A Source of Poetic Respiration
One billion poems. More than anyone could dream of writing in their lifetime…more, perhaps, than have ever been written since the dawn of writing itself. Ten times more, at least, than all of the books that have ever been published in the history of the world. And yet there I was, staring at an assignment asking me to create, at a minimum, 1,000,000,000 poems. How could this be possible? The answer is both simple and utterly complex: I would not write them myself… but no other poet would be involved.
Perhaps you are now thinking of the proverbial monkey who could produce endless streams of nonsensical lines before stumbling upon Shakespeare. In some ways, you are not far from the truth. However, putting questions of quality aside (as taste is subjective), let me clarify that I am speaking about “real” poems: ones that would be recognized as such, regardless of taste, by the average reader. You may also question the wisdom of completing such an exercise at all… wouldn’t the sheer number of poems render each individual one, however well-written, as just one grain of sand, invisible against the entirety of a beach? In response, consider the population of the world itself. Does the fact that you are one of nearly 8 billion people on this planet make your one life any less important, less valuable, or less unique?
I should explain that this is a typical assignment for a class in computational poetics: using computers to write poetry. The field has roots in the work of Surrealists and Dadaists of the early 20th century, but is on the cutting edge of creative writing, part of the emerging genre of digital literature. Some may consider “computational poetics” to be an oxymoron, arguing that no computer could ever write a true poem, one whose spark arises from, and illuminates, the heart of a human being. But such an argument overlooks the human being behind the computer, the one writing the code which makes such digital poetry possible. The computer is, in a sense, a semi-autonomous tool, a type of infinitely prolific pianola, revealing music that even the composer has not fully heard. We poets often speak of breathing life into our poems. I believe, through computational poetics, the computer can become an extension of ourselves — a mouth we may use to breathe life into our work.
And it is to breath that my mind went as I sat pondering this homework assignment…to the more than five million people whose lives have been claimed by the COVID-19 pandemic, their breath escaping into a hurricane that continues to sweep across the world, leaving physical, psychological, and economic destruction in its wake. It is not just their breath that has escaped, however. A 2020 article in The Atlantic describes one of the hidden, deeply personal losses of this pandemic: the absence of COVID-19 patients’ final words. Tethered to ventilators in semi-consciousness twilight, the pandemic’s victims have been slipping away in silence.
I have read a number of stories about the words of COVID-19 patients just before they are intubated. Some seem to accept their fate, others remain hopeful, perhaps even defiant in the face of death. It occurred to me that these truly are their last words, and many of these patients use these final conscious moments to send a message not only to their loved ones, but to the millions of others impacted by the pandemic. Before their physical death, they are entrusting a machine to keep them alive, to provide their breath. Could we allow a machine to speak for them as well? In those lost voices, I realized, lay one billion poems.
Japanese Zen has a long tradition of the jisei, or deathbed poem, typically a haiku or tanka that is said to be one's truest statement. Many of those dying from COVID-19 spoke about truth— too-late realizations that this virus is in fact real, and regrets of remaining unvaccinated. I felt that the jisei might be a way to capture their last words, honoring the desire to share their truth with others. Gathering a number of news articles where nurses or loved ones provided direct quotes from patients, I divided their words into phrases of seven, five, three, and two syllables, then created an automated poetry generator that “writes” haiku by combining these phrases in different permutations. With the words of just eleven patients, my code is capable of producing well over one billion potential poems.
After writing the code, I published my poetry generator, COVID-jisei, online. One must click a hypertext button to reveal each new, automatically generated haiku. Instead of simply labeling this button "Another poem" or "Another jisei", I decided to emphasize the very personal, human origin of these words by using the phrase, "Another life lost." Each time this button is pressed, the words on the screen disappear, instantly replaced by a new poem. The past haiku can never be recovered, just as the actual voices of the speakers, and the speakers themselves, have disappeared.
As would be expected, patients who are severely ill speak in short, monosyllabic statements, or instead write their final words, creating longer, multisyllabic texts. I found that some quotes worked best as shorter sections, even as one word, whereas others would lose or invert meaning if reduced... for example, "I thought this was a hoax" vs. "this was a hoax." Even the randomized juxtaposition of lines could alter their intent, an important lesson in how flexible and context-dependent language can be, a tangible example of how politicized, mis-interpreted, and often conflicting information has circulated throughout the pandemic.
I also found it interesting, although not surprising, that all references to gender are female ("mom", "mama", "daughter", "girl"). Fathers and sons do not make any appearance. This is true even though I have quoted an equal number of male and female patients. The poetry generator reflects our patriarchal society in general, where those dying are known to call out for their mothers (the typical caregivers) while female family members (wives and daughters) are still thought to be under the care of their male relatives. The generator clearly has its own poetic voice, influenced by our culture, but it did not write these poems. Nor did I, or the reader who clicks the button, or even the patients themselves — all of us, together, create each poem, just as the pandemic has revealed our global interdependence.
Computational poetics cannot, and will not, ever take the place of more traditional creative writing, but it also should not be considered a “lesser” or “artificial” form of writing. It is simply another option to convey meaning and emotion in our work, meaning which may be revealed by the very nature of how these poems are produced. In fact, computer-generated poetry might be exactly what we, at this time, truly need… much like the ventilator itself. Although this machine breathes for the patient, they are not dead, and not all of those on a ventilator will die. The ventilator is used to give the lungs an opportunity to heal, in the hopes that the patient will eventually be able to breathe on their own. The pandemic has revealed how critically ill our society really is, how, in many ways, we are already close to death. Perhaps, through machines, we may breathe life into our poetry in the hopes that we as a society may once again hear our own voice.
__________________
Jess Skyleson's poetry generator, COVID-jisei, can be found online at covid-jisei.glitch.me
Jess Skyleson is a former aerospace engineer who began writing poetry after being diagnosed with stage IV cancer at age 39. Their poems have been selected as finalists in the Tor House and Yemassee Poetry Prizes, and have been published by Oberon Poetry Magazine, Stillhouse Press, Nixes Mate Review, and Ponder Review, among others.