[alexpetros]

Hi all, I've been lurking in the comments all day and I wanted to let the HN community how much we appreciate the wonderful response. We were able to identify and fix a handful issues very quickly (communicating the rate limit when we first hit it, the text getting wiped if the request didn't succeed) because of the feedback we saw here first. So if you're one of the commenters who left a message about something broken: sorry, and thank you!

The purpose of the site, to the extent that it has one, is to interactively demonstrate the contours of what machine learning text generation can accomplish. There a lot of demos and news stories about what "machine learning" might replace, but if you play around with it in its current form, you get an understanding of its limitations—and potential for amazement. I'm sure there are others who could train the model better than we did, but I think the overall experience would end up similar. There are patterns in how humans justify our actions, but there are also wonderfully jagged edges. When used appropriately, computers can identify some those patterns, but they are fundamentally "guessing and checking" when they attempt replicate them. Those are just our ideas though—the experiences other people have been throwing out there have been far more interesting.

Anyway, thanks again for your patience and enthusiasm, please continue to let us know if you have issues, and here's a real world dilemma I've been having:

> AITA for reading a lot books about Common LISP lately?

YTA. It's just super confusing. You don't need the extra work that goes into common lisp. I wouldn't even call it a language, but for whatever reason people like to write books about it. It seems to be used as clickbait title more than anything else too

[brabel]

Just want to mention I asked a few "real" questions looking for advice, and got actually useful, illuminating answers!! This is very cool.