Since Raesh said he modified how FAE and teaching work, I did some testing.

TL;DR: With capped FAE, classes pulse almost twice as fast but teach the same amount per pulse. Which is odd, since it gives a scholarship boost to both teacher and student, which I would have expected to improve the experience per pulse. Maybe teaching is running off base ranks (which I would call a bug).


I started by testing pulse speed. For our control, a class without FAE pulses every 60 seconds like clockwork. I then tested teaching several different skills with FAE, and varied whether I started the class first or the enchante first. FAE was always cast at cap. Data available upon request, but here are my conclusions.

When teaching a class with FAE, we can divide the class pulses into two sorts: normal pulses, and pulses that accompany the message Bolstered by your music, your class becomes quicker and more effective (henceforth called bolstered pulses). Whenever you get a normal pulse, it occurs exactly 46 seconds after the previous pulse (it doesn't matter what sort of pulse the previous one was). Whenever a bolstered pulse follows a bolstered pulse, the gap between them is exactly 28 seconds. When a bolstered pulse follows a normal pulse, the gap can vary. I saw gaps of 2, 10, 12, 20, 28, 38 and 48 seconds.

I wasn't able to find a pattern in the sequence of bolstered and normal pulses. As far as I could tell, it was random whether any given pulse would be bolstered or normal. I also wasn't able to find a pattern in the gap size when a bolstered pulse followed a normal pulse. The net effect is that the timing of your class pulses is very irregular. However, the average time per pulse was always ~31 seconds. So that's almost twice as fast as a class without FAE.

Before the change, average pulse time with capped FAE was ~45 seconds. So this is an improvement.


Next I tested how much a single class pulse teaches. I would start a class for one pulse, then stop and wait until the experience had fully absorbed. I tested without FAE, normal pulses with FAE, and bolstered pulses with FAE. All taught the same amount (2.21% of a rank of Outfitting, in this case).

> I'd love to speed up classes by up to 2x just by area casting in the room

Sorry, only the caster's class gets bolstered pulses. All other classes in the area get normal pulses, but they're still getting the FAE speedup. That means a consistent 46 seconds per pulse, which means 1.3x speedup.

> I need to poke at FAE more. It looks like it's double dipping now and I haven't decided how to resolve it.

Yeah, I was pretty sure that wasn't intended behavior.

Any comment on the fact that the scholarship boosts don't seem to be increasing exp per pulse?

> Alright - this is resolved. I spend up the bonus in the core, but removed the chance for an instant pulse. Overall your experiences should be more consistent now (And better at low levels of mana). At higher levels it'll be about a wash and should be a little more flexible as well.

Cool. I'll test it out in a bit.

> I'd need to sit down and watch the example with debug code, but it's possible you're capping out the exp per teaching pulse before the bonus.

That seems plausible.

Unfortunately, it means there's no incentive for me train scholarship beyond tdps. Of course, scholarship's not unique in that regard.

> I don't know why my connection decided to suddenly soil the bed.

Found the issue (grumble grumble Steam deciding to start in the background and download Dishonored).

So it looks like the new formula is a consistent (56 - mana) seconds per pulse, except for at 1 mana where you get 56 seconds per pulse instead of the expected 55, for some reason. Vanilla classes are still 60 seconds per pulse, so that's a 66% speedup with capped FAE.

Also, I believe the way it's implemented is that every time the class pulses, it calculates the time until the next pulse. So determining the timing of a pulse depends on whether or not you had FAE up during the previous pulse.

> Your mechanics are right, your math is a little off. It's a 5-25 second reduction.

I don't know what to tell you; I was consistently getting 4-24. And I also did a test where I averaged over several pulses, to eliminate any effects of lag.

> One of us is looking at the code :) It's very clearly bounded between 5 and 25.

Fencepost error? You'd expect a measurement error to disappear in the average.

> Unless it's systematic. And it could be. Remember that DR is sequential code. That can do some funky things to timing.

But the measurement is being performed locally. Funky DR timing might affect the implementation of the time reduction, but not the measurement of it. The only way to have this sort of measurement error would be if I was experiencing 1 second of lag, then 35 seconds later 2 seconds of lag, then 35 seconds later 3 seconds of lag, etc.

I normally wouldn't do this, but, meh.

if (v3<5) then
set v3 to 5
else_if (v3>25) then ! Get some caps in place
set v3 to 25
.
set v3 to (60-v3) ! Final pulse rate

-Raesh

"Ever notice that B.A.'s flavor text swells in direct proportion to how much one of our characters is getting screwed?" - Brian Van Hoose

Hmm. Do you see the expected 55 second pulses at min? I'm wondering if somewhere we're not hitting the full 100% potency by a fraction, which would round it down to 24.

-Raesh

"Ever notice that B.A.'s flavor text swells in direct proportion to how much one of our characters is getting screwed?" - Brian Van Hoose

::squints::

I can only conclude that time passes differently in St Louis.

-Raesh

"Ever notice that B.A.'s flavor text swells in direct proportion to how much one of our characters is getting screwed?" - Brian Van Hoose

Sure. Let's go with that.

-Raesh

"Ever notice that B.A.'s flavor text swells in direct proportion to how much one of our characters is getting screwed?" - Brian Van Hoose