Sometimes, when one speaks at a conference, the conference organisers request that you write up a paper for the proceedings of the conference, which then get published somewhere. It seems to be increasingly the case that many people don't like writing conference proceedings for a variety of reasons, with the main one being that they are considered low-prestige places to publish, and consequently there are various drivers not to bother wasting time in writing them, or wasting interesting results that could be published somewhere else. There is even pressure from on high not to write conference proceedings, coming from the REF culture. I tend to write conference proceedings for those conferences which ask for them. For one thing, it's not hard to write about the work that one has been doing, and usually (at least in my case) I can show some calculations that are interesting enough examples, but something that doesn't make the cut for a paper destined elsewhere. They can also be good practice for PhD students to be involved in writing the proceedings and getting some publications on their CVs. Also, it's a sign of collegiality that if my friends and colleagues can be bothered to organise a conference (which I know from experience can be quite time consuming) then I can play my part in making it a success, and write the request proceeding article, sticking it on the arXiv, too, so that anyone can read it (indeed, one went up yesterday from me for a conference in Sofia in October)
Still -- I certainly don't mind when conferences decide not to include proceedings. They are probably an idea that has had their day. What seems to be replacing them, though, is that conference organisers want to put talks up on the web after the conference. This seems to me to be a total waste of time, unless one writes the slides that accompany one's talks to be completely self-contained, which is really a contradiction, since then there would be nothing to say during the talk. I tend to put things on slides that I can't say in words (pictures, graphs, movies etc) and say most of the words myself, rather than have people read them. Since most of the talks get put up as pdf files, movies don't really work at all, but what's missing is the commentary and the context. I really find it next to useless. Attached to this post is a screen grab from one talk of mine that has been put on the web by a conference organiser. The grab shows a snapshot from an animation of a calculation of the collision of two uranium nuclei. A snapshot of an animation is a poor proxy for the whole thing.