It's still a great data set because it's real-world, useful, high-quality, and an excellent size (~60k examples), and the "correct" classification isn't very su

It's still a great data set because it's real-world, useful, high-quality, and an excellent size (~60k examples), and the "correct" classification isn't very subjective at all - humans can easily get 99% accuracy.