The WaPo chart is for editions, not titles. If you list by decade for titles -- actual original works of creativity -- the 2000s utterly demolish each prior decade within this sample, and the 1990s surpass most, by the study author's own estimate (rather than manually identify duplicate titles, the author seems to havbe looked at some subsample and divided the whole public domain batch by 4).
See Figure 2 in the study.
It makes sense that public domain books have way more editions than titles; it is a market of duplication and redundancy, where you differentiate yourself using packaging and SEO type tactics. A spammy market, in other words. Printing a public domain book mainly involves some scanning, layout work using widely available software, and sending it off to the printer ("These tasks can be performed in less than a day," says the study). Oh, except you don't even need to scan anything for one of the many works already online, and you don't need to print anything if you use a print-on-demand service. So a lot of the "editions" in the WaPo chart are simply crap duplication garbage angling for a few pennies profit.
Now you can still make a case against lengthy copyright with this data, but it's trickier (because it's more honest). You can point out the population has grown tremendously and also as the study states: "In a world without copyright, one would expect a fairly smoothly downward sloping curve from the decade 2000-2010 to the decade of 1800-1810... Instead, the curve declines sharply and quickly, and then rebounds significantly for books currently in the public domain initially published before 1923."
The WaPo includes a paragraph summarizing this but the chart used is not the more honest Figure 2 but the misleading Figure 1.
Another issue? All the book figures in the study are based on an undisclosed set of subcategories of Amazon's "Literature and Fiction" category, a total of 7000 or so results later whittled down to 2317 based on a cross check with library of congress. What happens when all categories, the full range of textual creative expression, are included? We don't know.
If you want to make a convincing case that copyright terms are too long, use convincing data. "Look, public domain means more heavily duplicated garbage editions on Amazon!" is not particularly convincing.