Completely agree. In hindsight the choice of ES5 is questionable and in the study I actually conclude that another ES6-based experiment is desirable. Also, as o

Completely agree. In hindsight the choice of ES5 is questionable and in the study I actually conclude that another ES6-based experiment is desirable. Also, as others here have noted, you could start with ES6, see if it works for enough of your users, and only if not orthogonally introduce transpilation to ES5 as a production optimization.