobblestone, white+ are various advanced stone (didn't record type), mostly slate. Results:

Format: Total# Input/Input/Input -> # result, # result

102x White/Grey/Grey -> 45 white, 57 grey

100x White/White/Grey -> 51 white, 49 grey

56x Green/White/White -> 13 green, 43 white

21x Blue/White/White -> 5 blue, 3 green, 13 white

Thanks very much for reporting your results...

 

What is your runecrafting skill?  It would be interesting to see whether this has an effect.

if it is low, I might repeat this to see whether I see a statistically significant difference...


WAZ6Fov.png

"The cinnabar is a lie"

Share this post


Link to post
Share on other sites
4 hours ago, narsille said:

Thanks very much for reporting your results...

 

What is your runecrafting skill?  It would be interesting to see whether this has an effect.

if it is low, I might repeat this to see whether I see a statistically significant difference...

I have no crafting training or gear, only potions. I was also curious about the effect of skills, so I did half the grey/white tests with potion on, half with potion off - I left that part out of my summary because it didn't make much difference, although you can judge for yourself. Here's the more detailed breakdown:

52x White/Grey/Grey (potion) -> 20 white, 32 grey

50x White/Grey/Grey (no potion) -> 25 white, 25 grey

50x White/White/Grey (potion) -> 23 white, 27 grey

50x White/White/Grey (no potion) -> 28 white, 22 grey

The tests with greens and blues were all with potion.

 

I can't draw any real conclusions based on the small sample (so far), but if I had to guess:

The white/grey results are closer to 50/50 than to being directly proportional to the ingredients (i.e. 33/67). The results are slightly better with more white in the mix, but not as much as you'd expect if it was directly proportional. The formula could be based on ingredient proportions, but skewing them closer to 50/50 somehow?

The green/white/white skews towards lower quality results, it was consistently hovering around 25% green 75% white throughout testing, but it needs more trials. I need to do some green/green/whites, but that's harder to do in quantity for obvious reasons.

The blue/white/white is a tiny sample, but it does give a nice spread that's very close to directly proportional to ingredient quality. If you assign numeric values to qualities - say white (1), green (2), blue (3) - then the ingredients average 1.67 quality and the results 1.62 quality. More trials needed obviously, and I'd like to do some blue/blue/white, blue/green/white, blue/green/green etc. but scarcity of green+ stone makes this very difficult.

If crafting skill (aka potions) is having any effect, it's to reduce quality of results, which I find hard to believe unless it's bugged.

Edited by Avloren

Share this post


Link to post
Share on other sites

Thanks for sharing your results.  I found it interesting your blue/white/white recipes yielded 3 levels of product.  Given the starting resources for that sample you increased the quantity of non-white products from 7 to 8--paid for by a drop in blue quality from 7 to 5.  

 

Edited by Nasmyth
Updated info

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...