I think this way scales better - it's easier to visualize than the massive gap would be. And where we are in the news cycle is to focus on problems common to all classes, not the problems that divide classes.

Perhaps the visualization is just better this way, but your question is at least an interesting one to think about. How would the shape of the curve change, and might it suggest slightly different conclusions?

For example, it seems to me the absolute income version would be steeper at low incomes, perhaps amplifying the perceived impact of things like a living wage or basic income on educational outcomes. (I'm talking mainly about the psychological impact of the curve's shape, not raw statistical truths.)

I completely agree. Percentiles are a very unintuitive way of thinking for this sort of data. It's really easy to know how much one made last year, but it takes a while to figure out what percentile one's income was in. Differences in Percentile vs differences in Income is even harder.

I felt strongly about this the moment I read the article on NYTimes - I reproduced the data set to the best extent I could and actually went about plotting against absolute income:

https://docs.google.com/spreadsheets/d/1jD87QN3pPZjoefrAm4E_...

That's a far more intuitive chart, that a lot of us will understand better. I also graphed how much these children actually go on to earn against absolute income. I compared these results in a blog post - http://unside.t4you.in/data/intuitive-axes/

HTH.