I like it because you can get more granular data state by state, which is nice. But, I've noticed some errors that I think are due to automation. Probably no one cares about this but me. So, I'd recommend you stop reading now. If you finish reading this long narrative, I suspect you'll find that it wasn't worth your time.

This is harder to see now because they've made some changes. But, early on in the epidemic, when there weren't that many cases, that site used to number each new diagnosed case, and would show the number next to the notes. So, it would say, for example, "Man is his 80s in Seattle diagnosed" and there would be a column next to it that showed that man was, say, the 75th diagnosed case in the U.S. You could then go and see all the other reports about that man.

When it showed things that way, you could see that if there were multiple reports about an individual (say, a first report about a woman being diagnosed and then a second report days later about that woman dying), it would credit the death as occurring on the date of the first report, rather on the date of death. One obvious example of this that you can still see is the very first reported death for Washington. If you filter for deaths only, you'll see that it has two reports on February 29, 2020--one is of a man in Washington dying (the first U.S. death) and the other is of a woman in Washington being diagnosed. That woman later died sometime in March. But, it credits her death as occurring on February 29, the day she was diagnosed. That's why that site shows 2 U.S. deaths on February 29, where all the other sites show only 1.

Now that there are hundreds of deaths per day, little errors like this probably don't matter. But, I noticed it and it bothered me back when there were only a few deaths per day.

I warned you this wouldn't be worth reading.