There are more cases, they are on average much more mild, the real mortality rate is much lower, it is actually not doubling every two days (it is the tests that are doubling every few days). Lots of possible and very consequential conclusions.

Deaths follow a predictable exponent too.

But with several weeks worth of lag. The number of deaths today should be divided by the number of cases N>10 days ago. The number of cases was much more severely underestimated N days ago than today. The death rate in Italy is basically unknown because the denominator is just noise with a time lag.