From the examples, sounds like you're the slightly lossy clone of someone else.
Was interested at first but the tweets clearly came from a neural net. Who would consider changing Sphinx to puppy? A distributed word embedding, save as the other modifications. Look into word2vec, RNNs, etc.
My guess is that it's running similar to decaptha in n which their is a human/manual process involved. The pulling of the tweet is automated, scrubbed based on criteria, and then a human changes or confirms the tweet. Indians working off of a thesaurus would be rather inexpensive.
What if it's a bot that gets by with a little help from mturk?