How I Write - Conversation Transcript


Eric Roberts

Eric Roberts refuses to have a computer at home, he has vowed, in an almost religious sense, never to watch television, mulls over writing ideas on long walks, and he even confesses to "Luddite urges." Certainly, Prof. Roberts dispelled any of the lingering stereotypes about computer scientists that I may have harbored. He loves and values books, his home is filled with them, and he does not think they will be supplanted by the computer any time soon. It comes as no surprise, then, to learn that Prof. Roberts cares deeply about writing, particularly as a means for non-specialists to enter into the way computer scientists think. "I actually think that that's in some sense the most important thing I do," he says. "I've told people that, I think, in my role as faculty, I'm a teacher, and that's of course central, but I have to think of myself as a writer because you have a multiplier effect on getting those ideas out that's far greater than it can be in the classroom."

He is deeply committed to the writing process, and he believes that textbooks have a crucial role to play in the learning process: "When I see people with all sorts of colored tab along the side of [my] book and enormous amounts of highlighting, I know that there's an advantage of having that physical manifestation of a book." In this conversation, he describes the process of writing textbooks: how he develops proposals, draws from classroom experiences, accumulates literary, artistic and other references so as to interest a broad audience, evaluates reviewers' insights, and learns from field tests. Being inclusive is important to him, so he has developed strategies for how to write without using gendered pronouns. He describes how critical responses are so crucial to the writing and revision process, so much so that he chuckles, "All red marks are in some sense a gift from God." He puts his mouth where his red pencil is: he always offers students the opportunity to revise their papers.

Transcript of How I Write Conversation with Eric Roberts

HO: Eric Roberts has agreed to be the first person from science and engineering we're going to talk with. He's written Programming Abstractions in C, The Art and Science of C - I tried to buy these books: I almost had a heart attack, 80 bucks, it's a terrific business, I hope you get a little piece of that - and Thinking Recursively, and of course countless articles. You're the first person from science and engineering - I've asked Professor Sapolosky from Biological Sciences, John Bravman has agreed to talk. The creative writers, by the way, are not interested in talking about this is all, this is like trade secrets in their ball game, but for the rest of us regular Joes this is workaday stuff. I'm going to ask Professor Roberts some questions, then we'll open it up to the audience to raise anything about writing that you want. I begin by asking you some questions, but if you have something on your mind, go right ahead.

ER: Well, I thought I would talk a little bit about the kinds of writing that I do because one of the things I know that's true is that the kind of writing, and the audience that you're writing for, make an enormous difference in how you go about doing that writing. I write many different kinds of things, I've written about 75 technical papers published, and usually what I do there is I write down all my ideas and then I figure out how to shrink it down to the particular page limit that that particular journal has. There are lots of exciting strategies that you can use to fit things exactly in the space available which is of course hardly the thing you would do in writing a textbook or an article that is seeking to persuade - I write lots of memoranda that are trying to articulate a position usually for senate or university committees in such a way that it's hard to answer the data - those types of articles are very different from technical papers and they're very different from books that I've written. Textbooks are very hard to write partly because of the scale - I can write a paper in a week's time - I've done the research; it's not so hard - there's nothing I can do to get a book out in a week or a month or usually a year. The kind of discipline to write something on the scale of a book forces you into different patterns than for anything smaller. That factor of scale is probably the biggest stumbling block for most people working on their dissertations. Everything up to that point is smaller and therefore can be done using a set of techniques that one learns and practices over time, and all of a sudden you reach this point where that's not possible anymore.

One of the things Hilton and I talked about earlier today - he was interested in some of the parallels between writing code and writing text, writing English, and one of the things I thought was that property of scale was exactly the same. Problems that you have writing a hundred thousand line program are not at all the problems you have in writing a thousand line program. The change in what you can accomplish in a short amount of time - and therefore the need to plan time effectively - changes as the scale changes. So one of the things I encourage people to do is to recognize that when you are writing for different audiences and different scales you may need to learn different strategies, ways of approaching that particular writing challenge.

HO: One thing I noticed when I looked through the textbooks, reading Thinking Recursively, is that you try to be very clear in breaking concepts down step by step and using analogies. How did you get to do that? That doesn't come easily to a lot of people.

