I've spent some time doing that, as well as learning Toki Pona, and it doesn't seem to me that they're very similar. Latin and Greek have attested vocabularies

I've spent some time doing that, as well as learning Toki Pona, and it doesn't seem to me that they're very similar. Latin and Greek have attested vocabularies of thousands of words, many of which have a super-specific meaning and aren't analyzable as compounds. Just one example is that they both have words for specific plant and animal species, which Toki Pona doesn't. I agree that learning ancient languages and prefixes produces amazing results for our understanding of modern languages that are related to them and is a great idea. :-) I especially don't think that ancient language speakers felt that, just because they had productive prefixes like our re- or un-, they could always describe concepts with combinations of a few existing morphemes instead of inventing new ones or borrowing them from other languages. Also, even the ancient language speakers didn't understand the etymology or structure of their own languages very well; some borrowings and compounds were already opaque to them!