I always think one of the big probs with using twitter, facebook etc for storytelling is that if audience member misses the start of posts/tweets it pushes them 'out' of the experience.
He raises a good point, but it's not something restricted to social media like Facebook and Twitter, nor is it something that can't be solved. Ultimately, what it comes down to is point of entry.
A point of entry is simply a point at which a potential audience member can enter the storyworld and begin to (forgive the term) engage with the property.
In a single-medium story, let's use a novel as an example, there is only one real point of entry -- the start of the novel. You're not going to pick it up, flip to page 147 and begin reading, and if you did, you wouldn't have any reasonable expectation of being able to understand what was going on. However, the point of entry for a novel (or movie) is not time-specific. I don't need to start reading on a certain day to follow the story, I can read it whenever I want.
Let's look at another single-medium example, a television show. If the show features a long-ranging storyline, then it too only has one real point of entry: the beginning of the series. However, TV shows are (or should be) designed such that if you started watching partway through a series, you could still enjoy a given episode and begin to figure out what's going on as you continue, allowing for multiple points of entry. (With some particularly formula-based shows, you could even come in after a commercial break and pick up the story of an episode.)
Television is more time-specific, however. With new episodes airing every week, a viewer is pressured into watching at a specific time, and at a specific pace. (DVDs and services like Netflix are, of course, changing this. I rarely watch shows as they air anymore, preferring my own schedule and pacing.)
In a transmedia property, the landscape begins to look a little different. If your storyworld is presented across many single-medium stories (a movie, a book, a graphic novel, a video game) then each story only has one point of entry, but the whole property has as many points as there are stories.
The trick with points of entry, though, is this: in order for it to be a point of entry, it needs to be a place where an audience member can begin their experience of a story and still get the full experience. In the above example, each single-medium story must be a worthwhile experience on its own. If you want the movie to be a point of entry, it cannot rely on prior knowledge of the book or video game. Each point of entry must stand on its own, or it isn't a point of entry -- it becomes a barrier to entry, in fact.
Things get more complicated, of course, when your story is even more integrated across media -- unfolding in real time on Facebook or Twitter, for example. Paul Burke's concern was this: if I start telling a story through twitter interactions, there is a high risk that there is only one point of entry (the beginning of the story) and it is time-specific. If someone misses the start of the real-time interactions, there is quickly far too much material to catch up on to understand what's going on. It quickly becomes a barrier to entry. (I've written before, this has been my critique of ARGs for a long time -- there's so much material usually that joining after the start is very difficult.)
One solution, however, is to have some kind of updated summary of the story so far, constructed in a way that makes it as easy as possible for new audience to catch up and then jump in.
The better solution, though, is to make it possible for any point in the story to be a point of entry. If the story is unfolding on Twitter, make it possible for someone to browse the last few dozen tweets, then start watching in real time and be able to figure out what's going on. Moreover, any point in the story needs to be equally engaging so that people who do come in partway through decide it's worth thrit time to stay.
The Pandemic transmedia project started this weekend, and one of the main features (for people not at the Sundance Film Festival) is a twitter stream at #pandemic11. It features a whack of fictional characters (that's a technical term) that are basically just tweeting about their lives as they slowly realize Something Is Wrong (cue story). A lot of the twitter stream is basically white noise, very typical tweets about life and school and whatever. The benefit of this is that every tweet isn't integral to the plot. I can browse the last few dozen, and get a sense of what's happening (more people are sick, more characters are starting to realize there's a real problem...). I don't need to spend hours reading every single one. I can if I want, I can start to get a better understanding of each individual character (very cool, by the way, how each is unique and a full character, not just Another Twitter Account), and also start to uncover the ARGlike puzzles. But I don't have to in order to still get an understanding of the story.
The trick in any transmedia property is to have as many points of entry as possible. You want to make it easy for people to get involved in your story, because they aren't going to put in any excess time or effort without knowing what they're getting out of it. This goes whether you're dealing with Hollywood films and blockbuster video games, or an indie creation using blogs and Twitter. It's not restricted to social media -- but it's also not a problem that lacks a solution.
Look at your property from the audience's point of view. Would you be able to get into the story? Would you want to spend the time required? If not, rethink it.