I've been using them more and more as a kind of pseudo-code to improve readability. @Edward has said he's been going in the other direction, using @other more. I mean, something like this speaks more to the program flow than spelling out all the details inline:
<< init >>
<< load data >>
<< restructure data >>
<< create formatted output >>
<< write output >>
Yes, you could make all those sections into functions or methods of the same name, but that can add complication and take longer. Or it could be be done in a later refactoring if that seems worthwhile. or if the sub-block might be useful somewhere else, too.