I feel obligated to reply, but since I'm no expert at roguelike timing (I can just read lots of rgrd archives 
), luckily I don't feel like I have to make insightful observations.
Though I would like to hear more about that powers-of, I don't quite follow how that works.
I don't think powers-of is specifically used by anybody since I haven't actually read about it being used prior to mentioning it in a few threads. To me, it's intuitive and obvious and I assume most people already use something similar for that reason-- I come from a GUI background, so user-experience and comprehension, to me, is always the most important aspect of an application.
A problem with roguelikes is maintaining chronological consistency between entities in a comprehensible way. That is, the player's perception of the rate at which objects moves may not reflect the way the engine works. A speed of 100 and 99 are virtually the same in regards to ticks, but on that 100th tick the player's inference is shown to be arbitrarily wrong when one entity seems to get a free action.
For example, if you have an Orc with speed 10 and a Unicorn with speed 11- every 10 turns the unicorn will appear to perform an extra action. This inconsistency isn't easy to communicate in a meaningful way to a player in a turn-based game. In a real-time game, you can have all the arbitrary timing things you want- but in turn-based, you just can't get away with it unless you have some really slick graphics (that communicate subtle differences of speed via tweening).
A simple solution is to base your timing system on powers-of some base. So that a speed of 0 = b^0, speed of 1 = b^1, speed of 2 = b^2- etc. This ensures that all aspects of the timing system align in a coherent way. This can allow you to implement asynchronous timing schemes without losing coherence (that is, turn-order can change, but the meaningfulness of each action is measurable). If you perceive that an orc makes actions at half the speed that you do, you can infer that a special ability that moves at half your current speed won't incur any hidden or imperceptible/unexpected penalty- like a free hit- which could be the case in the speed 10/11 issue.
In this sense, you can use powers-of with a tick-based system and priority queue and ensure that all elements of the timing system function in an expected way. This can have far reaching implications if you choose to simulate simple physics- like being able to step out of the way of an arrow if you can move fast enough- and other aspects.
It's really just about applying a standards to a tick-based system so that the player isn't forced to think about timing elements that only obfuscate the fun of the actual game. I personally feel that the benefit of a powers-of system is much better, because the rules are consistent. If one object moves 3x faster and another moves 5x faster- we're already getting into largest common factors for referencing the sensation of a complete 'cycle' of actions. If everything shares a common divisor, these inconsistencies won't exist. IMO- Flexibility is less important than coherence.
