Not at all! You are the one who insisted that it was exponential, and I showed it was not.
You insisted that it was linear, and therefore not any kind of non-linear function.
The daily increase has been roughly linear. Which means that the overall numbers are not increasing exponentially. It's not that hard to understand.
It is not linear. I showed that in great detail in my post. Then you dishonestly ignored all that and switched to a red-herring about which exact non-linear function it is, when you're the one who started by casting linear and exponential as the only two options, as though you don't realize that Quadratic is non-linear too.
The 1day increase in deaths from April 3 to 4 was 35% greater than the 1day increase just 2 days prior. How is that "roughly linear"?
And a model presuming linear increase from March 31 predicts only 1/4th the total number of deaths 3 weeks from now than what is predicted by the actual data as of April 4th, when you claimed it was linear. When the data predict 4 times what a linear model predicts, then the data are not showing a "roughly linear" pattern.
"daily cases look very linear" "Which would be linear, not exponentiell."
Are you so upset you switched to German at the end?
That's a direct quote of your own words. Are you upset you've just mocked yourself for poor spelling? I have no need to misquote you to make you look stupid.
The difference between exactly which type of non-linear growth function with an increase in number of NEW cases per day is an irrelevant red-herring to the core discussion, especially since there isn't enough statistical power with only 5 recent days in question to test for a difference in goodness-of-fit between specific non-linear models. Plus, Quadratic functions have more parameters and thus are less parsimonious.
I am not arguing that the correct model is quadratic, only that that the growth curve looks that way when dailies are increasing linearly.
The thing is that models predict that the growth increases exponentially at first but then slows down. The fact that we are no longer increasing exponentially indicates that we have entered the slowdown phase. That's why exponential vs. non-exponential is significant. Linear vs. non-linear does not have the same significance.
Linear vs. non-linear has massive significance. Again, your claim of linear predicted 1/4 the deaths and 1/3 the cases of what the actual trajectory of the data predicted up through April 4th. Linear is very simple and the data from March 1 to April 4 showed a definitively non-linear pattern, which is any pattern where the growth rate is not constant.
Almost all other countries have had essentially linear or declining growth for the past 3 weeks, and non ever had the degree of sustained non-linear growth of the US.
Italy or Spain for example have also had a lot more per-capita cases and a lot more deaths. It is not surprising they are further along the progression curve than us.
They never had more than a week of non-linear growth. Most of their population is still uninfected, so their containment into a linear function is not b/c they hit some wall of no one left to infect b/c of a smaller population. It's because they employed measures to contain the spread.
Look at the graph below. It controls for when the disease took hold in each country by using "Day 1" as the day when 100 diagnosed cases were reached in that country.
But it doesn't control for the population size.
Each country's first 2 weeks look similar, then the US begins to separate and by day 18 the US slope takes off and is far steeper than all other countries for the past 3 weeks, including Italy and Spain. The steepness of those slopes correspond to the degree of non-linear growth.
No it doesn't. The steepness is the result of US having a much bigger population than Spain or Italy or Germany.
As you can see, all the other countries had only a about one early week of mild non-linear growth then flattened into linear growth (same # of new cases per day), while the US has had extreme non-linear growth for 3 weeks.
Again, the lack of normalization for population paints a misleading picture.
And your excuse about population size is irrelevant, b/c it has no direct impact on growth rate.
It's no excuse and of course it doesn't. But it does have an impact on the graph you posted. Not normalizing for population exaggerates the slope for one.
It only directly impacts how many total cases there could eventually be, which is the maximum height the line eventually reaches, but not how quickly it gets there.
Never claimed otherwise. How quickly it gets there is impacted by distancing efforts for example.
So, here you directly contradict yourself. You repeatedly disregard the data base on an argument that pop size needs to be controlled, when in fact it has no relevance unless you assume that pop size impacts growth rate, which you just claimed you don't assume.
If you take a line of 10,000 dominoes vs a line of 100,000 and keep other variables constant, then you push both starting dominoes over, after 5 seconds the number of dominoes knocked over will be the same. The longer line (bigger pop) simply has more possible dominoes that could be knocked over eventually if the process isn't stopped.
True for dominoes, but is not applicable here.
Yes it is relevant, which is why you just blindly dismiss it without argument. It shows that pop size does not impact growth rate, thus it is invalid to use per-capita stats when talking about growth rate. You have no logic as to why pop size should be used, it just manipulates the data to make it look better for the US.
