Jul. 14th, 2013

rinue: (eyecon)
If you ever want to guarantee that I will block all future e-mails from you (which, I don't know, you might have reasons to want sometime), try to draft something which combines Lacan with free-floating paranoia and ideally namedrops at least two different non-mainstream medical approaches and/or spriritual paths without relating them to the rest of the comment/e-mail. Pretty much regardless of context, that is going to be perfect trolling, and will be even more effective if the topic is science or current events and I have asked you a direct question about the factual basis for your predictions. (Because you are a Lacanian, they will turn out to have been free associations and not facts, in which words were given meanings unrelated to their expressed meanings, both denotative and connotative, and my attempt to approach a topic by talking about that topic will show how little I understand reality and have been co-opted by the system. What system? The fact I ask that question shows I am being literal again.)

I really don't understand where these people come from. If I, a poet and Absurdist who hangs out with Dadists, can't stand you, I don't understand who it is that tolerates this kind of behavior.
rinue: (Default)
A few days ago, Sofia Samatar's blog linked to something called the Poetry Assessor, which works something like a cross between that old Microsoft Word function that told you the reading level of your document (4th grade level of comprehension? 9th grade? Postgrad?) and the internet widget that tells you which author you sound like. Specifically, the Poetry Assessor tells you how professional your poem is (or how confident it is your poem is professional), on a scale of something to something (unclear, but I'm guessing -5 to +5). Positive numbers mean professional. Negative numbers mean amateur.

As I understand the Poetry Assessor, it was a machine learning project - a researcher fed a lot of poems into a computer, some of which were professional (for example published in critically respected journals) and some of which were amateur (source unknown to me), and from this the computer built rules to understand what professionals did differently from amateurs. I do not know how large a data set it used.

The code-breaking, pattern-finding area of my brain loves looking at computer outputs and working backward to guess its rule set, so I started feeding in a lot of poems. I know from the intro text of the assessor already that Sylvia Plath’s poem "Crossing the Water" scores 2.53. To further establish a kind of benchmark, I fed in two favorite poems of mine, to see if it was fooled by the casualness of Frank O'Hara's "Poem (Lana Turner Has Collapsed)" (+1.8, professional but not as professional as Sylvia Plath, which fair enough), and whether it agreed with the widely-held critical point of view that Yeats's "The Second Coming" is one of the finest works in English.

That one got a +1.6, more amateur than Frank O'Hara, which is funny since Frank O'Hara tended to define himself as amateur, kept a day job, and mostly left it to his friends to collect and publish poems he left on the backs of receipts and things. Whereas Yeats was a guiding force in the Irish literary scene and British theatre and won the Nobel prize for literature. So we are maybe using interesting definitions of amateur and professional. I next fed in my first published poem, which I wrote when I was 4, which is called Nutkin (after the Beatrix Potter squirrel character, obviously).

I'm a little Nutkin
In my hutkin
Eating futkin
In my hutkin

+2.5, way more pro than both O'Hara and Yeats. On the one hand, were they writing things of this caliber when they were 4? On the other hand, the Assessor has trouble with poems of fewer than 80 words, which is most of my poems and certainly "Nutkin." Following a hunch, I tried "Jabberwocky" by Lewis Carroll. +4.3, the highest score I have discovered for any poem. Taken together with "Nutkin," this seems to suggest that amateurs don't invent a lot of words. You have to have a certain level of confidence to believe that you will be taken seriously if you say a something like "futkin" or "snicker-snack." I conclude that nonsense is supremely pro.

At this point, I started feeding in a lot of my poems, despite the 80-word problem, to try to establish a baseline for "amateur." The most amateur I found was "Tomato, Avocado, Dill," (-2.4, fewer than 80 words), which is patterned after a jump-rope rhyme. So I theorized that perhaps some of the amateur poems fed to the machine were nursery rhymes. I tried Mother Goose's "Jack and Jill" (using both original verses but none of the edited or added ones) and came out with +1.6, Yeats level, which killed that idea. "Cinderella, dressed in yella" scored +0.2, meaning that the Assessor had no idea what to make of it. I gave up trying nursery rhymes and jump rope rhymes, given that all of them I know are fewer than 80 words and therefore couldn't be the amateur standard.

I have still not figured out the characteristics of amateur poems. The most amateur of my poems that is over 80 words is "Forecasts" (-1.2), which I suspect is because it uses "you" and "they" a lot, which seems to be of a pattern with poems I hear at amateur open mics. But I was not able to replicate this with further examples. Some of my other poems (mostly over 80 words):

Biology - -1.0
Third Film in the Trilogy - -0.9
Lloyd Blankfein's Testimony - -0.8
Science Has Been Found Wanting - -0.2
World Dryer - +0.2
Eulogy for Johanna Rasor - +0.3
A Bear of Very Little Ambition - +0.5
Dallas 2009 - +0.7
Hymn of the Nocturnal - +1.0
Loose Talk - +1.1
Summer and Austin Have Left Their Apartment For A House - +1.2
A Test of the Great Man Theory of History - +1.4
Remainders - +1.5
Ingmar Bergman Has Died - +2.2
When Giving Figs - +2.6
Testament - +3.0
Dog Whistle - +3.0

Taken together, thinking also of the Yeats rating versus the Plath rating, I think the Assessor may have a preference for Imagist poems and a bias against Symbolist poems, simply because you are going to get very few false positives with Imagist poems; there are not a lot of amateur poets writing them. Imagist poems tend to be vivid, journalistic, and unsentimental, whereas I would characterize most amateur poetry as cliched, autobiographical, and highly sentimental. The problem is that I would also characterize a reasonable amount of professional poetry as autobiographical, and it often uses cliches in order to subvert them or comment on them. (It is not, however, openly sentimental. Contemporary critical tastes do not run in that direction.)

Test case: "She Walks in Beauty" by Lord Byron. -0.8. Amateur. I am maybe on to something.

Test case 2: I search "romantic poem" on the Internet and find a poem posted in a forum by someone I know literally nothing about. That poem is called "A Moment of Truth". Rhymes. Uses "you" a lot and "love" a lot. Highly sentimental, a lot of cliches, presumably autobiographical but not particularly specific. Boom: -4.8. As amateur as "Jabberwocky" is pro. I have cracked it.

Alas I do not think I will be able to use this to automate Strange Horizons slush reading; I have run a few randomly selected submissions through (I won't say whose and don't remember) and do not find much agreement between the algorithm and my inclinations.

Profile

rinue: (Default)
rinue

April 2025

S M T W T F S
  12 345
6789101112
13141516171819
20212223242526
27282930   

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Jul. 1st, 2025 12:04 am
Powered by Dreamwidth Studios