ER: I think in large measure the structure of the textbooks reflects the structure and attitudes I've ended up taking to the class. I do not write textbooks for courses I have not previously taught. The genesis of all these textbooks was in some sense about what am I having trouble teaching, how would I prefer to get this material across, and so through the experience of having developed lecture materials and examples around given topics that form the genesis of the book but is by no means the same thing. I always have the optimistic idea that I've given all these lectures, the book will write itself, it doesn't. There are many, many things that happen when you try to piece things together, try to integrate different lectures into a cohesive whole. But the aspect of these books that I think you're looking at does come from the fact what I've tried in my teaching is develop a set of lectures and concepts and courses that are accessible to a wide audience and in making that possible - these two books - The Art and Science of C and Programming Abstractions that] are used in the CS 106 here which has historically been the largest course sequence at Stanford - over a thousand people took CS 106 in a given year - at the height, it's gone down a little bit since the collapse of the NASDAQ - and for many years it was the largest, and in being the largest it wasn't appealing primarily to computer scientists. That's not unique for Stanford, that's true for computer science courses in most institutions. But I was convinced that what I wanted to do was offer a course that was useful for anyone at Stanford, so that, no matter what your interests were, you would get something out of the course that would help you think the way that computer scientists think, and not just the general education course but a course that took a serious approach. And only 6 percent of the students who are in CS 106 go on to become computer science majors, a tiny fraction, most of the people are in other engineering disciplines or into the sciences, but a third or more of the students end up in humanities or social sciences or something entirely different. I don't think that's in any sense a loss, we need many more people doing those fields, and people should follow their passion. On the other hand I think it's good for those people to be exposed to that type of discipline of thought. In order to make that happen, though, you cannot frame the presentation in any way that would be off putting to those people, I have to draw them in, in some way. And so in writing these books, for example, I do try to break things down very clearly. I use more exposition than is typical in an introductory text. Some other faculty have criticized these books for being too chatty - more dense examples and less explanation. That's good for the people who respond perfectly well to those examples and don't need the explanation, but that's not everyone, the ones who can get it from the examples, the examples are highlighted - look at the things with the green background, and then you're done - but if you need the explanation I believe it should be there. The single most common comment I've gotten from people from other institutions about this book on why it's helped - I've probably gotten on the order of a hundred of such messages over the years - it was used in a 150 schools - is they appreciated the literary epigraphs at the beginning of the chapters, the fact that I would quote from such a wide variety of literary sources in ways that - yes, when you read the chapter this is actually appropriate to the material - although there is no sense in which presumably a 17th century author thought about computers - but by framing in a way that appeals to something that they can touch in their own interests - that provides a route into this material that otherwise might be scary.

I also taught - my first teaching job after getting my PhD was to teach at Wellesley College, which is a women's college outside of Boston, and in the 5 years I was there, one of the things that came natural to do was either to inverse the gender stereotypes - all my best students were women at Wellesely, ALL my students were women - and you can't use male-oriented examples, it doesn't work there, you can't use he pronouns and make it seem appropriate to the environment, and when I came to Stanford the courses that were introductory courses back in the eighties, which I completely redesigned used all those things. You were likely to have football metaphors in the class - we don't have those any more - those went away. There are no pronouns in this book - any of these books - any 3rd person pronouns whose gender can be identified - you can learn to write without them - and it's important to do so. And I have other little strategies, about ways I have learned to adjust the writing to make sure that you avoid putting anyone off, and those are things you acquire with practice, certainly the non-gendered language, the lack of examples that would be likely to alienate any particular audience is some thing I've worked very hard to achieve, and that's all part of parcel to the extent that I've tried to make these books successful.

HO: I know when I was reading Thinking Recursively right away in the first chapter you bring in the art of Mondrian and also quote My Buckets Got a Hole in It, so you're bringing a wide range, a field of references to explain things. Was this was a textbook used in a class too?

ER: This is an ancillary - it's used in addition to other books.

HO: For the textbooks, at least, you know you have a testing ground, you've developed a course. Now once you've developed a chapter and you want to get a sense of how explanatory it is, do you run it by students, do you run it by a friend who checks it.

