It's trivially easy to build a scraper for the new content too. And I think people really over estimate the relative amount of data reddit has compared to the l

It's trivially easy to build a scraper for the new content too. And I think people really over estimate the relative amount of data reddit has compared to the literal rest of the internet. Like sure, it's nicely categorized and ranked (somewhat), but leaving reddit out of a corpus probably won't drastically effect your final model.