If I were king with absolute ruling power, my "long term play" would go something like this.

#1 Restore Freedom to Low Risk Groups Right Now and Continue the Satus Quo for Middle Groups & Stricter Limits for High Risk Groups (starting now until #2 below is done). Based on the data from Italy on hospitalizations and deaths (their infections data is clearly way too low because of the unavailability of testing just like in China and the US), it appears that this is a very dangerous virus for people over 50 with risk factors, but children and young adults are a very low risk for hospitalization and/or death from this (even though they can still get really sick). It appears that basically everyone under 35 may be about as likely to die from a car crash or the flu as they are from this virus. Yet, we have very little information about how likely such persons are to be infected if exposed or sympotmatic if infected because we don't really test people who aren't highly symptomatic and kids and young adults simply are not filling up our hospitals from cornoavirus symptoms. It's insane to me that we shut down day cares (where are the medical staff's kids going to go, BTW?), as well as live instruction at primary and secondary schools and Universities when these groups are the least likely to be impacted. People can work from home and the economy can keep moving forward, but not if they have to watch their kids at the same time. It's tough to imagine a scenario where those decisions aren't viewed by history as a major blunder. In addition to going online, Universities in their exhuberance to show that they're taking this virus seriously, have gone so far as to close most all of the residence halls on campuses so that college kids have to go home. Keeping low risk individuals together so that nature can run its course among individuals who won't need hospitalization and can build herd immunity rather than sending them home to intermingle with a potentially elderly parent or grandparent would have been a far better approach. Not to mention Universities wouldn't need bailouts if students stayed in the residence halls to complete their coursework online and if we could guarantee better internet connections for our students then we wouldn't have to go to teh silly Pass-Fail grading (but I digress).

We're painting everyone with the same brush rather than making wise decisions based on the data we already have, which I think is a mistake. The justification for treating everyone the same is that the low risk people can spread the virus to the high risk crowd, but by taking a middle of the road approach we're keeping low risk people from developing immunity and destroying our economy as many low risk individuals cannot work from home. All the while, we're still exposing high risk people to the disease. We need stricter guidelines than social distancing for individuals who are clearly high risk (either over 70 or between 55 and 70 with significant risk factors like smoking, obesity, diabetes, etc.). Persons in the "high risk" group should be strictly harboring in place at home and making arrangments to do so for several weeks. I believe the high risk crowd is the only group that really should be holed up until the virus has run its course and this group should take precautions that assume everyone not in that group is already infected. Care facilities have put these measures into place, but individuals in other settings need to do the same. Also, caregivers in these settings (whether family members or healthcare professionals) should be permitted and encouraged to be regularly tested for the virus (weekly if possible) so the infections are identified early before spreading to other high risk individuals happens and any high risk individual could then begin treatment to mitigate the effects before the symptoms become severe. The 35-55 "middle risk" crowd should be taking a middle approach using mostly social distancing, hand washing, and working from home wherever possible without disrupting their jobs (this would be a lot more effective if these persons weren't also providing child care and home-school to their kids, BTW). This group should be doing what the less restrictive jurisdictions were doing early on and, while that does increase risks for a higher risk crowd, many in this group will build herd immunity as well. If we're completely cutting off the high risk crowd from the virus then the middle group should be getting most of the medical attention by hospitals and would spare our healthcare system from using up all the respirators and ICU beds on 80 year old patients who are unlikely to recover anyway.

#2 Test for a Month (Over the next 4-6 weeks). We should immediately drastically increase our testing ability (I'm talking tens of millions of tests ASAP rather than a couple hundred thousand) as well as the precision in those tests (how about testing for anti-bodies rather than just current infections???). If we did this then: (i) we could test random samples or even entire populations like a couple of towns or neighborhoods within cities intentionally identifying different demographics and living conditions so we can observe variance and drastically improve the accuracy of models with reliable hosptalization/mortality rates across different age and risk factors, and (ii) provide more widespread testing of individuals who are exposed rather than just testing individuals with severe symptoms so we stop flying blind.This testing is going to happen eventually and many papers will be published in the coming years on this topic, so why not fast-track that process with a healthy dose of funding so we have better data now rather than for the next pandemic? The ramp up on testing would probably be very costly and take several days (I don't know enough about supply chains and technology in this area, but I have to think best case scenario is at least a couple of weeks with hundreds of millions of dollars) and then once we have the tests the data gathering and analysis would take a couple more weeks, so I'd give this phase 30-45 days. Knowing how wide-spread this thing has already been is critical to making good decisions and we simply cannot know that without an anti-body test and testing of either random samples or entire population segments.

