The article says that I like to find ideas by “playtesting” then goes on to describe playtesting as a form of user testing.
Whereas there is some definition of the word playtesting that might be applicable to the way I design puzzles, it is not remotely the one in the article. In fact I think too much “playtesting” as described here will make your game boring, and will degrade your skill as a designer, and over time make you a boring designer. It is important to avoid this.
If you playtest too much it means you don’t really know what you are making or don’t have confidence to determine what playing it can be like. Even if you don’t have these skills or this confidence yet, you should be working on building them. Certain high skill levels of design will not be accessible otherwise.
The amount of playtesting we did on The Witness was very small — about 5 or 6 days over the course of 6.5 years, on a game that takes 50-100 hours to play through.
Whereas I do sometimes get puzzle ideas from bottom-up exploration of a space by myself (which is the definition of “playtesting” that would make sense here), in fact the vast majority of my puzzle ideas just spring to mind after I have built a sufficient understanding of the possibility space. So these are more like two distinct phases of design, and you might switch between them when designing one game. But I think most of the actual puzzle ideas have to come at least semi-intentionally, or the game will not feel very strong.
You have to know where you are trying to drive the truck.