Playtesting night in SLC

You're describing the difference between a gameplay playtest and a rulebook playtest (sometimes called a "blind" playtest). They serve related, but slightly different functions, and are merely different tools depending on the sort of result a designer is looking for.

A gameplay playtest usually takes place earlier in the process. The designer is interested to see if the ideas work; if the game has legs, so to speak, but is past the point where one person (himself) can fully evaluate the game and how it plays. He needs fresh eyes to see the game and wants to examine its emergent gameplay. Questions include: Is the game balanced? Does it "break" -- and if so, where? Where is the "fun"? What didn't work for the players? Since the focus here is on the gameplay, balance, and the like, it would defeat the purpose of the test to have players misinterpret the rules due to poor writing and then play the "wrong" game; likewise it's not helpful for them to spend time arguing about the meaning of rule [x]. In such a situation, when players are confused the designer WANTS them to ask him, so they can get on to testing the gameplay as he envisions it.

Obviously, however, at some point you won't be able to explain your game to every potential player that buys it, or every company to whom you send a prototype. To prepare for that, you do "rulebook" (also called "blind") playtesting, where you simply hand the rulebook to your players and watch them have at it without interfering. Ideally they have never seen this game's earlier versions, although that's not always possible.

You can still evaluate some of the design questions above at this stage, but you're now also looking for an evaluation as to the clarity of the rules. It's as much a test of your rule-writing skill as it is your design. If the players consistently miss something that you thought was obvious, or misread a rule that you thought was clear -- well, you may want to re-word something. If they say "this book is too long" or "this book didn't cover scenario [x]" then you'll want to consider implementing those changes as well. If they seem to be having a good time, then you're on your way.

Blind testing is great, and absolutely every game you send out (either yourself or to a publisher for evaluation) should go through that process, ideally several times with different groups. But it's not the only tool you as a designer have, and it's not necessarily the best one for a given iteration of your design phase.