Introductions, Papers, Abstracts and Beyond

Ph.D. Professional Seminar. 25 March 1997.

Nick Trefethen & Joe Halpern

This session was actually two shorter talks about writing. Notes for both Nick Trefethen's discussion and Joe Halpern's discussion are available.

Trefethen prefaced his remarks by pointing out that he was going to talk about his personal experience as a writer and professor, rather than attempt to lecture on general principles. He presented several points accompanied by humorous "grooks" by Piet Hein to illustrate those points. (To read the "grook" associated with certain remarks, click on the word GROOK whenever you see it.) GROOK

  • "This talk is about me, not about general principles."
    People will operate and do their work in very different ways. This talk will explain the way writing has fit into his life and some of his thoughts and observations about the process. GROOK

  • "My writings are what I care most about in my life."
    It could be suggested that life in academia is an attempt to escape mortality by creating something interesting, perhaps timeless. What we "leave behind" may be only our ideas, and, expressed well, these ideas can last for a long time. To some, writing is more important than, for example, teaching. Although, Trefethen acknowledges it would certainly not be a good thing if everybody felt this way about their writings. GROOK

  • "I cannot distinguish writing and research."
    The process of writing is different for everyone. Trefethen's preferred method is to work a bit, write a bit, work a bit, write some more, etc. Thus, the writing goes through many iterations. His feeling is that no one can truly think clearly except by writing, however he admits that this is obviously not true. For him, though, it certainly holds. He also observes that this tendency in himself has become more extreme with the years. GROOK

  • "I polish every paper endlessly."
    For Trefethen, each paper goes through multiple iterations, he estimates approximately 20, each of which consists of reading every word in the document. He made an analogy to fixed-point iteration, that is: Call the paper X, then Edit(X) should eventually be X itself again (Edit(X) = X). He wryly cautioned that within this framework, being a student of his can be deadly. GROOK

  • "I am obsessive about TeX and PostScript."
    While discussing typesetting, he mentioned his fastidious concern over fractions of a pt. in TeX, be it left, right, up or down, and also several pet peeves such as underlining spaces in an HTML hyperlink, using the correct umlauts and the distinction between en-dashes and em-dashes in TeX. The ultimate goal of this sort of attention to detail is to make the typesetting so perfect that no imperfections therein can distract the reader from the ideas being presented. GROOK

  • "I can't rest till it's utterly simple and utterly clear."
    Trefethen is more interested in explaining and expositing than creating, per se. He posits that each theorem has a Platonic proof ("God's proof, God's explanation") at least on a psychological level if not philosophically. In his mind, the wrong proof is nearly as bad as a wrong proof. However, he observes that this can be a risky course to take, especially in academia, where simplifying is somehow disrespected, and conversely making easy things hard is lauded. See "Dancing With Professors: The trouble with Academic Prose" by Patricia Nelson Limerick in the NYT Book Review. GROOK

  • "My papers are all equal, and all different."
    He feels about his papers as parents arguably should about their children. Virtually everything, especially in recent years is a new contribution. He avoids rehashings of already published material. He makes no distinction between proceedings and journals. He also mentioned that he finds the idea of "ranking" one's work (refereed journals as distinguished from conferences as distinguished from workshops, etc.) on a c.v. to be particularly repugnant. GROOK

  • "I only work on problems I consider important."
    He suggested that there is no point in working on something that doesn't have impact, whether it be long-term or short-term. However, this takes confidence and experience, and he pointed out that early on on his career he was more narrowly technically focused. In short, being bold is fun, GROOK

  • "Ignoring others' work is a crime."
    Trefethen argues that papers with a sufficient number of references, and without good references are "intellectually reprehensible." A paper must be authoritative as well as creative. He drew a comparison between academic fraud which is rare but receives lots of press and "cheating" (working in an intellectual vacuum) which is ubiquitous, but gets very little attention. This can cause catastrophic results at worse, and result in an extremely loss of efficiency at best, and yet it continues to happen. He suggests that the reason this hasn't stopped is that readers tolerate it. GROOK

  • "Not everybody is like me."
    Trefethen acknowledges that he is fairly pathological in his obsession about writing, and that not everyone will fit this pattern. There are other ways to function. His focus on the vagaries of the English would be very unusual coming from a non-native speaker. However, as English is becoming the de facto international standard for scientific discourse, his concerns about the quality of writing become more (arguably less, depending on one's perspective) urgent. GROOK

    Joe Halpern discussed his "biased" view of writing a paper. He first made the observation that the most important thing when writing a paper of any sort is the story. The story is not the main theorem of the paper or the system that's been built. Either of these will support the story line. The story is what will convince readers that they want to read your paper.

