Most of the times you don't "accept" results. You have to build something on them, like an extension or a similar version on other field. So usually the first step is try to understand the cryptic published version and do a reproduction or something as close as possible.

The exact reproductions is never published, because journals don't accept them, but if you add a few tweaks here and there you have a nice seed for an article to publish somewhere.

(I may "accept" an article in a field I don't care, but you probably should not thrust my opinion in fields I don't care.)