Magic Craftsmanship and Orb Change

[Griss]

And the quest to find out what the fuck magic craftsmanship is for marches on.

[Raldo]

I love how people doubt what's been officially posted.

[pompey_dan]

Then where is the hard evidence of people saying for example darksteel nuggets are standard synthesis, felt is magic craftmanship, most people still seem undecided or have different veiws on it,

I thought the dev post said that some synthesis will favor magic and others favour normal, then they could have meant a synth which you are trying to hq will be magic whilst a synth you are trying to just finish way above your level is standard. Thats just what i think, how about u actually try it out before blindly believing a vague dev post.

[Kestiel]

I don't advise anybody actually do any testing on what the stats do further until 2.0 comes out. We've been told the stats will change and be obvious as to their use. We've got probably a few months left of the game and any comprehensive testing would take a large chunk of that time.

The fact that people at one point were rampantly going about the crafting community saying mag. craft = orb color just goes to show that people say whatever they feel they see (which is often wrong). Our brains are hard-wired to see patterns, it helped our ancestors stay alive, stay well fed and rise to the top of the foodchain. So while one person sees Darksteel nuggets as needing high control (i've seen this suggested multiple times) another says craftsmanship (also seen this). I'd advocate balanced stats as high as you can go until 2.0 as the limited tests I've seen, and done don't show a massive swing in success rate/progress gain either way.

Please keep in mind you need ~1,067 samples to have relevant data, a few hundred is as good as worthless and eyeballing/gut/instinct is equally useless. Let's advocate proper testing rather than "he said, they said, dev said" please.

[tymora]

I don't advise anybody actually do any testing

Yep, can't be bothered at the moment.

a few hundred is as good as worthless

Not really worthless if you are only going for a general pattern and not a good value. Or it can be used as a starting point for further testing.

As above, too close to 2.0 to worth doing any detailed testing.

[Kestiel]

Worthless in the sense it doesn't give you an accurate representation and often will show you a trend that doesn't exist. It's why the usual accepted margin of error is 3% (which requires 1,067 samples) for most polling results to be considered "statistically significant". Giving a launching point though, I agree completely.

Edit: I felt I should elaborate a little. If you only do 100 samples your margin of error is a massive 9.7% which pretty much makes your work wasted since your white orb may be seen as ~16% while your prismatic could be around 34% drastically skewing your results and feeding you false information. Of course the percentage drops by around ~2% more when you get to 200 samples, and then falls to around 5% error at 300 samples. Which still could show you White at 20% and Prismatic at 30% leading you to believe something is significantly different than the truth. Of course had I stopped at one trial I could have been inclined to believe that the orbs were slightly in descending order from White->Prismatic since it seemed to line up that way. Or at least I would have had I not kept in mind that there was a 3% margin of error.

Point being, I don't think general patterns are helpful when there's little you can extrapolate in most things. If you're going through the trouble of testing, you might as well test properly. If you're okay with skewed results I'd have to question why you'd bother testing to begin with (Not "you" specifically to anybody, just the general use).

For this, knowing the orb rates is only mildly helpful. The true purpose was to see the effect of stats on orbs, and in that case a swing of even 5% could have led one to believe the orbs were influenced by one stat or another, more than a 3%. In an ideal world I would have done 9,000~ tests to get a 1% margin of error, but I'm not that masochistic.

[Green_and_a_Half]

when I craft class weapons, or any HQ items, I don't do 100 synths to get it.
I do 5 for the weapons/armor
and I do 1-2 for the HQ other stuff
So setting gear to what I think works Is what most people do.
Ive never had mag craft over 70 when hqing any item
until people started saying double craft lvl is best, now its at 120
same with crafting, never been over 100 until then, now its 120
and fro control, always ran with 150 but now its 130.
And still able to HQ the same amount of items.
I'm glad there are testers but ful 50 with "Crafting" gear on WILL make HQ items
Happy crafting
and not knocking any of the testers, thanks for all the work.

[tymora]

Haha, can't say I don't agree with Green on this.
TBH, Orb ratio isn't really an area of interest I am concerned because it isn't anything I can control and it only affects one skill...Magnum Opus.

I am more interested in:
- Action Success/Great Success Rates on each Color (or failure rate if you take the inverse) and how they vary by craftsmanship and DLvl.

-> You'll need way more than 1067 samples for those. As it is, my couple hundred samples do confirm my own observations on their variations. While I can't give you a formula on them or decide whether craftsmanship trumps stats or the reverse, I am able to do my crafts comfortably with the partial knowledge.

My best records so far are +1000 quality on Natron (47 lvls above) and + ~830 quality on Cobalt Ingots (I still do get ~400 quality or so once in a while, but I have learned to accept those "destined" to screw up synths) and I am pretty happy with that.

---------------------
My observations tallies with Green's: 120/120 craftsmanships should be enough to cover the requirements for most if not all current recipes in game.

---------------------
Not bashing on the need for a detailed analysis of stuff but most people just don't have the time to do it.

Maybe if I can finish this crafting parser I am working on...lol