Hamlet performed by computers. Algorithms rewriting Shakespeare. One actor in the performance. A Piece of Work at Brooklyn Academy of Music is definitely Shakespeare, but it doesn’t look like your usual production of King Lear or Midsummer. An unusual project that will surely alienate ardent purists, it feeds the text of Hamlet into set algorithms and uses the live results as the script, performed by computer voices and two actors on alternating nights.
The concept was first discussed between director Annie Dorsen and Scott Shepherd when they were talking about work projects, and she mentioned, “I’m going to n-gram [sequencing a number of n-items from a given text or source] Hamlet.” She notes the method is used often in data compression, as well as parodies for the nonsensical nature of the results. But she was excited by the prospect of “new poetry out of raw material.” Dorsen also knew it was going to be the story of the Danish prince. She states we have a very unique relationship to Hamlet: “[it’s] a play that haunts us…even when it’s not in front of us, [it] lives inside your brain in a weird way… [it’s] encountering a ghost.”
The dramatic and coding dramaturgy for the production is thorough. The stage is constructed based on elements of The Globe in London: the cosmology of the heavens above, the trapdoor to the underworld, and of course, all the world’s a stage on the middle performing platform. An off-side curtain serves instead of the center rear door, originally designed for big entrances and dramatic scenes (court entrances, Polonius behind the arras, etc.). Dorsen also uses an empty stage with a big projection screen that shows text generation as it happens, not available to Shakespeare at the time.
The first act is five percent of the total play word count, performed by computer voices with some sound and light cues, using the text to control the go. The second isolates lines by key words, which is more syntactical than the entire pool. Shepherd or Joan MacIntosh perform the third act; he or she is planted in the audience, and the words are taken just from the soliloquies. This is more directed on parts of speech, finding and replacing parts to hear more of the rhythm than the actual word meanings. The fourth act only pulls from five scenes, which creates a great combination of characters and situations. The final act takes only letters from Hamlet’s Act V, Scene II, which reflects the isolation and chaos as the play unravels, a “structured breakdown of semantics, of the words themselves.” It’s an experiment in structure, following the breakdown of status quo and character and constructs in Hamlet, as well as an easily-toured production that can reach wide audiences.
Dorsen doesn’t have an engineering or computer science background and taught herself the necessary coding, but that’s not the point. It’s “not about the tech magic,” trying to impress the audience with her computing skill and how it looks. “A lot of the sophisticated digital art is a demonstration of art… [here, the] audience is let in to the process of it… the effect is complex, but the technology is simple.” By using what are considered outdated algorithms, she’s able to do a lot of coding (Shepherd also did some of the original structure), but it brings the piece away from “a high priesthood of sacred knowledge.”
Both director and actor have worked with Shakespeare previously, sometimes in a nontraditional sense. Dorsen directed Antony and Cleopatra in the Yale Drama graduate program. She focused on plays that sparked immediately for her but quickly realized that Shakespeare’s language and construction is practically impossible to fully grasp: “the play is bigger than you… [it] contains a universe.” Shepherd has performed with The Wooster Group, which piped Richard Burton’s 1964 performance into his ear during Hamlet and various movie clips during Troilus and Cressida. Although he says he found it maddening, it was certainly good practice for this performance.
“You try so hard to forget or pretend to forget what you are already going to say” with a set script, and Shepherd appreciates the freedom that having the words in his ear allows. “I have this very developed familiarity [with Hamlet], and this total inability to predict what’s going to come, and no time to process because you have to speak what’s in your ears. It’s like talking: you have the material in your brain, and you have the mechanism for speaking, but you don’t know what to say.” This makes his performance less intentional and more spontaneous, but also less interpretive: more like a channel or vessel than a performer.
Dorsen always knew that she wanted to work with actors in this situation. She professes that if she excluded the human element completely, it would be simpler—all structure and algorithm--but missing a key point of connection. It would be harder to ask, what does the human give to this performance, why is it important to have a human here? Shepherd was an easy choice, with his background and his early contributions. She met Joan at a party, and had an aha! moment where she knew she had found a missing piece. Shepherd and MacIntosh give very different performances according to the director, but she notes that each has value.
The computer is a different animal all together. Shepherd acknowledges that his mechanical co-star can be difficult: “The computer is a hard act to follow, [it] will speak it with equal conviction. You have to take the computer at its word… [but] as humans we are constantly in a state of judgment ... we constantly try to explain or lucidate what we just said. To take the computer generated text, which doesn’t make sense, to take it as it comes [is] trying to be as good as the computer.” Dorsen knows the computer has a very different definition of what is a mistake: “When it makes absolutely no sense… [it’s] repetitive, dangling punctuation, not human, linear… [it] makes computer sense.” This breakdown of language allows us to see how our words come together to form meaning, something often taken for granted in our mother tongue. Many of the tour performances to date were in translation, which brings another set of variables to the experience. Perhaps native speakers will be more caught up in the language particulars, but it gives a greater opportunity to step away from the familiar starting point.
Dorsen says her purpose isn’t to be provocative but to explore the relationship between computer and human. “There are people who take certain texts very personally, aspects of our culture very protectively.” These viewers will be most outraged from the scrambling of the text and theatre (mostly) without humans: “Theatre was a place for human communion, not for technology. Those are not oppositional forces; where you have technology you have not lost humanity… We are formed by [computers], as much as they are by us.”
The heart of Dorsen’s project is what the computer is telling us about ourselves, how the computer is a tool rather a replacement: “Making tools is human… to extend our reach, our physical or mental or spiritual [reach].” It’s an immediate reflection of our capacity as creators. It also allows us to see what is irreproducible in human cognition and interpretation, our value as emotive creatures capable of incredible learning and skill—in other words, “how we are funny animals that talk.” She asks, is there another kind of beauty that depends on the viewer rather than the perfection of execution? Can the computer allow us to be our best— go past survival and ourselves to think of the future and the metaphysical and beyond—by doing what it does best?" She gets excited, talking about collaborating with the algorithms, compromising to find the best solution, which might make sense with nonsense: “We discover together.” And in that sense, it is a completely traditional Shakespearean production.