In the paper, they do say that the 4x4s showed more cardiac adaptations. To me that says to train all the intensities. Do some 4x4s and some 4x8s or 4x9:30s. 
They could have lit the training world on fire if they had a group that did both 4x4s and 4x8s. Or maybe 4x4s one day and 4x16s building out TTE on another day would have been the best overall?
The 4x8s may be the best bang for the buck if you are just going to pick just one interval.
In that discussion they also talk about fatigue which is interesting:
Based on the group difference in RPE reported throughout the study by the 4 4 min group, we can speculate that they experienced greater residual fatigue from the training intervention. If this is true, we might also speculate that they would have shown a greater improvement after training had they rested longer than the other two interval groups.
With a better taper, it’s possible that the 4x4s might have come out on top. The other interesting point that jumped out at me was this:
Independent of the training group, subjects who reported no weekly interval training (n 5 11) in the 2 months before study start tended to achieve greater average improvement in VO2peak, PowerVO2peak, and Power 4 mM
In the group of subjects, regardless of the interval group they were in, the subjects that hadn’t done intervals for 2 months had the greatest improvements. You read that and think ‘no duh’ as it sounds totally logical. Maybe the best intervals do do are always the ones you haven’t been doing?
It’s a great paper and a great discussion IMO but it’s not a nail in the coffin for other intervals. The 4x8s are the “best” by some huge margin. They all yielded gains.