It's a shame when potentially interesting papers don't hold up in practice. I've seen a few cases where the real-world performance didn't match the initial claims.