The tricky thing about discussing covid-19 mortality is that there are always confounding variables. The disease is deadlier for people with pre-existing conditions, particularly cardiovascular.

A common line of argument online is "it's just the flu, and we don't panic like this for flu season every year." Observing a spike in total deaths puts that line of thought to the test.

Because it would give people a better intuitive sense of how bad COVID is in a larger civilizational/historical context.

Over the past 12+ months, people have accepted enormous infringements on freedoms in their everyday lives. Many of the public health measures that we take as a matter of course in the "new normal" - lockdowns, social distancing, extended closing of churches, etc.- are things that would have been almost unthinkable as recently as January of 2020.

If you look far enough back, to some point in the twentieth century, you'll find a time at which the baseline infectious disease risk in everyday life was higher than it has been in the COVID-era US. But people at that time went to crowded theaters, ate in restaurants, packed together in churches and schools, and so on. Restricting these things could have decreased their all-cause mortality, but they clearly found that option unacceptable.

Note that this isn't an issue of whether people "believe in science" or "believe in data". It's an issue of risk tolerance. People a few generations ago understood the Germ Theory of Disease perfectly well, and they knew that they could reduce the spread of infectious disease by implementing the kinds of measures we've implemented for COVID. And yet they didn't implement these measures, except in very minor and temporary ways. We've become far more risk-averse than they were. And the question of whether that's a good thing is not an empirical question.

In a vacuum it would give you more of a sense about how much of an outlier this year was. But obviously enough other things have changed over the last 50 years that the numbers would stop being comparable. Perhaps looking at a 5 year sliding window to see how big the peaks usually are relative to their surrounding years could get at this with a bit less confounding.