IIRC there is at least one paper, maybe around 2005, where they were able to determine what was being typed in an encrypted ssh session, using packet timings co

IIRC there is at least one paper, maybe around 2005, where they were able to determine what was being typed in an encrypted ssh session, using packet timings correlated to collected human typing statistics. Looks like this adds noise to prevent that.