When people say “playtest” they actually mean something you might not mean and I might not mean, so I’m going to break it down a bit.
Playtesting mechanisms and system (a bunch of mechanisms that interact) is something I do in a closed setting: people I know and trust, guided by me. It’s not something I would ever do blind, by sending it out to other people to test in their own environment. This is because mechanical testing like this is still highly fluid: I may well want to change things up as we play. It’s a highly experimental and mushy process.
Because I do this within a scaffold, it’s really more like playstorming: we’re playing and designing simultaneously, and this is why you need a certain amount of trust in your compatriots. Revealing your half-formed ideas leaves you more than a little vulnerable. Killing your idea even when it obviously sucks is a difficult process. Admitting your baby is ugly is something you should only do with friends.
This is distinct from the blind playtesting that happens at the end. At this stage I’m not testing the mechanism at all: I don’t really care if someone doesn’t like a mechanism because I’ve already established that I and the gang of folks I test with do in fact like it. That’s all I really care about. So at this stage we are not testing the concept of the game–the ideal that I think or want the game to be. We are testing the text. Does the text deliver the game that I had fun playtesting?
So that’s the distinction. Mechanical testing is fluid, piecemeal, and requires trust and engagement. It’s best done with people you know, people whose opinions you value, and people you trust. And of course, people who are cool with dicking around with a new thing every week.
Textual playtesting is a process of validating the text. This is best done with strangers who’ve never seen the text attempting to turn it into play and comparing that play with the ideal of play that I’ve assembled previously.
I feel like this is best done completely separately or at most iteratively. But never both at the same time. People who are too close to the ideal of the game are poorly equipped to evaluate the text: they already have bits in their head and may well not notice their absence in the text. Similarly, strangers provide input on mechanical testing that is frequently not helpful or tangential: “I don’t like this”, “What if you did this thing I invented and love”, and so on.
So that’s where I’m coming with when I say “playtesting”. It’s not canon. It just works for me.