    There are several ways of figuring out what exactly the story is. If you write the abstract first (rather than finishing the paper and then coming up with an abstract) then you'll be forced to figure this out. Other methods include giving short conference talks (20 minutes, 10 minutes) to people who can offer constructive criticism. If this isn't possible, then attempting to explain your results to friends, partners, even parents may help to clarify things. Sometimes explaining things to people not well versed in the area will offer insights at a structural rather than technical level.

    Structuring the paper well, making it accessible and inviting and making sure that the story is clearly told will make it more likely that people will read your paper. A usual outline will include an introduction, definitions and background, technical results and a discussion/conclusion.

    The introduction is the most important part of the paper. This is usually where people will decide whether or not to keep reading. Here, you should present the big picture, suggest ways to formalize the intuition behind the big picture, mention your results which back up this intuition and also briefly discuss how your approach compares to other approaches. The reader should be able to read your introduction and have a good, clear idea of what the paper is about. Be careful, however, to make sure that the rest of the paper lives up to the introduction. There is a fine line between honesty and not downplaying your results. Halpern suggests that a good introduction is similar to a good advertising campaign. Make people interested in your product (the rest of the paper) but don't build up expectations that cannot be met.

    In a typical paper, the second section will give the necessary definitions and background. The question here becomes: How much should you assume that the reader knows? This will of course depend on the intended audience. However, even if the intended audience is expected to be fluent in the area, Halpern suggests it is always better to err on the side of including more rather than less. This makes life easier for the reader and will avoid sentences such as: Here we discuss X as defined in [1] and implement using the algorithm presented in [2] with the adaptations proposed in [3].

    The next few sections will include the technical results. Again, be sure that the results presented here fulfill the promises of the introduction. Also, keep in mind the story you are trying to tell and make sure the results are consistent with that story. Often, the story may suggest new results to prove or experiments to run. Scholarliness is key. It is important to include a detailed comparison to or discussion of related work. Also be careful to separate things into coherent, meaningful sections.

    The final section should wrap things up for the reader. Summarize the story again, suggest possible future research and applications, speculate on the impact and remind the reader of the big picture. Mention once again how the various pieces relate to the big picture and tie together the loose ends.

    The slides for Halpern's talk are available here in PostScript. Other references on writing and research can found on the CS706 homepage.

    Back to Ph.D. Professional Seminar Homepage.
    The Case for Obscurity
    On Thoughts and Words I.

    If no thought
    your mind does visit,
    make your speech
    not too explicit.
    The Egocentrics

    People are self-centered
    to a nauseous degree.
    They will keep on about themselves
    while I'm explaining me.
    To Sum Up
    A grook on finding a form
    commensurate with one's subject

    It may be observed
    in a general way
    that life would
    better distinctly,
    if more of the people
    with nothing to say
    were able to say it
    succinctly.
    A Tip
    to members of the literary profession

    Those
    who can write
    have a
    lot to
    learn from those
    bright
    enough
    not to.
    Twin Mystery

    To many people artists seem
    undisciplined and lawless.
    Such laziness, with such great gifts,
    seems little short of crime.
    One mystery is how they make
    the things they make so flawless;
    another, what they're doing with
    their energy and time.
    Ars Brevis

    There is
    one art,
    no more,
    no less:
    to do
    all things
    with art-
    lessness.
    Wide Road

    To make a name for learning
    when other roads are barred,
    take something very easy
    and make it very hard.
    Grook on Long-winded Authors

    Long-winded writers I abhor,
    and glib, prolific chatters;
    give me the ones who tear and gnaw
    their hair and pens to tatters;
    who find their writing such a chore
    they only write what matters.
    The Case for Obscurity
    On Thoughts and Words I.

    If no thought
    your mind does visit,
    make your speech
    not too explicit.
    That's Why

    Why do bad writers
    win the fight?
    Why do good writers
    die in need?
    Because the writers
    who can't write
    are read by readers
    who can't read.
    To the Reader
    A compliment
    with which to end a book

    The reader my expectant mind project
    was, gentle reader, not a bit like you.
    So if this book was less than you expected,
    believe me, I've been disillusioned too.
    Notes by millett@cs.cornell.edu.