Your article puts the high end of North America north of Mexico estimates at 18 million. With its own estimate at, indeed, 4 million.
1983 is not what I would consider a "modern" estimate. Snow (2001) puts it at 3.4M, Alchon (2003) at 3.5M, and one of the highest recent estimates was Thronton (2007) at 7M. So like I said, current consensus is in fact below 10M, and most put it below 5M.
Furthermore the article goes out of its way to explicitly state why that 18 million estimate is impossible:
"Dobyns's (1983) 18 million, couples optimistic estimates of environmental carrying capacity with unrealistic assumptions about the effectiveness of food-acquisition strategies ... The extrapolation has received its share of criticism, and Dobyns's (1983) assessment of the Timucua far exceeds figures based on more sober appraisals of historical accounts and archaeological evidence. High population estimates are consistent with maps that associate particular groups with irregular areas that collectively cover the land in its entirety, as if nothing was left unoccupied.
Maps showing contiguous occupation, however, have the effect of implying that all equally productive land, most importantly resource-rich shorelines and river valleys, was similarly and continuously occupied. A much different picture has emerged from archaeological work over the past several decades. Not only were there large and infrequently used areas between late prehistoric population aggregates, the vacant areas often encompassed highly productive land suitable for permanent settlement."