In principle, I don't see anything wrong with the idea. However, they aren't testing this in a very useful way. To really optimize Soylent, they need many people to use it, and they need base line blood work and vitals, and then track these people over time. One approach would be to take 20 volunteers, and maybe develop a 3mo development cycle. Every 3mo 10 of these people would use Soylent and they would be tracked. Anything changes they decide to make could then be tried on the other 10 for the next 3mos, and the original 10 would take 3 months off. It's not a great design, but it would be better than what they are doing.

The problem is that longterm issues are only going to crop up over time. Our nutritional knowledge is so incomplete, that there are going to have to be changes made.

Also, there is the possibility that humans benefit from not only complete nutrition, but nutritional variation. For that, they might consider different variations of Soylent or some sort of cyclical Soylent diet.

All that said, I bet replacing two meals with Soylent would probably benefit most Americans.