The article is not very good at expressing its own point, unfortunately. This wasn't it.
Take periodic task A, such as making dinner or sitting down to work. You do this task on a mostly regular basis. Doing this task creates waste: a dirty sink, an unclean workspace, and so on. When you complete the task, it often seems like yet another task (B) to accomplish when tired to do the cleaning. Thus your workspace starts to degrade. The next time you need to do task A, you have to do task B first. You're psyched and ready to do task A, but you can't, because task B wasn't done. And task B is not what you're psyched for.
All the article is pointing out is that you need to do task B as soon as you're done with task A. That way, when you come back the next day to go for task A, you don't have a preliminary step to go through before you can really get started.
> Although how one could test the idea proposed in the blog is unclear to me.
Take any such pair of tasks and measure the time it takes to get started on the task pair over the course of a decent period of time, such as a month. Try to have a balance of which task is performed first. By "time it takes to get started", I mean the the delta between the time you say, "I should do task A", and the time you actually start doing either task.
Of course, that's subject to the observer effect: noting down the time at which you say you should do it will probably in itself incentivize you. Thus, it may be helpful to regularly schedule a task pair and note the offset from the scheduled time.