tl;dr: Alacrity is working correctly by the math (.66s recharge per 1s of time, or 40% cdr with permanent alacrity)
So Pyro tested it last BWE, and believed that complete alacrity was only reducing cooldowns by 33%.
This would mean that alacrity is adding .5s recharge for every second of time (it’s supposed to be adding .66s).
I was crafting a test to double-check this, by using a stopwatch or whatever, and I came on a better test idea.
1. Take a skill, specifically Chaos Armor, with no cast time.
2. Calculate the final cooldown with 100% alacrity, for both Pyro’s numbers and the numbers as they are supposed to be (33% cooldown vs 40% cooldown). Turns out it’s 23s by Pyro’s figures, 21s as it’s supposed to be.
3. Rack up more alacrity than the needed time, and let it tic down until you hit that number. Starting with the 21s estimate.
4. If the alacrity and the cooldown end at the same time, that’s the correct estimate.
So, I did a full rotation of wells and shatters, getting to 28 alacrity, and let it tic down. At 21s of alacrity remaining, I activated Chaos Armor.
Chaos Armor came back up right before the Alacrity ran out, meaning 21s was the correct estimate.
To double-check, I then tested it at 23s, and found that Chaos Armor came back up about 2s before I ran out of Alacrity, as expected.
So, unless Anet’s buff timers are malfunctioning, Alacrity is correctly adding .66s of recharge for every second of time, and full alacrity will give 40% cdr, not 33%.
Great news