Scientists track down the source of mysterious radio bursts

You can also apply for a "target of opportunity" when you don't know exactly when or where the event will take place, but there is a good chance that some event will occur over the next observing cycle. If you execute on a target of opportunity you jump to the front of the queue. If there's a particularly high time sensitivity (like for an FRB) then you might even interrupt an ongoing observation.

Things are trickier at a smaller observatory because they generally don't use queue observing. What happens instead is that an astronomer is granted some number of nights and they have the whole telescope for that amount of time. All you can do then is call them up and ask nicely. Usually they will though because they get on a paper without too much work. (The most highly cited paper I'm on happened that way!)

Yes they generally get bumped and they will generally get that time back later on in the telescope schedule. Folks who do these "target of opportunity" proposals apply like usual, but don't specify targets. If awarded time, they get the ability to "trigger" their observations once an interesting target appears. So their time allocation is (generally) capped like everyone else's, but isn't scheduled. So presumably the people who get bumped will have their observations rescheduled for later.

In detail the question of priority is a bit more complicated, and sometimes the people observing can refuse to be bumped for a TOO, but that varies with the observatory and their rules.

Edit: Also, people are frequently happy to even "donate" a little bit of their telesocpe time to observe interesting transient sources (sometimes in exchange for co-authorship or acknowledgements in papers, sometimes not). This is partly because astronomy is a reasonably friendly field. And it's also because we can't design and carry out experiments like other fields of science – we have to take what the Universe gives us. So it's good to take advantage of opportunities to view interesting and new sources when they happen.

http://www.astro.caltech.edu/observatories/coo/TOO/ToOpolicy...

You can ask for ToO time (http://www.astro.caltech.edu/observatories/coo/solicit/2016A...), but it has to be approved as part of the semi-annual solicitation for observing time. When you use it, you have to notify the PI you bumped and a committee, and later report back to them on the results.

Postdocs don't have to give up their observing for a ToO, but faculty do.

In Caltech's policy, interrupted observers don't get payback time.

They are said to happen more frequently than a few times a year. I'm guessing it's just that we are not detecting the bulk of them because they're randomly distributed.