Let me try to shed some light.
At this point we still have more variables than known figures in our equations, but as time passes we are getting closer. Per the Lancet 2 weeks ago, the best overall current fatality rate estimate is about 2/3 of 1% (of course it varies with age, gender, etc.). Read this for more specific info; it starts with a summary that is pretty concise:
https://www.thelancet.com/journals/laninf/article/PIIS1473-3099(20)30243-7/fulltext
I think that is probably the most reliable statistic at this time. Previous fatality estimates were considered to be too high for a number of reasons, not the least of which was underestimating how many had been exposed and had minimal symptoms. That means that with 20K + deaths in the US as of last week, somewhere in the 3-4 million range had been infected. Given how we have thoroughly screwed the pooch on testing, that number makes sense. Bear in mind also that fatalities is a trailing statistic; people who die seem to typically do it (peak of the bell curve) in week 2 or early week 3. As long as the hospital admittances are still going up geometrically, that suggests that basing exposure #'s on fatalities is probably going to provide exposure estimates that are about 2 weeks old. So again, based on geometric admittance increases, it is probably safe to assume that at this point 4-5 million have been exposed & sick rather than 3-4 million.
Let's take the high figure....5 million exposed as of now. That is only about 1.5% of the population. Not even a pimple on the a$$ of herd immunity. If the whole social distancing game plan can flatten admittances at some point so we don't overwhelm the medical facilities, then we will gradually work through the population and be able to hold fatalities under 1%. Hopefully well under 1%. If the medical facilities are overwhelmed, then all bets are off and a fatality rate of 3% is not an absurd consequence. If we can actually cause the national admittances to decline...not just flatten...then we have a good chance at holding this at bay until we have a vaccine. Or, lacking that, an effective treatment. It looks now as though we are
at least 8-10 weeks from knowing if any of the current Hail Mary attempts at a treatment will work. That puts us to early July, when according to the start of this thread, it is conceivable that we may see football practices start. If we manage to flatten but not cause a decline in admittances...in other words, freeze the rate of new infections where it is right now...then the math would suggest that at that point we probably have somewhere in the ballpark of 20-30 million who have had the disease. Of course, if we continue to increase the exposure rate, then the number who have had the disease will be higher, and if the increase continues to be geometric, it could be a lot higher. Along with an increase in exposure comes an increase in deaths; an 0.66% fatality rate for 34 million exposed (about 10% of our population) works out to about 200K deaths at that point. Estimates of how many people have to have the disease in order to establish herd immunity are in the 80-90% range. That means that if we do nothing more than we are doing now and never develop either a treatment or a vaccine (and I expect we will eventually have both, but I have no firm bets on how long it will take), and 80% of our people end up being ill, we'll finish up with just under 2 million dead. That is where that 2 million figure comes from when you see it.
If we have a vaccine by July then I think we will see football in the fall under some sort of schedule, but I think a vaccine is probably October at the soonest, and maybe not until after Christmas. If we have a treatment that can be demonstrated to be effective by the start of July, I could see practices starting as soon as late August or September. If we have neither, then there will be no sports in the fall, and all classes will continue to be on line until we have one or the other, or we burn through 80-ish percent of the population. How fast we burn through the population depends on the success of the social distancing efforts, and in the final analysis the fatality rate will probably depend upon whether our medical system can be a "bend but don't break" defense. Because if they break, the fatality rate will go up.