The title doesn't make it at all obvious what or who this is about. It appears to be a followup to <a href="https://randomcriticalanalysis.com/why-conventional-wisdom-on-health-care-is-wrong-a-primer/" rel="nofollow">https://randomcriticalanalysis.com/why-conventional-wisdom-o...</a> , recently posted at <a href="https://news.ycombinator.com/item?id=41870858">https://news.ycombinator.com/item?id=41870858</a> (and also several other times since the original publication, but this is the first time it got traction on HN).<p>The thing that isn't clear to me in the author's model is why it should make sense to see this kind of power-law result (it obviously cannot extend indefinitely, since at some level of income, health care spending would be predicted to exceed 100% of income). This should be at least as much of a problem as the "poor countries are predicted to have negative health care expenditure" problem that the author points out with linear models.<p>(If we're concerned with out-of-sample data in judging the soundness of a model, then presumably we should also be concerned with these sorts of hypotheticals. The author does seem to note the concern, but IMO dismisses it too readily.)<p>A more intuitive model, IMO, would be a fixed percentage of income <i>in excess of some threshold</i> - but this is also problematic in that our "threshold" for covering basic needs is apparently highly variable (given how poor some countries are compared to the US).