> Recently I thought about it again and wrote a Python script that tried it 30 million times and... it never failed. There are enough possible card sequences t

> Recently I thought about it again and wrote a Python script that tried it 30 million times and... it never failed. There are enough possible card sequences that you might be concerned about your source of pseudorandomness failing to exhaust the space. Simulations can give you very misleading results when that happens. Even if you do have enough entropy, 30 million trials is definitely not enough.