I have a wife in the English department who checks it, which makes it quite wonderful. Both of these books were written entirely as draft textbooks that were used in the class before they were published, so they were field tested in their entirety here. Most publishers field test books externally as well, and both of these were indeed tested at other institutions. We get feedback before the books goes to press, so we have a sense of how that will play out not just in one institution. I just got done a 3 year project heading up the study panel that proposes curriculum recommendations in computer science, and one of the rules that we had for talking about cs courses and whether we would include experimental approaches in our report was that experimental approaches had to be used successfully in some institution other than the one that the creator lived in. Almost anyone can make an educational experiment a success if they bring their own energy to it - the question about writing a textbook is can you put that down on the page so that once it's divorced from that energy of the creator or even the institution in which the creator lives can it be successful there. For me it was important that these textbooks be field tested.

HO: There's a certain logic in explaining something, but there isn't necessarily a logic in writing something. In other words, do you start at the beginning and go to the end - obviously you have a conception. Where do you begin?

HO: The outline of every one of these books - what chapters are there - has changed radically from the time that I wrote the proposal to the publisher. That is the late binding decision of what the individual chapters are going to be. What I do is that I start these books around the examples that I want to use and you build out from there by adding exposition and then figuring out - you have a great example you really want to build toward in chapter 12 and somehow in chapters 5 and 7 there's got to be this particular topic so that example makes sense later on - and you also in any textbook want to make sure that there's a little bit of a spiral approach - you don't want to introduce all this boring dull stuff at the beginning and then use it at the end - you got to introduce some things and then build more on top of that and then go back and reinforce people's knowledge just as in a class there has to be some kind of repetition, some variegation of the modalities that you're using in order to make sure that people don't get bored. So I start with the examples, build around that, add continuity, rewrite, redesign, test things out, and in that respect it looks very much like programming as well.

HO: So it sounds like you write in concentric circles with examples at the core and then explanatory steps building to the examples that you want to use.

ER: You need this iterative quality. Usually the examples will stay but the structure will change considerably. Almost every case, just as in designing a program, you face trade offs, you can decide you can go this way and in that case it's going to have the following implications for the book, or I'm going to take a different strategy at the beginning and that will have other implications. You have to weigh those, balance those. The common ideas that you get when you look at initial reviews - particularly reviews of outlines - first of all, everybody who reviews them will write back the following paradigmatic form: "it's great, make it shorter, but add a, b or c" for completely disjoint values, those things that you are missing. The other common feeling that you get back - real pressure from potential adopters - is to put everything in chapter one. "That's so important it's got to be in chapter one," but so does everything else, and some things depend on other things, and if you have a prerequisite structure that forces - they wanted that chapter - none of that works. You have to find some way to break the inherent cyclicity of the concepts that are in one of these books and finding a way in a textbook to linearize them. Maybe in some future world where those books are electronic and hyperlinked in some ways - though I'm a little skeptical of that - you may be able to avoid having a linear approach to the topic. But strategically you need to be able to present it in one lecture after another - there's got to be a strategy for getting these ideas across in a way that doesn't force everyone to know everything at the beginning. Therefore you are going to have to break certain dependencies and then develop a linear flow based on that.

HO: You're working on a large book or even an article - where do you do this? Are you doing this in your office, in your own office at home? I assume you write on a computer.

ER: I do write on a computer and I invariably do it in my office although as often as not it's not this office. I find it very difficult to do the kind of writing you need for a book without going away. There are lots of things you can - I got 5 minutes before the next appointment I can get an email message written - it's useless to say I got 5 minutes what can I do on the book - nothing - you need 5 days, then I can get my head into it, I can start making real progress. The interrupt level processing, which is so much a part of my life as a faculty member, particularly as dean, doesn't lend itself to the long project, so I'm doing this in an office, but the ones where most work has been done at the offices was at Swarthmore, the office at Harvard when I was on leave from Wellesley and I wrote Thinking Recursively, the office at Reed where I finished up Programming Abstractions, and when I'm at Oxford on sabbatical in the spring will be an office wherever I'll happen to be for the current book. I do not have a computer at home - I believe in that - so if I need to work on something I will come in - home is for reading and for spending time with Laura, for all the other things that home is for and I think a computer invades on that, particularly for those of us who use it all the time during the day.

HO: It's interesting to hear that, the idea of taking a break, of having another space. Have you been working on another project?

