I've run and written con games. I just ran a six hour one-shot of the Feng Shui starter scenario for my group. The biggest thing is making sure there's a fulfilling experience in the time allotted. Here's things to do and to watch out for to run successful one-shots, the "Five P's."
You want to either provide pregens or have people do chargen ahead of time, if they have the patience for that. Even games that boast fast chargen, it's eating into your time.
Know the game you're going to run, know the adventure you're going to run. In other formats, if you fumble around in one session you can just pick up in another. In the one shot format, anything you do that's not "running the game efficiently" harms the experience.
If it's a new system to your players, try to get them to read something on it ahead of time. Like for my Feng Shui one shot, I put a rules summary PDF they provide in our Dropbox for them to read (they didn't of course).
Different games have different activities take different amounts of time. You don't want to stick to one plot format as it gets boring - I got turned off of RPGA Living games because all the adventures, to fit into the 4 hour slot format and meet XP goals, ended up having an extremely formulaic "2 fights 1 puzzle 1 rp scene" format. The less you understand the game and/or the more variance there is on how long it takes things to happen, the less well you'll be able to plan for the slot.
There's no "one" plot just like there's no "one" short story construction. You'll want a variety of plot types, too much of the same thing will turn people off (and won't really show off the different genres). "You should have a problem arising for the first hour..." is something you should do once, don't repeat yourself. Event based adventures are the best to fit into the format; you can have timed events happen at designated times regardless of the players being "ready." Puzzles are the worst; players can spend 1 to 100 minutes arguing and wrangling. Freeform roleplay can always just fill the time allotted. Sandbox can be good or bad - if the entire point is to just wander around and do some stuff, it's fine, if you have some goal in mind then sandbox may well push you way out of your time slot.
Other plot types, you should prepare for in terms of connecting one scene to the other and having contingencies so that you can cut pretty much all but about 3 scenes. You can safely rely on 3 scenes being must haves in a 6 hour slot, and 6 being the most reasonable, and if you have 6 then you need a contingency to cut up to 3 of them and have things make sense. Watch the clock and know "OK we're in hour 4 and we're in teh middle of scene 3; I need to cut scenes 4 and 5."
In short form adventures like this, don't waste time screwing around or being too precious. Too many twists and turns, or waiting too long to reveal what's really going on or who's really the bad guy, can be a turnoff. If you're playing Buffy the Vampire Slayer or Night's Black Agents as a one shot, if you only interact with vampires in the last scene, you're not really giving people what they expect out of that single dose.
Most of this an experienced GM can handle easily with timing. As you watch the clock, you can accelerate (or, more infrequently, decelerate) the action by brusquer roleplay, "a guy with a gun busting in through the door," opponents in combat breaking morale, use of speedier rather than slower tactics (entangle and summon? There goes 3 hours). The specifics depend on your game system but it's the same technique. Spin out detail as long as you're on time; cut to the chase, cut bits, cut scenes when you're not.
Know your players and how many there are. More players means more time per person and more complexity and you'll get through less content. Specific people may be fast or slow by nature. You can try to mitigate the slower ones in whatever way covers over their particular damage but in general you just have to account for it. "Oh, Chuck's in this session, I'm going to have to Taser him to get any roleplay scene to stop."