Crafting Level 6-8 Decons

[Kinerd]

I believe I have experimentally determined the following points:
1. level 6 decon - 24d3 lesser
2. level 7 decon - 32d3 lesser
3. There are widespread graphic bugs saying which level decon is which.
4. Greater decons are crazy.

.

We were told with update 15:
level 5 decon - 16d3 lesser, chance of 1d5 greater
level 8 decon - 40d3 lesser, chance of 2d4 greater
Each + value higher than +5 will yield a progressively higher amount of raw materials up to a +12 value.

We were told with update 9 patch 1:
Deconstructing low tier effects on high-end loot will now return more reward essences.

.

First, I recorded data for 316 level 5 decons, 318 level 4 decons, 318 level 3 decons, and 335 level 2 decons in the u9.1 era, getting average lesser values of 35.41, 18.16, 9.32, and 4.31. I also recorded that of these 100%, 88.8%, 69.5%, and 38.5% were from "high-end loot". The % increase over expected lessers from these decons were 1.11, 1.15, 1.24, and 1.20 respectively, for an average of 1.17. If I were to guess, I would guess that there is a 2 in 3 chance to use d4s instead of d3s, which would increase the lessers gained by:
[(2/3) * xd4 + (1/3) * xd3] / xd3
= (2/3 * 2.5 + 1/3 * 2) / 2
= 2/3 * 1.25 + 1/3
= 7/6 = 1.16

Next, I recorded data for 77 level 6 decons, 87 level 7 decons, and 23 level 8 decons in the u15 era, getting average lesser values of 63.51, 81.83, and 102.3. (100% of each were from "high-end loot".) Applying the 1.16 figure, this suggests base values of 27d3, 35d3, and 44d3.

However, I saw massive, shocking outliers for 6 and 7:
-I have received 110 lessers from a level 6 decon. Even if we supposed 27d4 and got the full value of 4 on every single roll (which is spectacularly unlikely), we only get up to 108. It cannot be overstated how impossible this return is for a level 6 decon.
-I have also received 135 lessers from a level 7 decon. This is at least technically possible with 35 * 4 (140), but again it is spectacularly unlikely. There's a reason no one has ever seen a Lightning Strike proc do 420 or 800 damage on a standard target.

Once I saw this data, I started recording specifically which effects correlated to which levels, and here is the (as yet incomplete) list I have compiled:
Level 6: Stat +5, Incineration, Steam, Magma, Smiting, Acid Torrent et al, Improved Cursespewing and Destruction, [Hiding, Jump, Search, and Open Lock] +13.
Level 7: Stat +6, Greater Twilight, Disruption, Banishing, [Listen, Intim, Swim, Open Lock] +13, [Open Lock, Disable, Eloquence, Repair, Hide] +15.
Level 8: Ambushing +2, Vorpal, [Focus, Repair, Hide] +15.

My theory is that [some? most? all?] Skill +13s are in fact level 7 decons, [some? most? all?] Skill +15s are in fact level 8 decons, and the graphical evidence to the contrary is a bug. I also suspect that Thaumaturgy staves fall into the same camp, but it so happens I have acquired few useless specimens of those in this period so I don't have the recordings to determine. I think this artificially inflates the above empirically determined values and that instead of 27, 35, and 44, the true values are the more intuitive 24, 32, and 40. (I personally have never seen a value above a level 8 decon that would demonstrate the possibility of inflation, but a guildy of mine swears he once saw a level 9.)

.

Greaters are even weirder. Because I have such a paucity of level 8 data I will focus on the 6-7 for this portion. Of the 34 recorded level 6 decons that produced greaters, I have 2 2s, 2 3s, 8 4s, 9 5s, 8 6s, 3 7s, and 2 8s. This produces a pretty gorgeous match for 2d4 (avg value 5.06 vs. 5)... except 2d4 is supposed to be the level 8 greaters!!! The high-end loot and graphical bug effects put together are nowhere near enough to explain this.

The level 7 decons are more of the same: I won't break down the bins, but they range from 4 to 11(!!!) and give an average of 6.79 over 34 recordings. (2d5? 3d4?)

The average return I got when greaters were returned for level 5 decons in u9.1 (which therefore includes the high-level effect) was 3.16 against an expected value of 3 over 99 recordings. Also worth mentioning is that in 0 of those 99 records did I see a value higher than 6 greaters, yet in my admittedly cursory recording since update 15 I have seen a 7 from a Greater Good decon (nominally level 5).

[RabidApathy]

First of all, thanks for gathering this data. I really ought to keep better tabs on my deconstructions (instead of just taking screens of oddballs). I have also noticed some strange outliers, especially with Thaumaturgy Staves, probably due to them having more deconstructable properties than most loot.

This is one of my more recent ones that left me scratching my head.