I've been known to say about game design that "Prototypes are sovereign", that you haven't really designed a game until you have a playable prototype. That's because, until the game is played, you just cannot really know what you've got. But I would be just as right to say "playtesting is sovereign".
When you design a game, you try to see in your "mind's eye" how the game is going to work, but until you play it, you simply cannot know what is going to work and what is not. The first few times you play, many things will change (provided, of course, that you're willing to make changes, which is a major requirement of a game designer).
Granted more experienced designers can foresee weaknesses and eliminate them before reaching the prototype stage. But we're interested here in teaching game design, so this is addressed to inexperienced designers.
Let's clarify something right now. I am talking about playtesting to improve gameplay, not testing to squash programming bugs. The latter is what is often meant by "testing" when people talk about electronic games, and this testing takes place late in the development cycle, when the gameplay and appearance are set in stone (because it's too late to make major changes). This bug testing ("Quality Assurance") is aimed at making sure the game works the way it is supposed to, not at whether the way it's supposed to work is good or not. "Bug testing" essentially does not exist in non-electronic games, although it is important (and often forgotten) to test the production version of a game, as converting the prototype into the published version can introduce its own set of problems. (For example, the boxes on Population Track on the FFG Britannia board are really too small for the purpose; this new version of the board evidently was not actually tested.)
So: here I'm talking about playtesting the gameplay and assorted details (such as user interface) that strongly affect gameplay.
There are three stages to playtesting: solo playtesting (also called "alpha"), local playtesting ("beta"), and "blind" playtesting (also part of the "beta" stage). (In electronic games, often the in-house testing is all called "alpha", and outside testing is called "beta".)
Few non-video games are meant to be played alone. Yet in solo playtesting, the designer plays the game solitaire, playing all the sides independently as best he can. At this stage the designer is trying to get the game to a state where other playtesters have a good likelihood of enjoying it, and of playing it through to the end. At solo stage the designer might try a portion of the game and then stop because something isn't working, or because he has a better idea. When asking other people to play a game I would never stop a game in the middle, or try something that might be so bad I'd want to stop, though I know of designers who think nothing of doing this.
Most video games can be played alone, and if there's a more-than-one-player component, it's usually impossible for the designer to play several sides by himself.
At the local playtesting stage, people are asked to play the game through, usually in the presence of the designer when it is a non-electronic game. Almost always, at the beginning of this beta testing I do not have a full set of rules, I just have notes about how to play, and some of the details are in my head. (This is a big reason why it is much quicker to design a non-electronic game. With an electronic game all the "rules" must be settled precisely before the programming of the prototype can be completed. The programming is the equivalent of the rules of the non-electronic game.) As local playtesting goes on, I make a rough set of rules, then finally write a full set of rules.
As the local playtests occur, I write down notes about what I see and hear, and especially about answers to questions that need to be incorporated into the full rules. By the time I have a full set of rules, I usually refer to the rules for detailed questions, to see if the rules cover that question and whether it is easy to find that information.
The third stage is "blind" testing, where someone is given the game and must play it without any intervention from the designer. This is a test of the rules, somewhat akin to "bug testing". Are the rules clear enough that people can play the game from the rules? What questions do the blind testers come up with, and how can the rules be improved as a result? Unfortunately, nowadays people are often poor rules-readers, so I advocate electronic tutorials to help people learn how to play a game.
I know from experience with published games, especially Britannia, that there will ALWAYS be people who misread rules, sometimes willfully. 99% clarity of detail is about the best you can get using the English language.
In a sense, electronic games can jump to "blind" testing quickly, because by their very nature these games hide the rules from the players, enforcing them through the programming. This is an advantage of electronic games over non-electronic, that no one needs to read and understand a set of rules.
Game design, when taken to completion, is highly interactive. Playtesting sets good games apart from bad, and playtesting is (or should be) interactive. In a separate post I list some of the things you must look for while doing beta testing.
There is no doubt that the last 20% of refinement of a game takes 80% of the designer's time. Playtesting is time-consuming, tweaking rules is time-consuming. In the non-electronic world, often a "developer", another person, does much of this testing and tweaking. I personally strongly prefer to do this myself, even though it is much less fun than creating new games, because I don't want someone else "screwing up" my game. (See http://www.pulsipher.net/gamedesign/developers.htm for some of my experiences.)
