Maybe someone can answer this for me: why are the estimates on asymptomatic cases so wildly disparate? This seems like a really important stat to know in terms of infection risk. Shouldn't a randomized sample be able to give a decently clear sense of that?
Because there are a lot of variables that cause that range to be significant.
It would be highly dependent on the "randomized" population which is a hard thing to control for. First, because it depends on whom you are infecting - if you are an asymptomatic relatively young person employed in a long-term care facility, it's a good bet that most of the people you infect will subsequently be symptomatic. Whereas maybe a 20-year-old warehouse employee who has 4 other healthy young, roommates may primarily infect those who would not be symptomatic. We also don't really fully understand the viral load but there is evidence that those exposed to a high viral load tend to get very ill - so likely there is a range of viral load from asymptomatic to somebody needing a ventilator. If you are infected and your viral load is high, the % of people that you infect who will be symptomatic will then be that much higher than average.