Hacker Newsnew | past | comments | ask | show | jobs | submit | hfhdndnfbdbd's commentslogin

I don't see why "western civilization" should be a relevant category for a technology release. You're leaving the EU out, yes, and that's relevant in itself. I don't get why leaving X% of western civilization or not would add anything to it.

(Btw, I consider latinamerica to be mostly part of western civilization...eastern europe too, so the EU is much less than 50%. However, people that care about the term usually prefer to exclude non white people for some reason)


The point is spotting the inconsistency within the document, not whether an original has a different form


I think people in the comments are missing the point.

I might be wrong, but the point isn't comparing a modified The Great Gatsby to the original one. Of course that's not impressive and it's an easy thing to do.

The point of the exercise is supposed to be[1] that the model has the entire novel as context / prompt and so can identify (within that context), whether a paragraph is out of place. That is impressive and I wouldn't know how to find that programmatically (would you have a list of "modern" words to check? But maybe the out of place thing is hidden in the meaning and there's no out of place, modern word).

[1] I say supposed to be because the great Gatsby is in the training sample and so maybe there is a sense in which the model "contains" the original text and in some way is doing "just" a comparison. A better test would be try with a novel or document that the model hasn't seen...or at least something not as famous as the great Gatsby.


Consider applying for YC's Summer 2026 batch! Applications are open till May 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: