IMHO every article should have an abstract. When you submit a
paper, even if the journal does not require or even expect an
abstract, put an abstract at the top of the article.
Sometimes an outline or table of contents serves the same
Here's this is needed: Suppose somebody is trying to find
your article, or having found a list of citations is trying
to decide whether it is worth time and money to obtain a full
copy of your article. If they can see the abstract, they can
much more easily decide whether this is what they were looking
There are lots of indexes that include the abstract.
Secondly, nobody has time to read all the literature. So
suppose somebody is /skimming/ a magazine. The title entices
them to read the abstract, and then the abstract entices them
to read the whole article.
Thirdly, suppose I want to recommend the article. I like to
cut-and-paste the abstract, so that people can see /why/ I
am recommending it. It's an ugly waste of resources if I
have to write my own abstract for somebody else's article.
The same applies to talks. Every talk should have an outline,
then the main body, then a summary/conclusion.
---> Tell you students to do the same.
Do as I say, not as I do. On my web site, some of the articles
have an "abstract" or "overview" section, but alas many of them
don't. I'm working on fixing this, but it will take some time.
Process suggestion: Even though the abstract comes first on
the page, do not /start/ by writing the abstract. Write
everything else, and then come back and write the abstract.
If you want, you can start by writing a provisional abstract
setting forth what you /think/ the article is about. Then
write the article, then go back and write the real, final
abstract. In many cases you will find that the final abstract
is significantly different from the provisional abstract.