#3 Restore Freedoms to Newly Identified Low Risk Groups and Phase in Restoring Freedoms to Higher Risk Groups Over Time (Starting in 4-6 weeks). There are likely many groups in the under 65 crowd who are not at risk of severe symptoms under coronavirus, but we're basically shooting in the dark on who those groups are because we don't have reliable data on exposure and infections and, instead, we are basically relying on hospitalizations and mortality numbers alone. The infection rate for an exposed person could be 10% or it could be 80%, but we don't really know because we aren't testing to find out. This means we're basically driving a car using the rear-view mirror and so policy-makers are going to drive very slowly when it comes to lifting restrictions (i.e. assume the worst and hope for the best). If we had better data we could restore freedoms and make recommendations to permit people to return to normal life much quicker than using the "confirmed cases" data curve to guestimate the right time for everyone to return to normal activity. We could also use testing to identify hot spots at the county or region level directed at imposing no greater restrictions than are necessary rather than making decisions at the Federal or State level.

#4 Back to Normal (probably in 2-3 Months). Finally, we could figure out when it's okay for the "high risk" crowd to return to normal life within particular regions (again by actually testing large random samples to ensure active infections are not a high risk) and figure out when it's okay for all restrictions to be lifted. As it stands right now, this date will come as a surprise to everyone because of all of the variance in the models. That's one reason why #2 above is critical. We need to stop shooting in the dark in hopes of hitting a target and/or focusing so much on "flattening the curve" when we're truly only flattening our economy and setting ourselves up for an inevitable outbreak later, which may be what we're doing now.

#3B Alternative Timeline (4-6 weeks). There's a genuinely possible alternative timeline that we may discover in #2 above. Some doctors guestimate that 20% of persons who are infected need hospitalization, but those figures are likely way higher than the true number because we're basing that on confirmed cases and not actual infections or exposure that we know are much higher numbers. Asymptomatic or low symptom persons are going unnoticed, which means there could be millions of people who have already been exposed and/or infected and either have natural immunity or have built up immunity through recovering from an infection with no symptoms and we don't realize it. We're primarily testing only really sick people (hence our "confirmed cases" data is understating infections due to selection bias and lack of testing availablity) and since we're using unrealiable data on total infections from Italy and China to extrapolate our hospitalization and death numbers into projections of total infections, we could be way off overall. If actual infections are 2-3 times the "confirmed cases" number and we are able to reasonably estimate natural immunity then our data and projections are probably pretty accurate, but if actual infections are more on the scale of fifty times the number of current confirmed cases and more people are resistant than originally estimated, then our projections for total cases aren't even close. Yesterday's US "confirmed cases" number was 163K, so if we're only catching one out of fifty infections (including those already recovered from the virus) then we have 8.1M past and present infections already (about 2.4% of the US population). If two-thirds of the US will be exposed and only 10% of persons exposed will actually catch the virus, that'd be 21.7M infections and I'd estimate between 50-150K deaths total. I don't believe a 1% CFR is credible at all and so I'm using 0.23%-0.69% mortality rate, which I think is reasonable based on adjusting the Diamond Princess Cruise Data to a population that is consistent with the average age of the US. That cruise had 3711 crew members and guests all which were exposed to recirculated cruise ship air for two weeks that likely had the virus and resulted in 712 infections (19%), 10 deaths (1.4% of infected, 0.26% of those likely exposed), but the cruise guests were much older on average than the US population so I'm adjusting the infection rate down from 19% to 10% and the mortality down from 1.4% to a range of 0.23%-0.69% to calculate the numbers above.  

If my guestimations above (which are honestly just about as good or better than many of the crappy models circulating around), then this would mean we could be approaching the halfway point on infections right now (though deaths are lagged at least a week or so on average after the confirmed infection date and so the death curve wouldn't hit its halfway point for at lease 2-3 weeks). If this alternative scenario is the case, then our hospitals will not be over-run for very long and our social distancing/stay-at-home efforts are actually only backfiring by slowing the spread when it could all be over in a couple more weeks if we just acted more normally except for protecting at-risk individuals. There are reasons I believe this could be the case, but assuming this to be the case without better data would be taking a horrible risk as minor adjustments to the total actual infections, the infection rates from exposure, and/or the hospitalization/CFR rates can have a huge impact and easily tip the apple-cart on our entire healthcare system. The Diamond Princess Cruise data is still a small sample and we can't know that 100% of the persons on board were exposed in a way that is similar to the way that 70% of the US population would be exposed (or even if 70% would be exposed), so there's a ton of variance there. With major public policy decisions and pandemic projections, if you put crappy data into them you get crappy outcomes out of them. This is why step #2 above is absolutely critical and it's shocking to me that this wasn't done in January when the virus was raging in China and we knew it was very deadly and likely to arrive here (or already was here) or in February when the outbreak in Washington was getting bad.