Writing Together on Surprise

The following co-authored post refers to an ongoing writing project I am undertaking with Katina Rogers in which we mail a journal back and forth to each other over an extended period of time. In those pages, we engage in an extended dialogue about writing, reading, and life. Posts here are a space to share excerpts as well as to reflect on the process.

Katina:

July 28, 2026

Writing brings up a lot of feelings. As Brandon and I have jumped into the writing experiment we wrote about earlier this summer, one thing that has been incredibly clear is that there is a strong affective component to working in this way. As the journal goes back and forth, we feel anxiety, anticipation, connection. In my first entry, I wrote: “I wouldn’t say this is a comfortable project for me, but that’s part of what feels exciting. I wonder whether and how my ideas will take shape across these pages. I wonder how your thinking will affect mine, and vice versa. I wonder how our lives will intervene as we go, what that will look like and how we’ll make space in our writing for it.” Anxiety and uncertainty were top of mind as we started, as I wrote without quite knowing how Brandon would receive my words or how our collaboration would unfold.

Those dominant feelings shifted as soon as I opened the journal after it made its return journey, because now there was a truck sticker emblazoned on the front of the plain black notebook. I laughed out loud with surprise and delight, suspecting that Brandon’s kiddo had probably had something to do with it. Without using any language at all, the sticker signaled to me a kind of agreement not to take ourselves too seriously. It worked best because it was entirely unexpected. It surprised me.

Surprise is striking both of us as an incredibly important part of truly-human writing. Since AI generates a plausible text through prediction at scale, it is hard to imagine a scenario in which it creates a text that is truly surprising in a catch-your-breath kind of way.

I reflected on this in later entries:

Knowing we each have a dedicated reader heightens all these associations for me. The feeling of picking up a book and not yet knowing how the pages will transport you is lovely. Seeing the package in my apartment hallway, the anticipation and uncertainty is similar but sharper, because unlike the bookstore book we have created an accountability to each other. Seeing that package means I receive a gift, and also that I will be responsible for an action. Maybe that interdependence and shared responsibility is part of where the anxiety comes from. We don’t want to let each other down.

At the same time it creates a heightened possibility for delight. I opened the package late at night after a long day and broke into a smile when I saw the sticker. AI writing can’t do that. Anticipation and delight require something unknown and unpredictable, whereas an LLM does its thing by finding the flattest possible framing, the most benign and bland simulacrum of accuracy. But it’s the rough and unexpected edges that bring texture and personality — our messy but unique handwriting, the implied influence of others in the stickers, the bits of ourselves that a machine would certainly polish away in favor of a uniform surface.

Academic writing tends to train the impulse for surprise out of us, too. But if that’s one of the things that makes our writing most human, how can we relearn and reclaim it? So much of what Brandon and I are trying to do right now is an experiment in creating the conditions for something to happen—something that may surprise and delight us as writers. If we are lucky (and skillful), it may one day surprise and delight readers as well.

Brandon:

August 4, 2026

Perhaps the most striking element of this new project that Katina and I are working on is how surprising it has been! I think that’s been true about every step of the process, including the fact of it happening in the first place. The fact that Katina happened to message about a writing project she was just launching on. The fact that she was willing to collaborate. The fact that we were both willing to take this weird multimodal approach. All surprising, and that’s to say nothing of the fact that we’re relying so heavily on the United States Postal Service. Every single time that I drop this journal in the mail, I wonder if this will be the last time I see it. It’s a miracle every time this writing appears in my mailbox.

This sense of wonder is also reflected in how I experience the object itself. Katina talked about how joyful it was to find a sticker in the book. My own version of this small joy came in the form of the plant life she included: leaves and petals pressed between the pages. Katina even gave me a small sprout bookmark that I will treasure forever. I could never have expected these gifts, nor have I ever experienced anything like them in other collaborative work.

All of this has me thinking about what surprise feels like and means in different kinds of writing and reading contexts. I tend to write as a way to learn what I am thinking, as a way to process my thoughts on the page, so there’s a sense in which every piece of writing is a kind of surprise for me. Similarly, my background is in working on experimental prose and poetry of the last century, so I’m quite familiar with the surprise you can feel when you turn the page and find something new and shocking. Katina has me thinking about what it is that creates those moments of surprise, and what kinds of writing are devoid of them.

When messaging each other about these small moments of joy, one of us jokingly said, “AI could never.” I want to think about this further beyond the scope of this particular post. I think the joke reflects an assumption that a composite piece of writing sourced from GenAI is likely to try and reduce those surprises, to smooth them out in the hopes of generating something that can plausibly pass as a final product that was requested. But LLMs are built on large statistical models that carefully weigh the probabilities of words and phrases occurring in sequence; chance is fundamental to their work. The temperature of the model, its degree of randomness, is also a parameter you can use to adjust the level of “surprise” in the output. And that’s to say nothing about the creative prompting one might give to specifically ask for these unexpected left turns. I want to think more about this as we build this series into a larger project, not least because I feel as though I need to understand these aspects of GenAI better to approach them thoughtfully.

It’s interesting to me that surprise and the joy that comes with it are so key to what we feel as human-to-human interactions. I wonder how much of these joyful surprises we get from reading and writing come from the mind of the writer or the interpretation of the reader. Perhaps all reading and writing is a play between structure and surprise. How could you actually practice surprise and make it part of your craft, such that it’s a tool at your disposal and not a happy accident?