ER: I've been working on - I don't know if it's my magnum opus, but this is the biggest project I've tackled - it grew out of a longstanding idea. If you go into the bookstore and you want to find out what's interesting about biology, you don't know anything about biology, but you want to find a book that explains why modern biology is interesting, there are lots of those books - people like the late Stephen Jay Gould have written a great deal of wonderfully popular, accessible material on very interesting subjects, the same thing on astronomy. You can do the same kind of thing for most modern sciences, you cannot do that in computer science. You go to a bookstore and every single title is some variant of "X for Dummies," of which there are over 600 different X's - this from the company that brought you Cliff notes - that's what's on the shelf or "Learn JAVA in 15 seconds," or whatever, and there's this notion that it's all tied to learning some skill and learning it in the shortcut way, how little effort can you put in to learn this skill without expending any mental energy that will somehow increase your earning potential by leaps and bounds. Now, I'm not interested in that, and what I wanted to do was to develop a book that people could go and see "That's what's interesting about computer science" for, as I describe it, the Scientific American readership, people who are not afraid of science, people who are able to think about issues at the level of a reasonably well educated college student or good high school student, I don't think that would be ruled out, it's not a book for young kids who are science phobic people, that's not the idea, it's a book for SMART people who are interested in finding out something about computers. And I've done this idea as a sophomore college course - John Hennessy taught it with me and has now taken over the course - called "The Intellectual Excitement of Computer Science" - but the book that I'm writing is precisely for that. I want to give an introduction to the history of computing and all the great ideas that have come out in it in a way that allows people to play with those ideas, so it's associated with a lot of online technologies so that you can use the web and can go and play with problem solving and go and use the machines that Babbidge designed in the 1820s and are actually simulated and work and all those sorts of things, but at the end you will have a very good overview of why computing is intellectually interesting, not why it's profitable, and that book it's been three years in the making and I only have seven chapters.

HO: Out of how many?

ER: I think it's about 40, but they're bite sized. The idea is to have one chapter per day in a semester course, and that's the project for winter, spring and summer while I'm away.

HO: Now, when you are going to be working in this office, what kind of environment do you like to have? It sounds like you like to be away, that's one thing, in a new environment, away from distractions here. Do you have a dark room, a light room, do you set up? Does everything have to be neat?

ER: I have never been accused of having a neat office. There's that notion that a cluttered desk is a sign of a disorganized mind, what's an empty desk? The point about being away is not about being in a new environment - in fact, that's probably not helpful - what is important is the ability to work long and interrupted periods, not necessarily in a day but over the course of a week or a month, to be able to put fulltime attention, i.e., 10, 12, 15 hour days, in to that process. Now, I won't be writing the book the whole time, I can't do that, but I'll be thinking or I'll be building tools or thinking about the examples or drawing pictures or something that relates to the book, and where I have a single-minded approach to the vision, so that I can get all of it in my head and then run with that, because if I'm distracted, if there are too many distractions, then that cohesive whole, that "eyes on the prize" sense that I'm moving towards that goal will not be sustainable for me, so I find it very difficult to work piecemeal on any large project. That's why I found it useful to have sabbatical time to go and do that kind of work. I don't - just to hark back to that person who listens to light rock - I don't listen to music, I can't do it while I'm writing, even music without words interrupts the flow of my words for some reason, so the office is quiet, the door is closed.

HO: And when you're thinking up ideas away from the writing process are there particular techniques or methods or ways that ideas are generated.

ER: I have some ideas in the shower, but my principle source for those ideas is walking. I haven't bought a parking sticker in years - it takes about a half hour from my house to the campus, and on those walks I have some of my best ideas. If those aren't sufficient I also like taking long walks. I remember not so long ago I walked from Canterbury to Dover and over the course of that 20 mile jaunt I figured out lots of things. You're also experiencing the countryside. Somehow there's enough time to do that and still think about those issues.

HO: And you take a little notepad?


HO: And you remember?

ER: And I forget as many things as I remember. But somehow there's still the sense that in that process of refining what it is that I want to say that it comes back. I think most people have had this experience of when your machine crashes and you lost two hours of writing - it doesn't take two hours to regenerate it - it usually takes about 15 minutes to regenerate it, whether it's program or text, because the real work was in figuring out what it was you were going to stay and in fact that's stored up here somewhere, so I think the same thing is true, that just the working it out, even if I forget details has reduced the amount of time that it's going to take me by a lot.

HO: OK, what I'll do is open up for everyone here to ask questions . .

A: While you are writing your next book are you looking into in any ways make it in a form, as you said, with using other technologies that are now available?

