No, I was trying to say what I said.
I cannot believe you still do not understand this. Or maybe you do and just really mangled your version of what you think I said.
The CDC totals on Rob’s page are all the deaths *known* on whatever date. They are what they are and they don’t get updated retrospectively. The update *is* the next total number known on the next day.
And so yeah, because if reporting lag, you can NEVER take the total number of deaths *known*on day X, subtract the number *known* on day X-1 and call that a daily death count for day X. Because some deaths that happened on day X-2 (or X-3,...), for example, weren’t known on day X-1 but *are* known on day X. Plus many reports for deaths that occurred on day X won’t be at the CDC on day X. You just get a meaningless mess when you do that.
If I want to know how many people died today, I need to add up all the reports of deaths that occurred today, ie whose reports are DATED today. These will reach the CDC over a period of 1-2 weeks, so the numbers before then are always at least somewhat incomplete. That’s why separate data sets are kept for this type of statistical analysis.