Human can learn about anything. It does not matter how badly designed, you can always find some individuals that will figure out how to make it work. As far as game design, this means that if you have a captive pool of people with which you interact about the game, at some point most of them will figure out how the game works, and that will tell you absolutely nothing about how good the design is.
What you need is to evaluate how quickly a fresh group of people will get the design. That’s what I call Kleenex testing, because you use those people only once. After that first use, they know too much about the product they can not be user again, they are dirty.
I strongly suggest to put them in front of the product with sometimes a quick explanation, and then to watch without getting involved at all until complete failure. Then you can restart them a couple of times and explain to them while they failed. In reality you have just learned where your design failed.
I have used this many times, and it has proven to have an extra value, that of convincing the team members faster about issues in the game that long conversations where everyone tries to rationalize everything to death. I highly recommend it.
On “The Sims” during some periods of the development, we have been doing it one evening every two weeks, to give us time to fix the code and test internally before bringing a fresh group.