ER: What I've done for my new book is make it so that book really does use the power that the web provides, in that every chapter is associated with not only an "applic" or more than one "applic" which is simulating the structure which I am describing in the chapter so that people can play with it and those are all linked through a web site at Addison-Wellesley or will be - we've got some prototypes up - but they also have lecture notes, power point. This is something that adopters want, but it's another way of approaching that same material. All that material will be publicly available and the idea is that material will encourage people, we hope, to get a textbook that describes it. I still believe that there's a wonderful role for texts. When I see people with all sorts of colored tabs along the side of this book and enormous amounts of highlighting, I know that there's an advantage of having that physical manifestation of a book. And there will be many things in that book, particularly narrative flow, that sense that I want to make computer science a part of an intellectual tradition that has connections to all sorts of other domains - I've taught lots of courses that make those crossovers. It's important to recognize that computer science owes a great debt to other fields of knowledge and vice versa, and that computer science changed the way other sciences and other disciplines proceed, and it has in turn been changed by them. And I think that those are much easier to develop in the context of a book, although there may be some ways of expressing them on the web as well. Somehow I think it may just be my Luddite urges but I'm still very much committed to books and print. I read two or three a week.

A:Could you briefly describe the process of getting a textbook approved?

ER: I can certainly talk about it. I've done several and I've advised a number of people in the computer science department about the process. What most textbook publishers - which is a different part of the company in most publishing houses from 'trade,' which is what you sell at a bookstore. The textbook division is looking for the marketability of some textbook to a suitable audience which they hope is large, so what they are going to ask you for are your sample chapters, but more importantly in their minds, a survey of the market, what audience are you addressing, what other textbooks cover that market, and how your textbook differs from what is already available, what is the value added that your proposing. Now, it may not be much in the sense that it may be that one publisher has a particular title on their list and you have found a publisher that wants a competing title, and it may not be doing anything differently but probably in order to get it accepted you'll have to come up with some rationale for why it would displace the market leader in that particular sector. What I've found - and certainly it's true in most scientific disciplines - is that you can take that proposal and prospectus to multiple publishers. For The Art and Science of C I sent five copies of the prospectus out and got six offers back because one of the agents shunted it off on the side to yet another publisher, and then entered in a process of negotiation with the different editors to see who would provide the best support for the development of the book. It is certainly true that the time you will get the most attention from publishers is during that time when they're trying to sign you and after that time they tend to disappear a little bit, but they do provide a great deal of value. For one thing they provide all of that field testing structure, they can assemble focus groups, they can assemble reviewers, they can find other institutions to test out work, and primarily a textbook division of a major publishing house will handle marketing and distribution which I have absolutely no interest in, so I want someone to take that on. I'm been very happy with Addison Wellesley, I think there are many very good publishers out there. I don't know if that answers your question.

A: Pretty much. So these reviewers, what do they review for? Subject matter, writing style?

ER: What they do is, publishers have a stable of reviewers that they use for particular areas of work. The larger publishers will have of course more focused reviewers who can provide better feedback. I do a lot of reviewing of textbooks and typically you get honoraria of 250 to 500 dollars - that range - for preparing a review. So you want to do it with some seriousness. I've very much appreciated, I would say, the responses from half of my reviewers, the other half I just ignore because I don't think they've done a good job, same thing on conference papers: What is this? You get enough good feedback that it makes the process worthwhile, and - one thing that holds the same about all writing, people have to get over the fear of being reviewed or the ego sense that a criticism or a suggestion about a particular piece of prose is somehow a negative reflection on you as author. I remember learning this very - I didn't get very good feedback on my writing as a college student. I would get lots of high grades and not very many comments and I was happy with that, that sounded good to me. At my first research lab job I had as a graduate student over the summer I was asked to write a few papers for Bolt, Bradick and Newman, which was the company that built and managed the arpanet, the predecessor of the internet, and I remember circulating one proposal around and just having it come back from four different people on the same corridor all covered with red marks and I was just scandalized and hurt. But in reading through those comments I thought they're right or this one is completely off the wall but somebody else said something that was useful in that same section. Now I feel if I don't get a lot of red marks on what I write, primarily from Laura my editor, I won't be able to produce the finely honed prose that I think is so important in writing. I care about how these words come out.

