1.6% is a reasonable estimate, overall, taking into account un-documented cases that are not part of the official count.
Another estimate is to take the average 'days from diagnosis to death', and divide current deaths today by the number of cases that existed that many days ago. Using that metric, and a 4-day window of average diagnosis to death time, the death rate has hung around 4%.
That does not mean the final rate will be 4%, it simply places a reasonable upper-bound on it. And it is a biased estimate, because it is only using cases that were tested for and symptomatic enough to get a test.
If we assume there are 2x as many cases out there which are not documented, then that instantly drops the rate to 2%; very much in line with the 1.5% estimates out there. If we assume there are MANY more undocumented cases, well it is difficult to rectify the outbreaks we are seeing with multiples more cases, without the outbreaks occurring more "in sequence" or more "all at once". That means the estimates of 4x or 5x more "asymptomatic" cases than we are documenting are problematic, and something else has to be going on. Case tracking has not "skyrocketed" with testing, it has been tracking along a linear geometric curve now for almost 4 weeks before we are seeing some tailing off.
What is more realistic is the undetected cases are no more than equal to the detected ones, and our 4% number dropping to 2% is a reasonably sound number at this point.
That is likewise supported by the 'positive test rates' being on the order of about 10% of those tested in most places. Yes, there are a small number of people truly asymptomatic, but that number is more like 5% of the total, not 50% or as many as detected. The "80% don't have any symptoms" nonsense is a bastardization of the actual statistic that "80% do not require hospitalization", and have symptoms ranging from mild cold to really really bad flu, but never having breathing difficulties.
Based on my tracking tables, we've been hanging around that 4% mortality (using 'backward' case numbers of 4 days) for about a week now. Thus, 4% is becoming a stable statistic - 4% of those who are symptomatic enough for a test end up dying of this disease.