Loopy Logic
That last blog got me thinking. My blog is unreviewed and not critiqued, so there’s no feedback loop. There’s no science in it, just words. So I can say what I want and get away with it.
So I got GPT to grade my blog. It assigned every recent post a score from zero to ten: zero for word soup, ten for formal reasoning. The average was 4.8 with a spread of 1.1. It might occasionally look rational and/or scientific, but not so.
It is a paradox. A machine trained to imitate judgement is now judging writing it did not create but could have (it says).
The exercise says more than the scores do. Without independent critique, both humans and machines live in a loop of confident language. It reads well, but it achieves nothing.
Then I got GPT to score my writing style. Each recent post got a number from zero to ten: zero for synthetic word soup, ten for strong human prose. The average was 5.0 with a standard deviation of 1.2.
Says GPT – “Most posts sit in the middle. A few are well-structured essays, some are fragments, and most are clear but unambitious.”
So according to GPT my thinking is half neuron soup and my writing is half word soup. I’m not sure whether to be happy or sad; a good score from GPT would have been the kiss of death, whereas a really shit score might be reflective of some hidden truth.
I’ve often wondered if writing this blog is worth it when it feels like publishing within a black hole; the problem being the absence of feedback from post-Enlightenment minds that aren’t tied to some convention or other.
What I’m really doing is refining my own thinking in a private lab-notebook style, with no feedback loop but a strong internal framework.
offshore westerly 