[D] why does the Mamba paper seem to distance itself from the apparent connection with RNN (a la Schmidhuber et al)?

Maybe I missed something, but the similarity with RNN is hard to miss. Yet the paper seems deliberate in mentioning RNN. Is it because the convolutional optimization makes it far more efficient than sequential RNN? Or is Schmidhuber nowadays considered an outcast in ML community due to his overzealous self attribution earlier?

machinelearning mamba optimization paper rnn something