T2K, north of 95% seems way too high for asymptomatic cases. Not sure where 4Real got the 83% number, but that seems too high also.
Preliminary reports for NYC show approx 25% of those tested were positive for antibodies. Although the testing isn't a random sample (since they only tested those out and about in specific locations), let's assume that 25% of the city has antibodies. That is about 2.1MM people.
NYC also has about 19,000 deaths and 69,000 hospitalizations (per Hopkins dashboard). If we assume 2.5% of symptomatic cases result in death, the deaths imply that there are 760,000 symptomatic cases in NYC. If we assume that 10% of symptomatic cases result in hospitalization, the hospitalization numbers imply 690,000 symptomatic cases in NYC. Those numbers are fairly close for back-of-envelope estimation. Let's split the difference and say there are 725,000 symptomatic cases in NYC.
That means of the 2.1 million cases, 725,000 (34.5%) are symptomatic and 1,375,000 (65.5%) are asymptomatic.
While 65.5% is a solid majority, it is still far away from 95%.
Interestingly, the Pacific Princess data showed about 50% of cases were asymptomatic. The Pacific Princess data may have been low, due to missing people who were asymptomatic and did not have enough virus load to test positive (either because they were recovering, never that sick, or just testing error). On the other hand, the NYC 65% estimate may be high due to false positives in the antibody test, or lack of random sample. But I think it is reasonable to assume that somewhere in that range is correct.