Incoming weblog traffic volume has been all over the place recently. Some posts that I barely recall even writing from 10 or more years ago are getting a lot of attention. Initially, I wondered if people were just digging around (which happens sometimes) or if it was some type of automated web crawling traffic. Generally the top posts receiving traffic get listed out and sometimes that information is useful. My traffic analysis is really just in visits and does not have a metric for time spent. That would make it easier to figure out if the uptick in traffic was natural based on people reading things or just some automated scraping traffic.
I have been working on training a GPT-2 model on my entire online corpus. Twenty years worth of weblog content is enough to reasonably train that model to do things. Telling the difference between synthetic text and automated traffic is a lot harder now that it used to be based on the quality of earlier models. My efforts are generally aimed at seeing how a well trained instance of the GPT-2 model would write weblog posts based on topics. A real test would be for me to sit down and start to write about something and pick a topic to compare my efforts to the output from from the trained model. That would be a very simple A vs. B test that would give me a good indication of the difference between my 20 years of personal training in writing weblog posts and the training of a model based on the output of that 20 year weblog corpus.
Another behavior of mine has changed a little bit in the last few days that is worth noting. Those posts that seem to be getting a little more traffic are getting some grammarian style grooming. I’m not really editing them for content in any way shape or form, but they are getting a few words filled in to make them more readable. I really do have a bad writing habit of leaving out certain filler words when I pick up typing speed. The faster I seem to go, the more prevalent it is that a word or two will be missing from a sentence. That does not generally change the meaning of what was written, but it does make it harder to read. Looking back on something that I wrote over a decade ago those words seem distinct to me anyway. Doing a little cleanup is probably the right thing to do for readability and the general comfort of the reading public that might consume that prose. Yeah —- that feels like a radical change in position from my previous thoughts on the subject matter. It probably is a radical change in position. It might be a milestone of personal growth or something.