moxilot and anonyme: it's called the bystander effect, and it happens a lot. Ironically, the more people witness an incident, the less likely any of them will do something, especially if none of them feel directly threatened.
I once saw a fight break out while I was working at Bumbershoot, and none of the dozen or so people watching, or the dozen or so more who could see what was happening as they passed by, did anything. Certainly none of them approached me, a festival employee (wearing a staff shirt, festival badge, and radio), to ask that I intervene. I had my finger on my radio button to call security when a couple of them happened to walk by. The two people involved in the altercation were ejected from the festival.
Most people aren't trained to react to violence in any useful way. One way of countering this if you're in a bad situation is to directly address individuals when asking for help. This often works. People won't intervene on their own because they're worried about the consequences of getting involved, and often don't know what to do.
Each individual is afraid that if they step in, nobody else will step in as well to help and they'll be left dealing with a violent situation alone. Even police officers get partners, and they're trained and armed. I would've hesitated mightily to get involved in an altercation involving five people unless I knew I had two or three friends with me.
Toward the end of the video, you do hear people shouting to the driver to stop the bus. It was an articulated bus and crowded, with the fight happening at the back, which means it's not beyond the realm of possibility that the driver had no idea what was happening until someone called his attention to it.