I surmise that the speed of communication now exceed the speed of publication (or even the speed of the typesetting/proofreading cycle). So, you prove something, send the rough draft to people who, in your opinion, may be interested, start preparing it for publication, and in the middle somebody answers one of the questions in your paper. The best thing is if his answer makes your paper obsolete. Then you throw it to the garbage and waste no more time on the boring writing routine. But more often than not the answer is given to some not so important question that doesn't really change much but still adds an interesting twist. To publish it separately would be ridiculous. To ignore it would be strange. So what to do? Well, it becomes an "appendix". Now, the gentleman's way is to offer the full co-authorship. Also the gentleman's way is to reject it. Now we are stuck with either an extended thanks, or a separate "appendix authorship", and here goes. I believe we'll get even weirder things like "This paper supercedes the work of X who improved the result of Y, which was a generalization of theorem by Z proved in responce to my question (all 4 private communication)" if the speed goes up a bit more. I personally am in favor of pushing the gas pedal to the floor and shaking the current ideas of priority and authorship off entirely but I know that my mathematical opinions are very non-orthodox...