Maybe but "Imagine not having" sure sounds like you remove the deaths entirely.
> you're cherry-picking the years
That's a quote of an article that only compares those years. I'll thank you not to accuse me of cherry-picking here.
> It's not simple, there's a lot of complexity and path dependence to how we've ended up with these car designs and there's no reason to believe it would happen in the same way and the same order for carriages. But regardless, you have to average over such variations rather than picking one unrepresentative point.
You wanted to consider "modern safety regulations" as a confounding factor.
I see three options here.
1. Let's assume carriages would have undergone the same evolution of safety. That means modern carriages would have a lower death rate than old ones. It also makes the possible cherry-picking matter a lot less because the increasing menace of transportation over the last 20 years would be the same either way.
2. We say it's too complex and ignore those factors.
3. We invent an entire alternate history in excruciating detail.
Option 2 is bad, of course. I think option 1 is fine. If you want option 3, you're on your own.
That's like saying it's fine to compare the cost of houses in New York and San Francisco by comparing the cost of a house in New York in 1998 to the cost of a house in San Francisco in 2009, because house prices change at more or less the same rate everywhere in the country. It's true that house prices tend to go up or down at a similar rate everywhere, but it's still a problem if you cherry-pick the two years you're making a comparison with!
https://www.accessmagazine.org/spring-2007/horse-power-horse...
You can criticize their choice of sources, you can dig deeper and provide more comprehensive data, but going post after post admonishing them for cherry-picking data that is just gauche. They didn't pick the years!
But since they are just presenting you with the most commonly available data, nothing you say will make them produce different results.
Not necessarily, they could just mean many fewer deaths.
> Well I'll assume if we were using carriages en masse they'd also be going through design changes to kill more people since that seems simplest
It's not simple, there's a lot of complexity and path dependence to how we've ended up with these car designs and there's no reason to believe it would happen in the same way and the same order for carriages. But regardless, you have to average over such variations rather than picking one unrepresentative point.
> and still supports my point that there would be many deaths.
No it doesn't; you're cherry-picking the years when motor traffic was least deadly and comparing to a random time (or perhaps even the most deadly time?) for horse traffic.