Bradd W. Szonye wrote:
>> However, this [suggestion] doesn't address the major objections I
>> have to most RPG turn systems (including HERO's). Most systems
>> feature frequent, inconsistent turn changes -- you take your turn in
>> bits and pieces, and you don't always follow the same player.
Klaus Æ. Mogensen wrote:
> How does Hero make characters take turns in more bits and pieces than
> other games? In fact, Hero allows characters to move and attack in one
> phase, which is more than many (most?) games allow in a round.
HERO isn't worse than most; indeed, as I wrote above, it's quite
typical. Most RPGs that I've played allow players to make one move and
one attack per turn. It's also typical for the turn order to vary over
time, because the turn order is semi-random (AD&D), because some players
get more "turns" than others (HERO), or both (Shadowrun).
D&D3's turn system is atypical: You set the turn order at the beginning
of a fight, so that you consistently follow the same player (except for
held actions), and faster characters get all of their extra actions at
once, rather than getting extra turns spread throughout the round.
In theory, giving more turns should improve fairness and waiting time,
while giving more actions per turn should improve overall combat speed.
This is a well-known property of scheduling algorithms in computer
science: If context (turn) switches are free, you can improve
responsiveness by breaking the workload into small, well-distributed
chunks, without hurting throughput (overall speed). However, context
switches generally aren't free, and RPG turn changes are no exception.
That creates a trade-off between overall speed and average wait times.
Also, if you break up the workload too much, all the context switching
takes so much time that throughput and latency both suffer. I'd need to
do the math to be certain, but I suspect that happens whenever the time
slice (how long it takes to take your turn) is shorter than the duration
of a context switch (how long it takes the next player to get started).
In my experience, that's exactly what happens in some RPG turn systems.
The turn length is very short (one move, one attack), and turn changes
take a long time (e.g., calling out action numbers, or frequently
re-rolling and re-sorting the turn order).
Even though the new D&D turn system chunks many actions together, it
seems much faster and more responsive than most RPGs, in my experience.
I suspect that's largely because the context-switching time is very low:
The "bigger" turns means fewer switches, and the simple order means that
I can just run down a list calling out character names. This is such a
great advantage that it outweighs the potential (but largely unrealized)
responsiveness advantages of other systems.
That said:
> The main feature of the HERO turn order is that characters with many
> actions don't get their extra actions top-loaded or bottom-loaded, but
> spread out over a turn. I think any decent replacement should keep
> this feature.
Correct; there are potential game-balance problems with clumped actions.
HERO has one other major advantage over other RPGs: The turn order is
deterministic. That means that you can prepare a "turn list" in advance,
calling out names instead of using the slow "Who goes on 3?" approach.
That in turn reduces the context switching time, which might actually
make the "small turns" approach viable.
Therefore, I think I'll try the standard HERO turn system (with prepared
turn lists) and see how it goes. I don't care for the extra bookkeeping
aspect of it, but it may be fast enough.
Thanks for the comments; they got me to thinking. I'm somewhat
embarassed to admit that I never thought of this in terms of scheduling
algorithms before, despite the fact that my day job includes working on
the scheduler for an operating system. Oopsie.
--
Bradd W. Szonye
http://www.szonye.com/bradd >> Stay informed about: [Fast HERO] Simplified turn order 
FAQ
Profile
Private Messages
Log in
