EQdpm is not a totally broken yardstick.
It's not a yardstick at all. And that's my point and that's why the discussion is continuing. Perhaps I should have said it's not a micrometer which is the only thing that would settle this discussion (read real data here).
"Worse than useless" describes an untestable hypothesis...
Would you settle for equivalent to useless? What P99 hypothesis can you test with EQdps? When one is hypotesting, one requires relevant data. EQdps cannot falsify a hypothesis on P99.
"So eqdpm and p99 have a transitive relation. You act like Eqdpm is a random die."
In a transitive relationship, the curves would be the same shape and one could extrapolate accurately. This is not the case with EQdps. For the purposes of this thread, I'd say that EQdps is worse than a random die for the reason that people put unwarranted confidence in it (for this question). Were we to roll a die and give an answer, at least no one would have any confidence in the conclusions.
"You don't seem to get the precision/accuracy thing in this situation, but I also see that I didn't explain myself well. EQdpm uses an equation that, because it is not an utterly failing model (use it and tell me its data does not correspond somewhere around "very well" with your experience), collects data that shows some amount of precision. It's outputs have undeniable correlation with the "real" values and with each other, although we can be dead certain they are not accurate. I guess I should have said "an equation that describes reality precisely but inaccurately" instead of "a precise but inaccurate equation?" "
I'm not misusing any terms.
You don't seem to get that the way you used accuracy and precision (and the link that you provided) pertain to measurement and not modeling. Nobody I've worked with would conflate these two uses. Those are two different animals. If you me to agree to use the words in the same way that you for this conversation, I'll go with that and give you that EQdps will ballpark okay-ish. But the question of this thread requires real data to make a determination. Ballparking is not good enough. If it were, we'd be done.
I figured I didn't even need to refer to your offhand to shoot down the idea of a 1h combo involving a 9/20 weapon outdamaging a tstaff. There is no level after 20 where that can happen.
Pure assertion. Conflicts with what I have experienced.
The combat system will never treat different starting delays as equivalent when hasted, because this would utterly break itemization. Assuming there is a delay floor, haste would calculate as a percentage of the difference between weapon delay and the floor instead of as a percentage of just your weapon delay. This would ensure that a 9/20 is never as good as a 9/16.
Support this, too. I don't see any reason why the floor would not be a hard stop.
I don't know for sure which setup is better, but I can say for sure that monk epic fists are not some kind of undisputed champ of the dps mountain.
Support this with data.
____________________
It is an empirical question. And with Tasselhofp99's latest post, it gets murkier.
So let's test it.
I have a L58 monk epic-ed with CoF.
I propose I try 10 min runs on the PoD as well as a L50-ish player character (thinking this will give an insight into variation in some way) with the following loadouts:
Fist/ADC
SBH (is it warhammer?) - anyway, the 9/20 and ADC
T-staff (if I can find one to borrow)
IFS (if I can find one to borrow)
RFS (if I can find one to borrow)
Max haste with chanter or shaman haste. No other buffs.
No kicking, but lotta clicking to keep the fist haste fresh.
I'll post the results (and a link to the data) in a day or few.
Will separate procs from the dps but will included them in the dataset.
Any other suggestions for the test? What am I missing and is there anything I'm thinking about wrong in the proposed methodology?
Anyone going long on T-staffs and IFS? Could be a value changer.
|