Great question kismet, and nice to see you in the nootropics section
Would be interested in hearing your speculation about this when you've had time to consider it. I still haven't looked that deeply into the working memory/Gf and psychometric testing literature, so I can only offer some vague impressions from my skimming.
The minimum training time for participants was 10 minutes a day, 3x a week, divided across 6 different types of training for each of the 2 study groups. This is obviously a very different approach from the DnB studies, which do something like 20min x5 days on a single prolonged task targeting a single domain (WM).
The problem of how cognition was measured is always central to this kind of investigation. It seems the tests were pretty short, and references are only given for a few of the 12. This bears looking into, as more accurate tests of Gf (like Raven's) take like 40 minutes for a reason.
The supplemental figures from this study show enhanced performance on the training tasks with more sessions, but no transfer to general cognitive measures. If DnB does indeed result in far transfer from working memory to Gf, it seems to be a fairly unique property.
The design of this study seems to have tested whether exercising your brain in general increases intelligence in general. They may have been casting their net a little too wide to come up with anything useful. Their failure to include DnB, the most well-studied brain training game (and AFAIK the only one shown to transfer), suggests to me that this was intended more as a 'Mythbuster' endeavor (indeed, the study was commissioned by the BBC show
Bang Goes the Theory) than a rigorous test of what is possible. If anything, it may show that simply 'exerecising' the brain with things like Lumosity may not be as useful as DnB.
A full PDF of this study is available
here, and the supplemental figures on the
abstract page. The tests used for this study can be taken online at
Cambridge Brain Sciences (requires free signup).