1. The GM. If the GM goes away, the plot will have a lot of trouble continuing. Even if the GM remains, if they choose too lenient of a gaming style ('Monty Hall' GMs) it can get boring, and too sadistic or restrictive can kill interest.
2. Players. Even a great GM can't necessarily hold things together if multiple players leave. If enough leave, most of the city/NPCs will need to be redesigned (e.g., in FtF, since 4/5 of the original cast has left, most of the NPCs and folks that were developed in the city itself aren't really applicable anymore). At some point it becomes more useful to just do another 'setting the scene' round where the city and characters are setup. There's also the issue of where good 'jumping in and out' points are; if players leave during a sequence where it's not plausible for new players to be added, the whole thing can bog down and make it take longer to get to a point where new people can enter, which increases the chances of game death.
Another note on players -- players whose posts are hard to read/use poor grammar can drag down a game, simply because others won't always be as inclined to read or respond to such posts.
3. Consistency of posting. The longest game that I'm aware of on the board is the chat game, which meets twice a week at set times to post. This ensures that the game happens.
4. Setting. Some settings don't lend themselves well to being RPGed. It can work at first due to novelty, but eventually the logistical difficulties are too strong to be overcome. (This can happen in games set in other fictional universes, e.g. a Potterverse game or a Death Note game, because the magical objects in those games don't really lend themselves well to the DFRPG.) Other problematic settings are those where the main characters have to compete against each other (e.g., in the extreme, a Battle Royale game) because this makes it harder to work as a team or accomplish much of anything significant, and games where characters don't have well defined motives to continue being in the game (i.e., no explanation of why they don't quit and go home).
5. Particular difficult scenes/characters/etc. If a game features something really new, it'll take more work on the part of the GM, which can make the GM want to drag their feet a bit as that scene approaches. They might also be forced to pause the game if they can't stat or create something in time. (E.g., in my game, the characters may eventually need to fight armored vehicles, which are challenging to stat because there aren't examples in the books. So, if I didn't stat those ahead of time and suddenly the characters had to fight a Blackhawk or a battleship or something, I might have had to pause the game to hash out the stats, maybe even run a combat test or something). Long enough pauses can derail things.
6. Inclusion of gimmick elements. If Slenderman shows up out of nowhere, great, but it just got a lot harder to run and play the game because you've got a gimmick that risks overwhelming the whole plot. This can be mitigated if the gimmicks are built into the campaign (e.g., in the above, a major adventure, designed from the beginning, is dealing with Slenderman), but then it's not really a gimmick anymore.
7. Mismatched setting/character level. Some settings are better for, say, 'Chest deep' campaigns than 'Snorkeling' or 'On the Beach.' If the plot is 'the characters are generals in Winter's army as they march against summer', statting them as 'On the Beach' will limit what they can realistically do against Summer's forces. On the other hand, 'Snorkeling' characters are liable to walk all over, say, the characters from some of Harry's earlier adventures, and this can make the game boring.
