Defenders quick attack speed
Is this still normalized to an exact time? I swear I see a difference in timing compared to for example, water gun and confusion. I'm sure I missed something somewhere along the line but no search I've done has provided information about it. Maybe it's connection?
Answers
My understanding is that mons in gyms have a 1.5-2.5 second delay added to each attack (both quick and charge). The delay is calculated and added on a per-attack basis, meaning that for each attack the delay is generated randomly. So, a quick move which hits every 1 second when attacking (say, Ember) will hit every 2.5 - 3.5 seconds (1 second plus a random 1.5 - 2.5 seconds) when defending (and the duration will vary for each attack.
Most people simplify and say there's a 2 second delay because that's the appropriate average given the above.
Correct and I think that's likely intentional to ensure that dodging is a bit tougher to do.
It's also why most people prefer quicker quick moves (i.e. with a cool down of 0.5 seconds), because they're guaranteed to be able to use 3 of them and still be able to safely dodge. (Since they've only used 1.5 seconds and the quick move takes some amount of time.)
Obviously, if you're using a slower quick move (say Zen Headbutt with a CD of 1.1 seconds), you are only guaranteed being able to use one of them before you have to worry that the defender's attack may come. If you try to squeeze in two, you will have used 2.2 seconds and it's possible that the attacker had a delay of 1.5 or 1.6 seconds coupled with a fast quick move like Lick which will hit you in less than 2.2 seconds.
That said, 2.2 seconds is still faster than the average defender attack, so the risk may be worth it. Particularly if you know the CD of the quick attack the defender is using.
The basic formula to determine the number of safe attacks to launch before thinking about dodging is:
FLOOR((Defender Quick Attack CD + 1.5)/Attacker Quick Attack CD)
Thus, a defender with Lick versus an attacker with Zen Headbutt:
FLOOR((0.5 + 1.5)/1.1) = FLOOR(2/1.1) = FLOOR(1.82) = 1
However, a defender with Zen Headbutt versus an attacker with Lick:
FLOOR((1.1 + 1.5)/0.5) = FLOOR(2.6/0.5) = FLOOR(5.2) = 5
So in the former case you can safely attack only once before you may want to dodge, whereas in the latter, you can safely attack 5 times before you should dodge.
HTH
Happy to help, but to be clear, those formulas aren't in the code (that I know of). It's just a way for me to explain myself in (hopefully) a clear way.
The reason I included "FLOOR" is that if you rounded the remainder up, you'd be assuming you had sufficient time for that last attack, which wouldn't be the case (1.82 seconds in the first example is not enough time to launch two attacks which take 1.1 seconds each, for example).
That would be very interesting if so. I know that there are "Damage Windows" for each move which vary by move, but my understanding is that the 1.5 - 2.5 second delay was random (and on top of) the Damage Window. I didn't include the Damage Window to the calculation specifically because it does vary by move (and only adds time, so the calculation above is still accurate, but not necessarily optimal).
But, if you find testing results that shows that the 1.5 - 2.5 second time is not random but is in some way related to the move, I'd love to see it!
You were right. My recollection was revolving around the fact that certain moves tend to be either longer or shorter, but the assumption is that some of these outlier moves (like blizzard) are bugged and not working as intended.
For some reason I also remember someone doing a test that Gyarados had more variability than others. But can't find it yet.
