Tag: machine learning

My polit­i­cal the­o­ry of poet­ry is fair­ly straight­for­ward: poet­ry is the most prim­i­tive form of com­mu­ni­ca­tion. Poet­ry is the cre­ative use of a lim­it­ed resource. So the non­sense that comes out of mouths of any­one learn­ing the vocab­u­lary & syn­tax of a new lan­guage is also, clear­ly, poet­ry. As the lim­i­ta­tions of lan­guage lessen, the need for cre­ativ­i­ty lessens, peo­ple opt (typ­i­cal­ly) for the most com­pre­hen­sive method of com­mu­ni­ca­tion.

What a mar­vel the first bird must have been!

I’m going to test this the­o­ry by teach­ing a machine to write poet­ry. I hypoth­e­size this will be much eas­i­er than teach­ing it the intri­ca­cies of typ­i­cal speech. This is a fair­ly easy sup­po­si­tion for me to make, as most machine-gen­er­at­ed texts already read like they are a new lan­guage. I have no idea how long this is going to take, but I’m start­ing here: Machine Learn­ing Mas­tery. Once I can swing the ham­mer cor­rect­ly, I’m going to start my dig­i­tal child pound­ing on the 1000 most com­mon words in Eng­lish.

If I raise it up as it should go, maybe it’ll even take over Poet­ry 4 Free duties.