A key aspect of the Gnoetry software is the ability of a human operator to intervene in the language generation cycle, helping to “guide” the artistic process and to produce a result that is a true collaboration of equals.

We can of course enjoy these poems without knowing the details of the collaboration. But of course part of the fascination with these poems is about who wrote them, specifically that the machine was a co-author.

With generative systems that collaborate with humans, in order to “believe in” the machine as a co-author, is it enough for the naive reader/viewer to see only the results of the creative process, or does the reader/viewer instead need to be the one in the creative process, the co-author herself?

Or at a minimum, to witness the creation, to see who did what.

Or, along with reading the poems, play with the system to get a feel for its authorship potential, which you can then project into the reading of the poems, to give you the understanding of the nature of the collaboration.

This is a little bit like a teacher grading the homework of an elementary school student who got their parent’s help…

While he does afford Gnoetry agency in his first paragraph—to go so far as to ask what or how much the machine wrote is to evince some faith in the machine as equal collaborator—Andrew’s misgiving is of the sort one might feel in the face of a story co-written by Famous Author X and Unknown Author Y: Is X just trying to give Y some credibility? Did Y actually do anything? Is X or Y a fraud? Are they just sleeping together?

In fact, the question about the machine as co-author is just bad faith and could be leveled at any author at any time. Why assume that any author wrote anything she claims to have written, given the nature of influence and the writing process and the fact that any time an author writes it is, in some respects, a collaborative process?

That said—perhaps a little glibly—Gnoetry wrote everything in The Dublin of Dr. Moreau; Elshtain edited the writing: he changed punctuation, fixed some verb tenses, transformed no more than six words per sixteen lines and, by the very nature of the Gnoetry program, chose some lines over others, often researching dozens of possibilities before landing on the “right” line. Gnoetry acts like what the late Terence McKenna called a “self-transforming machine elf”: it throws out linguistic possibilities, asking “is this it?” “do you like this!” “how about this one!” all within a fixed form that the human must take some care to maintain the integrity of.

As I write this, Jon Trowbridge—the Programmer—is working on a new version of Gnoetry which will enhance the collaborative process, lending more power to both the machine and the human co-author.

Actually when I wrote the comment I did think about how this compares to reading the work of two collaborating human authors. At the time I considering writing, with humans we can suspect who wrote what, but we don’t generally have to consider that one of them might not be capable of human-level writing in the first place; instead we wonder did one of them do most/all the work and the other just share credit.

Thanks for the further info about the collaboration. Have you considerend appending a similar an explanation to the poems themselves? Or would that feel too much like explanatory texts on art museum walls?

How do you (and anyone else who wants to comment) feel about the usefulness or even necessity of such explanations? Are they unnecessary? Should readers take it on faith, or be purposefully left to wonder and try to figure out, the nature of the human-machine collaboration? As I suggested earlier, I think one of the interesting things here is that readers can become the collaborators themselves if they wish.

I like the idea of readers as collaborators (not in the sense you mean, although I’m certainly in favor of anyone having access to Gnoetry as a tool) which I think follows quite logically with what happens to meaning when one reads a work of human/machine collaboration; because there is no “psychology” as we understand it in relation to artistic creation, the gnoems (or any other readable computer-assisted language) are more open systems, allowing for the reader to decide where the meanings are. The reader, placed in a starting position of wondering what’s human and what’s not, has eventually to decide what the gnoem means without the groundwork of intent; talent; social, historical or political formation; any of the typical governors of meaning a reader brings to a purely human text (even the most occluded, difficult work of a Language Poet hard-liner). Too much explication would mitigate that productive “wonder” you point to: a wonder, perhaps, at the fact that a machine was involved at all (speaking of the typical humanities laden reader of poetry, perhaps) or a wonder that begins with the question of “who/what did what” and then moves on from there.

I’ve read gnoems at poetry readings, and people often mistake the gnoems for my own productions. Someone could randomly find a gnoem, I imagine, and not see anything machine-like in it, just a good or bad poem. This differentiates Gnoetry from bot language; people are easily fooled in that context, which probably says more about the level of linguistic expectations than it does about the ability of the bot to “sound human”–it takes much more to get verse to sound like verse. With poetry, the stakes shouldn’t stay within the realm of what’s human and what’s not, in any case. I would hope that readers would not spend most of their time trying to de-code the poems with only that scope in mind. More explanation, I think, gives the reader to many opportunities to be dismissive, rather than to take the gnoem on its own terms. Like when a poet introduces a poem with far too much “background”: the need for the poem is all but eclipsed when a poet begins with “this poem is about…”