netpoetic.com

Here is the author’s description:

“The Longest Poem in the World is composed by aggregating real-time public twitter updates and selecting those that rhyme. It is constantly growing at ~4000 verses / day. You can see more verses by clicking the three dots at the bottom (• • •) Made by Andrei Gheorghe.”

It’s interesting to read a bit of this. And if you reload it, it changes. So perhaps it is being added to top-up rather than bottom-down.

The almost-perfect Twitter longest poem in the world. Shoot me now. No, wait; it has not quite unfurled.

We read at edition.cnn.com that “Gheorghe said he doesn’t consider his work art. “It was just a random idea that popped up and I played with it. And it is what it is.”" Gheorghe is a web developer from Romania.

I expect it is art. There are all sorts of different types of art. From pop and viral to esoterica. This is definitely viral. And possibly pop. And definitely computer and net art.

It’s a really good code idea. Random tweets one after another would be dismal, but tweets with the same syllable count that also rhyme, why, they twitter quite delightfully, as it turns out.

Here we have a feed-based piece that saves Twitter from itself. It’s doing something interesting with a very odd database.

When I read it on Friday afternoon, it was all about people getting ready for the weekend.  A poem in rhyming couplets that may chronicle the time but is continually atwitter with the buzz of the moment. Remarkable.

Wallace Stevens said that poetry is a matter of making silk purses out of sows’ ears. I don’t know how silky this one is but surely it’s made of almost a hundred percent sows’ ears.

And how would you implement the idea? You’d need a rhyming dictionary (like this one). That is stored on the server. It isn’t downloaded to the client. All the writing happens server-side. It isn’t clear what language the server-side processing is done with. PHP or Perl or ASP or Java or whatever. The rhyming dictionary would probably have an API for that language. So that you could, say, send the dictionary a word and get back all the words that rhyme with it. Or you could send the dictionary two words and get back a yes or no as to whether they rhyme.

If it was a good rhyming dictionary, you could also send it a word and get back a number. This number would be a label for the set of words that rhymes with the word you sent. So you get the same number back if you send, on different occassions, two words that rhyme. And you get different numbers back if, on different occassions, you send two words that don’t rhyme.

Rhyming is an equivalence relation.  It partitions the set of words into disjoint sets. Each of the sets in the partition is made up of words that rhyme with one another.

Of course you also need access to lots of current tweets. I’m not sure how this would go. Would this be limited to a set of your friends or what? Anyone have a sense of how that would go?

If you wanted to make it so that each two tweets in a rhyming couplet have the same number of syllables you have to be able to count syllables in a word. In English, that’s usually a matter of the number of vowel clusters. For instance, ‘beautiful’ has 3 syllables and 3 vowel clusters: ‘eau’, ‘i’, and ‘u’. There are exceptions to this rule, such as ‘estuary’, which has only 3 vowel clusters (‘e’, ‘ua’, ‘y’) but 4 syllables, or ‘one’, which has 2 vowel clusters but only 1 syllable, but the rule works broadly enough to be of use.

So for each tweet, you have a few pieces of info.

1. The tweet itself.
2. The URL of the tweet. 
3. The number of syllables in the tweet.
4. The last word in the tweet.
5. And all the words that rhyme with the last word in the tweet.

Alternative to 5 would be the number label for the set of rhyming words, if the dictionary had that feature (discussed above).

Given two tweets, to determine if they can make a couplet, you compare item 3 in each tweet to see if they’re the same. And you compare item 4 to make sure they’re different (so you don’t rhyme x with x). And you compare item 5 to make sure they’re the same or that the one word is in the set of the other word.

If that pans out, you write the couplet to the poem, which is maintained on the server. Top-up. In other words, when you add a couplet to the poem, you add it to the top of the poem.

And when somebody visits the site to look at the poem, you feed them the most recently generated couplets.

by Jim Andrews
vispo.com

Tags: computer art | e-literature | Jim Andrews | literature | materialism | poetry | programming

You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.