Another thing I used to do, I used to do this at breakfast, when the Good Earth restaurant was down there, I would sit at the counter and I would take my draft chapters and I would read them out loud, not so that anyone could hear, but to hear how they sounded. This is what a poet needs to do. I believe that you need to do that really for textbooks. You will find all the errors but you will also find the ungainliness. I have made lots of red pencil marks on my own prose but you have to understand that all red pencil marks are in some sense a gift from God - Boy, they're going to make this that much better - and it's hard to get over that sense of ego investment. You need to do it.

A: So it seems like textbooks are unique in that the purpose is to light the light bulbs in one mind so when you are trying to have them reviewed the reviewers may not be able to look at a revision and say whether or not it would light that light bulb too. I was wondering how you went about having reviewers look at different drafts.

ER: I don't think I've ever had the same reviewer for instance look at multiple drafts. Editors look at multiple drafts, that's more for the writing style and clarity. You'll be amazed how much someone can assess about the clarity of your presentation when they don't know the technical details, in fact it's often better. I want to be able to write for something that anyone can recognize it's coherent whether or not they can understand the whole flow. So that, in that editing process I have many iterations, and, of course, many iterations with myself in that editing process. Isaac Asimov is said to have been able to write complete drafts start to finish and never edit anything - amazing. Not in my experience and I will never be, and I won't get as many books done as a result. On the other hand I am convinced that that process of continual revision makes things better. The reviewers come on at the beginning when you're looking at the acquisition and when you're getting large-scale feedback on the effectiveness of a particular chapter or section of a book.

A: What are the most common errors you see in student papers?

ER: Spelling is terrible (laugh), no. I've read and reviewed a great number of papers because I've taught a lot of courses in which papers, not programs, are the principle deliverables. I teach a course in Computers, Ethics, and Social Responsibility most years where people write papers about the societal impact of technology and policy decisions that follow technology. I've always taken a section of that course and read the papers, just as the TA's do, for my section. Argumentation is not as strong as I'd like. I feel that most of the papers are written at the last minute and don't in fact get the benefit of multiple rewrites or even single reads. At the same time I actually think that Stanford students write pretty well - they write better than Wellesely students do. Laura and I disagree as to whether they write better than Swarthmore students did. And all of those, of course, being elite schools, are better than you would expect in the run of the mill writers. On the other hand, I think that people could become better writers by putting more time to the project.

In my current introductory seminar, for example, which is a prototype for an Introduction to Humanities course on the subject of technological visions of utopia, when people submit papers I write long pages of commentary, not on the sheet, I just take my computer and write my assessment of that. I actually have a whole bunch of canned paragraphs for when someone has failed to do a particular type of thing, so that I can assemble and mix and match those things which simplifies my job a bit because there are repeated problems and grade that paper and give that back and say, OK, if you want to rewrite it, go ahead. And every course that I do with writing involved I give students the opportunity to rewrite and that's essential to get good writing.

HO: Do they take the opportunity?

ER: Yes - of course, the people who got A's on their first submission, even if I have comments to make, didn't. People who did not got A's - I think I gave 3 A's out of 16 students - of the rest 9 rewrote, so it was good. There's altogether too often, though, a temptation for those students to rewrite merely - I said that this word wasn't good, so they changed the word, but the rest of the comments were the flow of this paragraph is a little bit weak, those are harder, I haven't given them quite as clear a roadmap on how to fix that piece. Some people take it very seriously. I think that's the only way to get people to do better writing.

HO: I've suggested to students to take a look at that last paragraph, it's pretty good, start there. That's a rough pill to swallow. How has reading affected you as a writer, particularly non-scientific or even scientific reading?

ER: I do lots of reading. I haven't watched any television at all in many, many years - this is like walking to work, it's a religious sort of thing for me. In the time that I'm not watching television, one of the things that I like to do is read, and I read broadly, and partly, having been married to Laura whose interests are in English and writing and literature, I've always had some interest in that area, but of course sharing a life with someone broadens your interests. And so I'm interested in her poetry and through that the sources of her poetry and I read a lot of poetry. I've read all of Shakespeare and I'm hoping to complete my cycle of watching all of Shakespeare when we're in England. All of those things provide you with resources to use as metaphors, as references to look back, it of course helps you in reading because there's so many references in literature to those common sources of Shakespeare and others, and knowing those references makes it easier to get something out of the reading. But one of the things I do while I'm reading is if I run across a quotation, a little section, I think, ah, a fine epigraph some day, I type it into the computer or find it, these days, on line in the Guttenberg project's archive of electronic sources for things that are in public domain, and just take that snippet, organize it, put some key words on it and I have a huge collection of potential quotes. So when I'm looking for epigraphs, it's not a question of going to Bartlett's as much as it's a question of - ah, I remember, I thought about this once upon a time, and there was this perfect quote, and I still have it, and it's indexed well enough for me to get it.

