A/B Test: Responsive Email VS Non-Responsive Email

Recently we read on the Action Rocket blog a post about A/B testing of Responsive emails. The headline is "Split testing responsive email design: You're doing it wrong"and I was saying that a test of a responsive email VS a non-responsive one doesn't make much sense.
The main thesis of the article is that the main objective of responsive email design is to improve the user experience and therefore measuring the results numerically is a chimera. Eliot Ross, author of the article, maintains that the differences in openings, clicks or even conversions between the two versions would hardly provide reliable data.
However, the goal of responsive design, a better user experience, must be reflected in measurable campaign results.. After all, the goal of optimising the user experience is to make our marketing efforts more effective.

When a user receives an email that is not adapted to their mobile device, their behaviour should be different to if they receive it and the email is rendered allowing a comfortable interaction (large texts and buttons, no need to enlarge or scroll sideways...) Therefore, how can we not get useful data from the analytics of this type of A/B tests?
We can intuit that the interaction of our audience with a responsive email will be better than if the email is designed only for one type of device (desktop/mobile), but the results of a test may indicate that the effort to design and layout responsive emails does not bring significantly better results than designing emails exclusively oriented to desktop or mobile.
In any case, from Digital Response, given the circumstances and the data of apertura on mobile devices that we handle, we recommend in most cases the Responsive Email Design, however, we must not forget that it involves a extra effort with regard to the non-responsive layout, and that we often have to sacrificing the display in the Gmail App. This test (perhaps carried out several times) may show us that our responsive email shows better interaction results, but it may also tell us that there is no significant improvement and therefore, we can save the effort (and Gmail's displeasure) in case we are short of time to send a message, for example. Yes, there are many factors that can influence user behaviour, but a Split-Test, or better, a testing routine, will always give us information about the behaviour of our users, allow us to get to know them better and adapt our efforts to optimise and monetise our email marketing actions.
We also recommend, before this split test, to know the devices from which the users are opening the email. You can obtain this information from your sending platform or from analytical tools such as Litmus.
What do you think? Have you done this type of test, have you obtained interesting results, or, like Eliot Ross, do you think it is not useful?