It's a prisoner's dilemma. If you (researcher) defect (publish in Elsevier journals), you perpetuate Elsevier's business, a shitty situation. But it's not as bad as cooperating (not publishing), because there's a chance that everyone else defects and you screw up your career, a really shitty situation. So out of self-interest everyone defects and perpetuates Elsevier's business.
Framing this essay as apologia for Haskell's laziness is a weird because Haskell was created exactly because there was a proliferation of non-strict functional languages and the Haskell committee recognized that having a common language would be beneficial. So it seems disingenous to separate Haskell and laziness: the latter partly defines the former.
Haskell's evaluation strategy is not something you can just change and have the rest of the language stay the same. If Haskell were strict there would be a good chance that it wouldn't be pure (see: ML variants); if it wasn't pure then IO would not be a problem; if IO wasn't a problem then Phil Wadler wouldn't have needed to invent typeclasses, etc.
