Every game is a set of rules consisting of goals, the tools for achieving them, and risks. If players do not know the rules, they cannot play. This is fairly obvious, so there is no need to spell it out at the level of mental modeling. Every game faces the problem of teaching the player, but each solves it in its own way. Let’s look at a few approaches.
I was prompted to write this post when Masha decided to play The Witcher 2. Half an hour after launching it, she told me the game was impossible to play and she could not see where the promised great story was. I was stunned: when I played it, I thought its opening was one of the best I had ever seen in a game. Looking into it, I discovered that after I had finished the game, the developers released an update that added a tutorial.
This is what the game looks like with the tutorial:

Watch the video on its original platform


Distinctive features:
* No plot. There is only a small story clearly created to serve the tutorial.
* Constant pop-up tips interrupt the gameplay and dump walls of text on the player.
* A high density of information per unit of time.
* Once a mechanic has been learned, the game moves on to the next one. There is no opportunity to use the mechanic just learned.
* It teaches concepts the player does not need early in the game.
* You cannot progress until you do what the game demands, even if a bug prevents you from doing it.
Now let’s look at the game in its original form. The video is long, but you do not need to watch all of it; skipping to a few points is enough.

Watch the video on its original platform


Distinctive features:
* Story hooks appear immediately: who is Geralt running from and why, why is he in a dungeon, and who is interrogating him? There is more to come: the castle siege makes it clear that this is not a small-scale story about a wounded knight.
* Interactivity increases gradually. At first the player simply watches a cutscene. Then they select dialogue options. Next they can walk around the camp. Only after that are they given the opportunity to fight.
* No pop-up windows. Every action is intuitive, and there is no opportunity to do something else by mistake instead.
* Once the player has learned a mechanic, they start playing with it. Learned to choose dialogue options? Talk to the characters. Learned to walk? Go and explore the camp. Can you do both? Here is your first quest about some fighters and a mysterious amulet.
* All the mechanics are optional. Do not want to talk? Then do not. Cannot kill an enemy one way? Try another. In the very first battle you can simply stand still and watch friendly soldiers kill all the enemies.
* Information is spread out over time. In the first half hour I did not brew potions, change equipment, or kill monsters with spells and bombs — the game will tell me about all that much later. In other words, as I play, I will keep gaining new possibilities.
Frankly, the choice of which version offers a better introduction is so obvious to me that I do not understand how the creators of such a wonderful game made such a serious mistake. Perhaps I am missing something, and I would welcome an explanation of this decision.
Another example is this analysis of how Mega Man X teaches its player. Just watch the video and you will understand.

Watch the video on its original platform


Finally, here are the slides from a GDC 2014 talk in which the creators of Splinter Cell: Blacklist explain how they built their game’s tutorial. Of course, it is better to watch the talk itself, but that requires a paid subscription. The slides nevertheless illustrate everything quite clearly. They are not just a set of abstract images, as presentations sometimes are.
http://gdcvault.com/play/1020212/Super-Spy-Crash-Course-Teaching


P.S. The header image is a
randomly generated comic from the creators of Cyanide & Happiness.