I have two points to address here. The first is a proposal or hypothetical idea, and the second is a question about one detail of that proposal.
First: what if day/night cycles mattered in an MMO game world? What could you do with that goal? How might you make them matter? Part of the line of thinking that I'm following here is the idea that disruptions to a simplistic static stability are a good thing. This doesn't mean the world is inherently unpredictable; the basic mechanics are still there, players just have to learn and adapt to a world that is in motion, rather than standing still.
In fact, having various things like the time of day and weather affect things like combat could make it easier to do away with simple random number generators for things like dice rolls. In other words, if there are a host of conditions in the world around you that could modify the outcome of various things such as a spell, is there as much need for relying on simple randomness? Instead of throwing a virtual die, and deciding what happens when that mage throws a fireball, how about looking at the sky, seeing that it's raining, and deciding that his fireball fizzles down to half it's starting size by the time it hits the target?
We can discuss weather another time; this is about day night cycles. I came up with a list of some basic things that could be modified by whether it's day or night out.
- Dodging (can't see it coming, can't dodge)
- Aim (can't see the target, can't hit as easily, especially if they're far away)
- Animal movement (nocturnal creatures come out, such as wolves hunting)
- People sleep
- Guards become more alert (more dangerous, and attacks are more common at night)
- Magic (light-based magic stronger during day, weaker at night)
- Movement (slower at night, faster during day)
I'm sure there are more creative effects out there, but those are some simple things off the top of my head.
Now for the question: How long should a day/night cycle be? My thought is that it shouldn't follow the real world's time. True, it's a nice feel to see it night outside for real when its night game, but if the world is going to change based on this simple cycle, I think it'd be better to shorten the cycle so that someone playing at the same time each day could experience different times.
To this effect, I determined that a multiple of 5 hours per cycle would be a pretty good number. Why? Because multiples of 5 do not divide evenly into 12 hours. In fact, it's a regular unevenness. If you're curious, just go look at the remainder, add that into the next division, and see the remainder after that. If you keep going, you get every remainder possible from 0 to 4 hours if you divide by 5 hours. Different multiples have different remainders, but they all have the same pattern: the same time in RL each day will be a different time of day in the game world, and the pattern won't repeat until at least 5 days later.
Note, a 5 hour day/night cycle would mean 2.5 hours for a day, and 2.5 hours for a night, while a 10 hour cycle would be 5 and 5. I nearly confused myself while writing this, and ended up having to replace a bunch of 2.5s with 5s, since I definitely don't think having 1.25 hours each for a day and night would be a good idea. An hour just seems a bit too short. I could be wrong though.
The downside to this shortened cycle is that it makes coordinating times between players a bit harder. You could have a "game time" and "server time" to make it easier, or just let players figure out their time zones on their own. The other potential problem I can think of is that 5 hours is so short that players could get disoriented, or upset at the fact that the world is changing so quickly on them. This could be a good thing, or it could be a bad thing. I think it is a potentially useful mechanic to force players to vary their playing, and keep them on their toes.
That aim can be viewed as a negative thing if players don't like the implied hostility directed at them. I'd rather this wasn't a mechanic that was only popular in Darkfall-style games. However, I think forcing players to adjust regularly could be a good way to help keep them engaged in the game. Plus, if you have 2.5 hours for a day, a large battle might last for a whole (or large part of a) game day. This would create certain potential imperatives to finish fights before factors change in a way that might put the initiator at a disadvantage.
Perhaps 5 hour cycles could do that, but perhaps 10 hour cycles would be better.