There's a logical inconsistency here. If we assume that the total number of cases is much higher than reported, then the hospitalization rate per infection is driven down by the larger denominator. There's a point at which there are so many total cases that the number of hospitalizations is problematic and noticeable. If we continue to assume that the total number of cases is much higher than reported, the death rate per infection (IFR) is also driven down by the larger denominator.

But if we had a large number of undetected cases in the past, then we should also have a large number of undetected cases now, contrary to the study. (Unless we already have herd immunity, but there is no data at all suggesting this.)

Perhaps in the past, earlier in the virus's exponential growth, many people had it but still an order of magnitude less than now. And that order of magnitude is the difference between a slammed hospital and a regular day. And because an order of magnitude people less had it, groups like the elderly were an order of magnitude less likely to be exposed en mass.

I'm not necessarily saying people got infected a long time ago. If anything I think we're just missing way more cases than people assume at this point.