The 'idea' is that a post ... of any length ... is stripped of all characters other than ATCG, and that what remains constitutes a possible genome sequence.

bark moth couple dingo

becomes

atcg

Then you run this 'sequence' against a database of genomic data, and get a nearest match to some species.

The actual problems with this are many.

Actual genomic sequences are orders of magnitude larger than even a lengthy, stripped down tumblr post.

Actually running a nearest match search of a short string / sequence will likely match many, many different species.

Running an exhaustive, ie, accurate, nearest match search against an actually comprehensive database would require using a supercomputer, or at the very least, a lot of powerful, networked computers.

...

EDIT: There's no real relation to the content or context of the post or thread to the 'genomic match' species. None.

Its about as legitimate or useful as trying to decipher the 'Bible Code' by running any number of pattern matching algorithms on a modern, English Bible.