In the last six months, I have changed things up.
I now train with low volume per muscle group (on average, 4-6 sets per muscle group, reps per set between 6 and 12, pushing very hard, close to exercise, not muscle, failure, which means I stop when I cannot complete 1 or 2 additional reps with proper form).
I do full body workouts, averaging 4 workouts a week (I also do jiu jitsu 2 or 3 times a week, and try to run 1-2 times a week). Almost all the exercises I do are machine-based, following the path opened by the late Arthur Jones (if you haven't read his biography, I recommend you at least skim it: it's wild, disorganized and fun).
With this new protocol, I have seen excellent results in terms of strength and muscle growth, with less soreness than I expected. I also increased my protein intake throughout the day to (roughly, I don't use a scale) 1 g per lb of body weight. I find it quite easy to reach this amount by mixing protein powder with Greek yogurt: it is a protein bomb. I have also started taking magnesium glycinate, about 800-1000 mg per day.
As for exercise research, scientists should start, and should have started 30 years ago, adding variance measures to their results. A 3% increase vs 1% when testing competing exercise protocols says nothing. I want to know the variance, because individual variability in response to any stimulus or insult is typically high.