Even when you don't intend to change the rules, rewriting them introduces unintended consequences (as evidenced by the Britannia Second Edition rules rewrite by FFG--and apparently having no testing of the new version of the rules compounded the problem). When you rewrite to change a rule, the repercussions are often larger. So a remarkable amount of testing is needed.
In the electronic world it is difficult to quickly and cheaply make big changes in a prototype. This is one of the problems that all makers of electronic games face, and a major reason why some electronic games are not very good. By the time the development studio has a playable prototype, it is too late in the schedule to make the changes that playtesting reveals are necessary.
At some point during playtesting of a game, the designer must decide if "there's something in it" (as I put it): if the game is really good enough that people might play it, like it, and would buy the finished version of it. There's really two times when this should happen, once during solo playtests (alpha testing), the second time during playtesting by others (beta testing). The "something in it" point in solo playtesting is an indicator that it's about ready for others to play. The "something in it" point in beta testing comes when observing people playing the game and their reactions during and after playing.
Usually I need to tweak a game quite a bit from its state at the end of solo play, before I can reach the "something in it" stage of beta testing. Sometimes there doesn't seem to be anything in it during beta testing, and I set it aside for further thought. Sometimes I realize, from solo playing, that there isn't "something in it", at least not yet, so I set it aside at that point.
I strongly suspect that novice designers rarely understand these stages. Their egos become involved, and they assume that because they took the time to make the game, and it's their idea, there must be something in it. In extreme cases, the "designer" thinks he has "something in it" when all he has is an idea, that is, when he has virtually nothing at all. The number of people who think they've successfully designed a game, yet haven't playtested it at all, is remarkable. Playtesting is the meat of successful design, not the end. (I confess that I don't think of "development" as a process separate from design.)
So how do you recognize when there's "something in" a game? That's hard to say, unfortunately. Surveys or written feedback won't necessarily reveal it. In alpha testing, the "something in it" stage is a gradual realization, coming from observing my own thought processes as I play. My games are, almost without exception, strategy games. When I "see" myself thinking hard about the strategies, and liking the options, then I may think there's something in it.
In my case, in beta testing when spontaneously (without any urging) people say "I'd buy this game", I know I've got something. However, this is rare, and I don't remember anyone ever saying that about Britannia, or Dragon Rage, or Valley of the Four Winds, but they have all been quite popular. Perhaps better, if people want to play the game again, in this day of the "cult of the new" when hardly anyone plays a game twice in the same session, there may be "something in it".
I am very low-key in beta playtesting, preferring to watch reactions of people rather than try to solicit opinions, in part because people (being polite for the most part) won't say negative things even when asked. I also try not to play, as 1) the designer playing in a game tends to skew results and 2) when I play, I do a worse job of playing, and a worse job of evaluating the playtesting, than if I did either alone. As I'm that strange sort of person who enjoys watching games as much as playing, why play?
I do not "inflict" a game on players until I think it is good enough to be OK to play, that is, I've reached that first "something in it" stage. Evidently some other designers playtest with other people very early: not me. My playtesters play games to have fun, not as on obligation, and most are not hard-core boardgamers, so I do what I can to make sure the game MIGHT be fun before I ask them to play.
As I said, playtesters tend to be polite. It's hard to find out what they really think. I am skeptical that a feedback sheet will make a difference. Rather,
I sometimes try the "Six Hats" method (devised by Edward de Bono) when playtesting; specifically I'll ask players successively to put on their black hat (the judge), then the red hat (intuition and emotion) to see how they assess a game, and then the yellow hat (the positive side of assessing an idea) to see what they like about a game. With local playtesters I sometimes ask them to think of ways to make the game better (the green hat). Google "de Bono" or "Six Hats" for more information.
Also see the following article on Gamastutra: http://www.gamasutra.com/features/20050913/sigman_01.shtml
This includes tips on constructing prototypes.
"The worst form of inequality is to try to make unequal things equal." -- Aristotle
"Always do right--this will gratify some and astonish the rest."Mark Twain
"A designer knows he has achieved perfection not when there is nothing left to add, but when there is nothing left to take away." Antoine de Saint-Exup'ery
"Not everything that can be counted counts, and not everything that counts can be counted." Albert Einstein
"Make everything as simple as possible, but not simpler." Albert Einstein