A: What are your other sources for finding quotes - you've mentioned your wife.

ER: At our house there are bookshelves on every wall and they go all the way up to the ceiling, there's literature and poetry, and then there's the whole raft of political theory. There are almost no technological or scientific books at home, they're all at my offices. I don't know how many shelf feet there are of political theory but it's probably - hundreds -- but I read those. I read those and they provide a lot of insight. The question of how they come to me - I get recommendations from lots of people. One of the things I've done in this introductory seminar on utopias is I asked people what have they read that had given them a sense of how the world could be better, and I read every one of those things so that I could talk to them about their interests. For instance, going around that first day, four or five students mentioned a short story which turns out to be a story that's read by lots of teenagers in the last few years, "The Giver," which I had never heard of. Frankly, it came out long after I was a teenager or after I was forty (laugh) so this was not something that occurred to me, so it was a very quick read and I picked that up and, yes, I see how this could influence people's sense of questions about utopia. So I get from my students, from my friends, from my communities - we get together with people to read children's stories, to read poetry. We're part of several literary circles that make that much easier. Browsing in Kepler's, that's a good thing. Bookstores are a good thing, and we have to make sure that bookstores survive, so it's good to buy things there and not entirely on line. Powell's!

A: For your textbooks, do you use a word processor or do you use . . .

ER: I have written and formatted for camera ready copy in Microsoft word - it seems crazy because it doesn't have good page formatting capabilities and some sense you'd like to use a layout manager like Pagemaker but I've found that Word - at least the good versions of Word 5.1 - the newer ones have taken away most of the features that I need, but I found that I can develop a very good style with just Microsoft Word.

A: When you say it takes a week to write a conference paper is that a week when you're doing other things or . . .

ER: It's a week when I'm doing other things, and it only makes sense if the work has been done, that's the time to write it up. Let it sit for a day or two, then come back and look at it - but as I said at the beginning - writing what you need, discovering that it's twice as long as it's allowed to be, and then finding a way to say only the essential things.

HO: And you don't feel bad about writing more and then cutting.

ER: No, I think it's much more efficient.

HO: Some people, it's their precious words, and they feel very attached to them.

ER: Get over it! It's that same attitude which is going to make you hate red marks.

HO: One last question before you go, and I think I know the answer to this question: Do you enjoy writing?

ER: I do enjoy writing. I actually think that that's in some sense the most important thing I do. I've told people that I think in my role as faculty, I'm a teacher, and that's of course central, but I have to think of myself as a writer because you have a multiplier effect on getting those ideas out that's far greater than it can be in the classroom. As a New Year's resolution many times - and I think I need to do this again - I said I need to give fewer talks and write more. I like giving talks and one of the things I have find interesting, there are interesting crossover effects about the way I give talks and the way I learned to write. For example, I always talk without notes and over time learned how to get over that terrible problem that comes when you're talking without notes, so what happens when you can't figure out what to say. The mechanism that works in that case is to learn to have multiple strategies for finishing each sentence. So you've gotten yourself over here but you know you want to be over there, there has better be some stratagem for taking a sentence that started there and moving it over to the place that you'd like to have it go. That turns out to be extremely valuable in writing because it allows you to make the text evolve in different ways, and the more options you yourself have for putting something down on the page means that you can avoid some of the pitfalls, it makes it much easier to take away the gendered language. None of these books, for example, ever use the word 'data' in a way in which the editor can tell what number I thought that had because computer scientists have insisted for years that it's singular and editors believe that it's plural, and I believe that it's both. I believe that data should be considered plural if you mean statistics and singular if you mean information, and since I cannot convince any editor to edit it that away I just don't write it. Just put in an auxiliary verb - might, could - it's insensitive to the number or add a noun or you do something - I don't think about it anymore, I just write.