It’s a truism sometimes attributed to da Vinci, W.H. Auden, or Paul Valery, sometimes about painting, sometimes about poetry, and it goes like this: “Art is never finished, only abandoned.”

Anyone who has ever tried to write a play will understand. You will never feel completely satisfied. There will always be those lines, speeches, or turns of plot that you know you didn’t get quite right. So the question won’t be whether to abandon your play but when: When do you admit you’ve pushed the thing as far along the asymptotic curve towards its Platonic ideal as it’s going to get and vow to do better on the next one? While this question is always operant, it plagued me very specifically on and off for three years, and then another four, regarding Completeness, my play about university science nerds maybe falling in love.

Completeness was born when I was offered a commission by the Sloan Foundation, which supports plays that deal in some way with math and science, at a time when, as it happened, I had an idea for one. I’d become fascinated by a well-known problem in computer science (well-known to computer scientists, that is, and obscure to everyone else) called the “Traveling Salesman Problem,” which posits that even a powerful supercomputer running a sophisticated algorithm cannot tell you the best possible path for a traveling salesman among a given number of cities when the number of cities, and thus the number of possible paths, gets high enough. Rephrased as another truism, this one about love: Needing to know you’ve made the best possible choice leads only to paralysis.

And so I wrote a play about a computer-science grad student named Elliot working on this problem who meets a molecular biology grad student named Molly, whose work on synthetic wound healing metaphorically mirrors the difficulty of recovering from past heartbreak (which is, in my experience, another major obstacle to romantic bliss). Over the course of the play’s nine scenes they meet, shed their current partners, and get together (not entirely in that order), then run into problems in the form of new temptations and their own emotional baggage and ultimately falling apart…before maybe, just maybe, starting to reach out to each other again.

In April 2011 Completeness premiered at South Coast Repertory. An Off-Broadway run at Playwrights Horizons followed that fall, and the script for that production was published and sold—is still sold—in the lobby of the theatre. So the thing was done, right? Ordinarily, after hitting some, let alone all, of these benchmarks, I would let a play go, for a variety of reasons, some good (e.g., seeing your play performed for multiple audiences may truly have allowed you to refine it to the point where you are reasonably satisfied), some less good (once a play has been exposed in these ways, certain psychological barriers to working on it seem to get higher, i.e., if the play did well, People know this version and it’ll muddy the waters if I keep changing it, and if it didn’t, Isn’t it pathetic to keep grinding on the same doomed thing?), and some outside of your control (e.g., the passage of time may have turned you into a different person from the one who first wrote the play, so you can no longer access it authentically).

But even as the play’s Off-Broadway run wound down, my brain was still spinning over a couple of unanswered questions, and in this case they weren’t the usual lower-order ones about the wording of a joke or a few stray beats—the kinds of things a playwright is likely just to tweak or massage whenever they happen to be involved in some hypothetical future production. No: Of the aforementioned nine scenes, though I was basically happy with the first six and the last, I had the strong sense that in between, in the vicinity of scenes seven and eight, I’d gotten something flat wrong.

I could feel this wrongness emanating from the text whenever I read it, or thought about it, or sat in the same room with it. And this feeling was confirmed whenever I watched a performance and, during that very section, felt audience engagement—that palpable buoyancy or electricity or tension that, when things are working properly, throbs in the air between a play in progress and the collective group-mind watching it—ebbing away. When Elliot and Molly unpacked their emotional baggage for each other in scene seven, the whole thing seemed to meander and take too long. And when they were tempted by attractive new alternatives in scene eight, there didn’t seem to be sufficient dramaturgical real estate to properly establish the new characters.

Also, a meta-theatrical device that ended scene eight (an interruption to the play meant to underscore the weird tension between a play as both a scripted event meant to unfold in a particular way, i.e., like a computer algorithm, and a one-time live event with all the mess that that entails, i.e., a biological experiment) seemed to fall flat, but I couldn’t tell if this was because it was incorrectly constructed, because it was unnecessary, or because its impact was being harmed by the erosion of interest caused by errors I’d made in the scenes immediately preceding it.

And finally, while all this meandering and meta-theatricality made the play long enough to require an intermission, its status as essentially a romantic comedy seemed to call for a sleeker, intermissionless experience; as I remarked to Pam MacKinnon, who directed both the regional and the New York premiere, “If I could cut 10 minutes from this play I could cut 25.”

