TITLE: OOPSLA Day 1: Writing Exercises at Extravagaria AUTHOR: Eugene Wallingford DATE: October 19, 2005 10:11 AM DESC: At what other conference would I be able to say that I gang-wrote traditional Japanese poetry with Guy Steele?? ----- BODY:

I am nobody:
A red sinking autumn sun
Took my name away.

-- Richard Wright

As I noted before, I nearly blew off Sunday, after a long and tiring two days before. As you might have gathered from that same entry, I am happy that I did not. The reasons should be obvious enough: cool ideas happen for me only when I am engaged with ideas, and the people and interactions at Extravagaria were a source of inspiration that has remained alive with me throughout the rest of the conference. In the afternoon of the workshop, we did two group exercises to explore issues in creativity -- one in the realm of writing poetry, and one in the realm of software design. Gang-Writing Haiku Haiku is a simple poetic form that most of us learn as schoolchildren. It is generally more involved than we learn in school, with specific expectations on the content of the poems, but at its simplest it is a form of three lines, consisting of 5, 7, and 5 syllables, respectively. If I understood correctly, a tonka is a poem constructed by following a haiku with a couplet in which line is 7 syllables. We can go a step further yet, by connecting a sequence of tonkas into a renga. John called the renga the "stretch limo" of haiku. Apparently, the Japanese have a traditional drinking game that requires each person to write a verse of a growing renga in turn, taking a drink with each verse. The poems may degenerate, but the evening is probably not a complete loss... Our first exercise after lunch was a variation of this drinking game, only without the drinking. We replaced the adult beverages with two features intended to encourage and test our creativity. First, we were given one minute or less to write each verse. Second, when we passed the growing poem on to the next writer, we folded it over so that the person could see only the verse we had just written. Rather than start with scratch, John seeded our efforts with haiku written by the accomplished American novelist Richard Wright. In the last eighteen months of his life, Wright became obsessed with haiku, writing dozens a day. Many of these works were published in a collection after his death. John gave each of person a haiku from this collection. One of them, my favorite, appears at the top of this essay. Then we were off, gang-writing poetry. My group consisted of Brian Foote, Joe Yoder, Danny Dig, Guy Steele, and me. Each of us started with a Wright haiku, wrote a couplet in response to it, folded Wright's stanza under, and passed the extended poem on to continue the cycle. After a few minutes, we had five renga. (And yet we were sober, though the quality of the poetry may not have reflected that. :-) The second step of the exercise was to select our favorite, the one we thought had the highest quality. My group opted for a two-pass process. Each of us cast a vote for our two favorites, and the group then deliberated over the the top two vote-getters. We then had the opportunity to revise our selection before sharing it with the larger group. (We didn't.) Then each of the groups read its best product to the whole group. My group selected the near-renga we called Syzygy Matters (link to follow) as our best. This was not one of my two favorites, but it was certainly in line with my choices. One poem I voted for received only my vote, but I won't concede that it wasn't one of our best. I call it Seasons Cease. Afterwards, we discussed the process and the role creativity played.

Most of us tried to build on the preceding stanza, rather than undo it.
This exercise resembles a common technique in improvisational theater. There, the group goes through rounds of one sentence per person, building on the preceding sentences. Sometimes, the participants cycle through these conjunctions in order: "Yes, and...", "No, and...", "Yes, but...", and "No, but...".
Time pressure matters.
Personally, I noticed that by moving so fast that I had no chance to clear my mind completely, a theme developed in my mind that carried over from renga to renga. So my stanzas were shaped both by the stanza I was handed and by the stanza I wrote in the previous round.
Guy was optimistic about the process but pessimistic about the products. The experience lowered his expectations for the prospects for groups writing software by global emergence from local rules.
We all had a reluctance to revise our selected poems. The group censored itself, perhaps out of fear of offending whoever had written the lines. (So much for Common Code Ownership.) Someone suggested that we might try some similar ideas for the revision process. Pass all the poems we generated to another group, which would choose the best of the litter. Then we pass the poem on to a third group, which is charged with revising the poem to make it better. This would eliminate the group censorship effect mentioned above, and it would also eliminate the possibility that our selection process was biased by personal triggers and fondness.
Someone joked that we should cut the first stanza, the one written by Wright!, because it didn't fit the style of the rest of the stanzas. Joke aside, this is often a good idea. Often, we need to let go of the triggers that initially caused us to write. That can be true in our code, as well. Sometimes a class that appears early in a program ultimately outlives its utility, its responsibilities distributed across other more vital objects. We shouldn't be afraid of cutting the class,but sometimes we hold an inordinate attachment to the idea of the class.
To some, this exercise felt more like a white-board design session than a coding exercise. We placed a high threshold on revisions, as we often do for group brainstorm designs.
Someone else compared this to design by committee, and to the strategy of separating the coding team from the QA team.

Later, we discussed how, in technical writing and other non-fiction, our goal is to make the words we use match the truth as much as possible, but sometimes an exaggeration can convey truth even better. Is such an exaggeration "more true" than the reality, by conveying better the feel of a situation than pure facts would have? Dick used the re-entry season from Apollo 13 as an example. (Aside: This led to a side discussion of how watching a movie without listening to its soundtrack is usually avery different experience. Indeed, most directors these days use the music as an essential story-telling device. What if life were like that? Dick offered that perhaps we are embarking on a new era in which the personal MP3 player does just that, adding a soundtrack to our lives for our own personal consumption.) A good story tells the truth better than the truth itself. This is true in mathematical proofs, where the proof tells a story quite different from the actual process by which a new finding is reached. It is true of papers on software system designs, of software patterns. this is yet another way in which software and computer science are like poetry and Mark Twain. A Team Experiment with Software Design The second exercise of the afternoon asked four "teams" -- three of size four, and the fourth being Guy Steele alone -- to design a program that could generate interesting Sudoku puzzles. Half way through our hour, two teams cross-pollinated in a Gabriel-driven episode of crossover. I don't have quite as much to save about this exercise. It was fun thinking about Sudoku, a puzzle I've started playing a bit in the last few weeks. It was fun watching working with Sudoku naifs wrap their games around the possibilities of the game. It was especially fun to watch a truly keen mind describe how he attacked and solved a tough problem. (I saved Guy's handwritten draft of his algorithm. I may try to implement it later. I feel like a rock star groupie...) The debrief of this exercise focused on whether this process felt creative in the sense that writing haiku did, or was it more like the algorithm design exercise one might solve on a grad school exam, taken from Knuth. Guy pointed out that these are not disjoint propositions. What feels creative is solving something we don't yet understand -- creativity lies in exploring what do not understand, yet. For example, writing a Sudoku solver would have involved little or no creativity for most of us, because it would be so similar to backtracking programs we have written before, say, to solve the 8-queens puzzle. In many ways, these exercises aren't representative of literary creativity, in several significant ways. Most writers work solo, rather than in groups. Creators may work under pressure, but not often in 1-minute compressions. But sprints of this sort can help jolt creativity, and they can expose us to models of work,models we can adopt and adapt. One thing seems certain: Change begets creativity. Robert Hass spoke of the constant and the variable, and how -- while both are essential to creativity -- it is change and difficulty that are usually the proximate causes of the creative act. That's why cross-pollination of teams (e.g., pair programmers) works, and why we should switch tools and environments every so often, to jog the mind to open itself to creating something new. -----