Disclaimer: this is entirely a personal viewpoint, formed by a few years of publication in a few academic fields. EDIT: Many of the comments are very worth reading as well.
Having recently finished a very rushed submission (turns out you can write a novel paper in a day and half, if you're willing to sacrifice quality and sanity), I've been thinking about how academic papers are structured - and more importantly, how they should be structured.
It seems to me that the key is to consider the audience. Or, more precisely, to consider the audiences - because different people will read you paper to different depths, and you should cater to all of them. An example of this is the "inverted pyramid" structure for many news articles - start with the salient facts, then the most important details, then fill in the other details. The idea is to ensure that a reader who stops reading at any point (which happens often) will nevertheless have got the most complete impression that it was possible to convey in the bit that they did read.
So, with that model in mind, lets consider the different levels of audience for a general academic paper (of course, some papers just can't fit into this mould, but many can):
The least important audience. An interesting title may draw casual browsers in, but those likely aren't very valuable readers. Most people encountering an academic article will either be looking for it, or will have had it referred to them from some source. They will likely read more of it. So the main role of the title is to not put off these readers, and to clarify what the paper is about, and what field it belongs in. Witty titles are perfectly acceptable, as long as it fulfils those criteria. So in-jokes for the whole academic field are perfectly acceptable, in-jokes for a narrow subfield are not - unless you're not aiming beyond that subfield.
The most important audience of all. Most people reading a paper will only read the abstract, and will then proceed to dismiss the paper or accept it and move on. The abstract thus plays three roles:
- It presents the paper's results. The abstract must be crystal-clear on what the paper says; abstract readers must be able to describe the results correctly.
- It establishes the credibility of the result. It can do this by briefly outlying the methods used, and by its general tone. It must thus be serious, and use the correct vocabulary for the field. No room for impressive rhetoric here - dry and descriptive is the model of the abstract.
- It can draw the reader into looking into the paper proper. Because of the first two points, it cannot achieve this by teasers or rhetoric. Instead it must present strong results that cause the reader to want to read more.
This audience will skim through the paper to see what it says. Most crucial for them is the introduction and, depending on the field, possibly the conclusion or discussion section. These must tell the skimmers everything there is to know about the paper - what the problem is, what the results are, what methods were used, why these results are valid, why they are important. As long as all these points are covered, rhetoric and wit can be used, in moderation, to make the reading more enjoyable and salient. But be careful to use these in moderation, lest you give the impression that the paper's results depend on rhetorical tricks. Rhetoric is the flavouring, giving out the information above is the main goal.
These are those readers who will go through the whole paper, though they may skim some parts along the way. The important thing here is to get the structure absolutely clear - it must be easy for them to see what the crucial steps or arguments are, what implies what, what relies on what. To do this, lay out the structure of the argument and of the paper clearly in the introduction or in the second section. Emphasise the important results through the paper (consider the layout for this, it can often be used to draw attention to the main points), and connect them together ("combining this with the results of section 2.3x.iii..."). Some rhetoric can be used around these important results, especially if it emphasises their importance.
These are your greatest fans or your more hated critics. They will go through the whole paper, taking your argument apart to understand it completely and figure out how it ticks. No fancy rhetoric for them, just careful attention to detail, clarity, and rigour. In mathematical terms, these are the people who will be reading the proofs of your minor lemmas. Don't waste space with anything that doesn't help you establish your argument or your results. These are the lawyers among your readers, looking for the tiniest of flaws. Don't give them any of these, and don't try to hide them with weak arguments.
Writing the paper
The different audiences above give a structure to the paper, but they can also give a structure to writing process. Looking back, I realise that I start by writing for the full readers, getting the important points and structure correct. Then I fill in the details for the deep readers. I then write the introduction (and conclusion, if appropriate) for the skimmers, and conclude with the abstract for the most important audience. The title can be chosen at any point in this process.
Hope this helps! I think I've been following this advice implicitly for a long time, and it's got me a few publications. Feel free to ignore it, of course, or to post your own preferred approach.