Some papers are guaranteed to fail because we try to write on topics that are far too broad, too narrow, or inherently illogical or in other ways problematic. When you choose a topic, think realistically about what the chances are of producing a coherent and viable paper of the appropriate length on the topic.

Decide upon the level of scholarship at which you will operate. To start with the most demanding level, papers that are publishable in scholarly journals are both original (the material and ideas have never appeared before) and exhaustively researched (the author knows and either can or does cite all previous relevant publication, in whatever languages). Of course, such an ideal is hard to attain, and it is in fact often not attained, even in scholarly journals. Nevertheless, even in an undergraduate paper it is incumbent upon you to know in general the bibliography on your topic, to know more or less the state of knowledge in the area, and (especially) to cite any ideas that you use that are not your own.

Most teachers of writing suggest using an outline. The advice is good, but sometimes it is easy to start outlining a paper before one is ready. A helpful thing to do before outlining is simply to sit down with a pencil and paper, and write down everything that comes into your mind that you think is relevant to the topic and that you might want to include. "Free associate" as you write ideas down, so that one suggests another. You can weed out superfluous and irrelevant ideas later; the process described here is helpful in getting a critical mass of ideas with which to work, and often in developing a sense of how they should be organized.

Having an outline is useful, and even essential. But don't trap yourself with it. In the process of outlining we are frequently not aware of details of an argument that become apparent in the actual process of writing, and that demand that the original outline be scrapped and that we start over. Sometimes it is helpful to start with a general outline, then gradually flesh it out in more and more detail until one gets close to the eventual prose of the essay itself; this process might make continual writing and throwing away of new outlines less likely.

It may help, in the arduous process of moving from either a general or detailed outline to the writing itself, to imagine yourself telling a friend or teacher what your paper is about; or imagine that you are teaching a class on the same material. Doing so may help to put the materials in order and to achieve logical succession and continuity, as well as suggesting actual sentences and turns of expression.

The best writers often write quickly and painlessly, at old manual typewriters. (Two of the most accomplished essayists whose work I know, Richard Taruskin (music) and Stephen Jay Gould (history of science, evolution) work this way.) For most of us, however, the task involves continual rewriting, reformulating, and correcting. For those of us who write slowly, it sometimes works––assuming that we have achieved a reasonable command of the material and general organization of what we are doing––to force a fast writing: that is, to refuse to dally over particular words, sentences, transitions, and so forth, and just get everything down as fast as we can, as though we were taking a timed test. This process often produces a credible first draft, which may have a good sense of continuity since it was done quickly. (Of course, it may also be utter chaos and completely worthless; use your own judgment.)

As you write, try to imagine for whom you are writing, and keep that ideal reader in mind. One of the most difficult aspects of writing is establishing context; how much do you have to say to keep the reader oriented, and what can you leave out? There are no hard and fast rules here, but one must always keep the issue in mind, and, whatever decision you make about who your reader is, stay with it consistently.

State your principal thesis early in the paper as precisely as possible. Stick to the thesis, and make sure that every succeeding paragraph elaborates, develops, or demonstrates it. Each paragraph should have a clear topic sentence that focuses in on one aspect of the subject. No paragraph should be shorter than three sentences or longer than a page.

Corollary to the above: Try to make your paper as user-friendly as you can by informing the reader early on what its purpose and scope are, and how you are going to set about accomplishing what you propose. The experience of the reader in expository prose (as opposed, say, to modern poetry) should be that of being guided gently and securely along an at least partially predictable path. The reader should not be jolted by sudden changes of point of view or by the introduction of new and unfamiliar concepts that are left unexplained. He or she should never have to stop and ask what you are trying to say, and thus be forced to infer from context what you must have meant, even if you didn't say it particularly well.

Once you have a draft, try reading it out loud. Flaubert did this with enormous success in Madame Bovary, and it may work for you, too. Awkward expressions, poor prose rhythm, gaps in logic or continuity, irritating repetitions of words or phrases often leap out in actual speech in a way that they do not when you are reading your own writing silently.