So when the Off-Broadway production closed, rather than letting the play go and moving on, I felt myself continuing to work on it. I say “felt myself,” because it seemed to be happening, if not against then at least without my conscious will. I’d be walking down the street, or exercising at the gym, or trying to work on a different, newer play, and my mind would return to the “problem scenes” in Completeness and turn them over, trying to “fix” them. It kept me up at night, figuratively speaking, and also sometimes literally. And because the best and really only way to confidently make changes to a play is to try them out for real in the laboratory, as it were, or full production, what followed was a years-long slow-motion odyssey of periodic chances to get back in the ring with the play and take another swing.

L.A. Theatre Works, which records audio versions of plays, did a recording in 2012, for which I tinkered mainly with scenes seven and eight. The following year Theater Wit, in Chicago, whose artistic director, Jeremy Wechsler, had worked as a computer scientist, staged a production for which I tinkered further. The play was published in a collection from Oberon Press in a version that differed from the version published by Playwrights Horizons and also from the L.A. Theatre Works recording, and was then licensed by Samuel French in a version that differed from all three. And still I had the feeling there was something I hadn’t cracked about the play, that I was simply rearranging deck chairs on, if not the Titanic then a boat that continued to list slightly to one side.

It was not lost on me, nor perhaps is it on you, as you read this, that there may be an eerie parallel between my ongoing preoccupation, nigh obsession with getting this play “right”—with “solving” it—and Elliot’s attempts inside the play to solve the unsolvable problem in his field. More than once I wondered whether this over-identification with my protagonist was all that was going on and if I, like him, was eventually going to have to conclude that certain things can never be completely solved, and that’s okay.

Then in November 2014 I happened to be in Los Angeles during a run of Completeness at VS. Theatre Company there, directed by Matt Pfeiffer. And suddenly, I had the lightning-bolt moment I’d been waiting for. I saw how and where to cut scene seven so that it would land properly and not meander. I saw the device that would allow me to more efficiently establish the new characters in scene eight in less time, and how this would allow me to evaluate whether the meta-theatrical breakdown that followed worked on its own terms or not. And I saw that all this would take enough time out of the play that I could finally cut the intermission too.

Why could I suddenly see this so clearly after straining so hard for clarity for so many years? At the time I thought it was simply that I had finally achieved sufficient distance from the play or had at last grown enough as an artist and/or person. And all of this was true, up to a point. But something much more specific was also going on, and it was this: Since the last time I had seen this play, I had finished a new one.

In June of 2014 I’d taken part in the first workshop of my play The Whistleblower at the Colorado New Play Festival in Steamboat Springs, marking the first time I’d been in a rehearsal room with a new non-musical in something like four years. The act of removing a new script from the private safety of my hard drive and placing it into the bodies of actors and in front of the eyes and ears of an audience had done something to the way I was now able to watch my older one. It was as if knowing for sure that Completeness would not be my last play freed me from needing it to be and say everything it could, and allowed me simply to help it be and say only what it wanted to—as if having to care for something new that was struggling toward a final form rendered the final form of Completeness completely obvious.

I told Matt what I was thinking, and that if he ever wanted to direct the play again—say at Theatre Exile, in Philadelphia, where he is associate artistic director—that my rewrite would be ready. Then I went back home to New York and, some three years after the play had closed there, did the rewrite in four hours. So the rewrite took three years and four hours, which was as clear an example as I’d ever experienced of the fact that sometimes the most important thing we can do for our art is something else. And I knew that it was right—or right enough—because from that day the splinter in my brain that been troubling my sleep all that time dissolved, never to return.

It would be another year before Matt and I managed to schedule a reading in Philly to confirm my suspicions that I was on the right track, then three more until, in October 2018, a production of the new draft went into rehearsal at Theatre Exile. Even then it took watching rehearsal room run-throughs, another bout of rewriting, and attending a preview and another performance near the end of the run to further evaluate and refine the new sections and at long last determine the ultimate fate of that meta-theatrical disruption (I turned it into an optional appendix).

Now, some eight and a half years after the world premiere of Completeness at South Coast Rep, Samuel French has (kindly) agreed to license the new version going forward and (very kindly) to republish the play. At last I feel ready to let Elliot and Molly go. Not really finished, maybe, but abandoned, they will have to travel on from here without my help.

Which is a good thing, because I premiered a play at Berkeley Rep in 2008 called Yellowjackets and, you know what, I never felt like I got that one quite right…

Itamar Moses wrote the book for the Tony-winning musical The Band’s Visit and is the author of the plays Bach at Leipzig, The Four of Us, and Back Back Back, among others.