Thanks for running these Ashaton. I had hoped to be more directly involved with the thread, but got busier than I expected.
Bunch of thoughts, many of which have already been addressed. The ones that I don't think have:
1) I didn't say that buffing Sacred Duty to 30% would definitely fix it. I hadn't run any sims to test it yet, so I was just estimating based on the rating conversion. Under-estimating obviously, since it took ~40% to get there. But the concept was solid: if the reason haste is weak is due to rating conversion changes, buff rating conversions (either directly or indirectly).
2) In any event, as nice as the 40% SD is, I doubt it's going to fly just due to class symmetry. I'll propose it, but I don't think they'll be keen on the idea. That said, I agree that it's probably the most balanced of all of the solutions.
3) I think Divine Riposte could be made to work depending on how things go. While it's good, it's not necessarily too good. I don't think we require the sort of stat parity a DPS class has. Presumably other tanks don't anyway. That said, we have more knobs than you're using.
First, I think you're still using the parry version there, yes? Parry's DR coefficient is significantly more generous than dodge's. It's entirely possible that the Dodge version of Divine Riposte would not scale as well with gear, which might give us the perfect solution.
Second, diminishing returns coefficients are class-specific. Since we have no other sources of dodge, we could actually treat that coefficient as a free parameter and adjust it specifically to control the value of haste across gear levels.
I think that Divine Dodge, potentially with an adjusted dodge DR coefficient, should be our next attack vector. I can show you where those constants are in the sim if you need help finding them.
4) I think the SotR mitigation angle is good too, but has more problems. First, it's not that intuitive. Not that mastery->mitigation is, I guess, but mastery is more abstract. Haste has a very specific connotation, and "adds mitigation to SotR" isn't it. That said, you can probably make the same argument about Divine Dodge, so....
The other problem with the SotR mitigation angle is that it's harder to control than Divine Dodge. There's no obvious way to curb the scaling you're observing.
That said... there is one way it can be counteracted: successive nerfs. They've done this before, so it wouldn't be hard for them to e.g. reduce the scaling coefficients to 0.004 or 0.003 in the next tier to keep it from getting too strong. I just think it's a less elegant solution than Divine Dodge with appropriate DR. In my experience, the devs really like elegant solutions.
5) Rather than estimating gear sets, you can actually just scale the ilvl of the T17 gear sets. The
scale_to_itemlevel command lets you set the item level of each item of the gear set. So for example:
- Code: Select all
scale_to_itemlevel=700
will automatically scale each item to ilvl 700 when you sim, giving you appropriate stats for an ilvl 700 gear set. Of course you have to guess at the ilvls of future raid tiers, but that should be relatively easy to ballpark (and we don't need to be all that exact, to be honest).
6) I'm not really all that worried about the odd haste breakpoints. Again, i think they're just artifacts that should be ignored. That said, if you want to try and smooth them out yourself, there are already two settings in the sim that you can use to attempt that.
On the Options -> Globals tab, in the first column, there's a "Player Skill" selector. It defaults to Elite, which means the actors don't make mistakes. Dropping that to "Good" will cause the actor to occasionally slip up - there's a small chance they won't use an action that's available even though the APL says they should, which usually means they'll accidentally use a lower-priority ability. The further you drop that skill setting, the worse the player performs (and I suspect, the smoother that haste curve gets).
Also on the Options -> Globals tab, in the third column, is a "World Lag" setting. If you read the tooltip, this artificially extends the cooldown of skills by a random (normally-distributed) amount based on the selection. Adjusting this should vary the fillers available somewhat (the /wait lines after CS and J should maintain the CS-J-X-CS-X-J-CS-X-X cycle though).
Even though we may not be interested in results that are extremely "sloppy", seeing what those plots look like may tell us something about those haste anomalies.
One thing I intend to do is add an "advanced" Scale Factor calculation mode. Instead of calculating Scale Factors from two sims (one at +0, one at +350; or +/- 175 if you have Center Scale Data selected), it would run seven or nine at intervals (e.g. 0, +/- 100, +/- 200, +/-300) and average the results. This should give a more representative Scale Factor since it's essentially just sampling more points on the plot.
