Again, based on what? I have seen nowhere near that level of estimate assuming peak hospital overload and unmitigated spread.

Stats from Italy are going to be heavily skewed towards fatalities for a number of reasons that have been documented by epidemiologists.

Based on the death rates in Wuhan and in Italy where this actually happened.

Careful with those death rates, my guess is that they likely exclude a ton of mildly or asymptomatic patients who never got tested (skewing the death rate upwards).

Sure. And yes, Italy's population is more elderly as well. So I agree it's possible the death rate will be less than 5%. (Although I believe they're currently at ~8%, so we don't know that.) But those are the closest parallels we've seen so far as to what happens when a modern heathcare system gets overwhelmed by this, so they're the best estimates we have to work from.

And regardless, it would be very, very bad. If it's only 5M people instead of 15M, that's still catastrophic. Also while the hospitals are completely overwhelmed with COVID-19, people will die of other things that would normally be survivable.

And finally, we have an alternative: keep employing distancing and, where necessary, lockdown measures while scaling up testing. These lockdowns will get the cases under control, but they would flare back up as soon as they're ended. However, with the expanded testing, we can instead perform sufficient testing across the entire population to catch outbreaks before they start. With sufficient testing, things can go back to normal while remaining confident that we're not spreading the virus undetected and seeding a new outbreak. It needs to be orders of magnitude more than we have now, but it's entirely possible.