Creating Better Tutorials
Tutorials could be so much better, and Celeste is a shining example of how
How do you learn something new? It’s a fascinating and poorly understood topic. In schools, we cram kids’ heads full of factoids, and they generally retain some of it. But they, by and large, don’t necessarily enjoy this style of learning. It only “works” because we force them to stay there.
If we apply that approach to a game—walls of text explaining everything the user might need to know—the user might simply stop playing, turned off by the boredom.
Though it’s possible to push through a bad tutorial for a good game, it’s never a pleasant experience. I’ve come to expect mediocrity from tutorials, and I simply hope that it might get better as I learn the game more.
This is why Celeste completely blew me away after my first playthrough.
Celeste doesn’t really have a tutorial in the usual sense. Instead, you’re thrown into the world, with little explanation — just a few words of text:
With that, we’re in the world — and what a world it is. Soft, melodic piano music plays in the background, telling us that this is a safe haven.
In these initial moments of the game, there are no tooltips telling the player what the control scheme is. Celeste essentially forces the player to figure it out as they go along. As you can see in the above and below screenshots, the first real interaction the player has with the physics of the game is totally risk-free.
Next, there’s a small risk: a hole the player must jump over. They do, with ease, but it forces them to discover another mechanic.
If they tap the jump button and then release, they won’t make it over the pit. Only if they hold the jump button momentarily will they get enough airtime to make it over.
On the next platform, something new happens. Here, we see an ice block, disconnected from the main ceiling area.
With only a second of warning (and the same calm music), it falls.
Celeste has introduced us to danger. Evading this danger is trivial, as the user has already learned to run and jump. Not only this, but it places an element of time pressure on the control as well.
This falling block also serves to show us yet another new mechanic: clear and distinct lines between objects signify clear separation. This will come into play later when distinct chunks of walls are breakable.
In the first screen, the first thirty seconds of gameplay, the player gets a real feel for what it’s like to play Celeste. They have been tested, by the falling ice, on their ability to smoothly and quickly combine these controls. They understand the physics, they know how to run and jump, and they are ready to move on.
The second screen introduces climbing in much the same way.
After a brief visit with the Old Woman, the third screen gives us another lesson in time pressure with a collapsing bridge.
Finally, as it looks like you’re doomed to fall into the abyss, you are given your final power—the dash—and saved.
Three screens. Three mechanics. Each one has a very specific role in teaching the player. But the player hasn’t even noticed any of this. They are learning in the way that (I believe) humans learn best: by doing.
Learning by doing is the secret of Celeste.
The entire game is a tutorial because, on every screen, you are learning. Sometimes you can feel the learning. Sometimes you’re just breezing through. But at just about every turn, there is a new learning opportunity, and this opportunity gives you a chance at a breakthrough.
Anytime a new mechanic is introduced, or even a new type of item, there is a brief moment of respite where the player is given a chance to discover the properties of each one. Players are taught something very specific about the item, building from basics, and soon, the player has mastered it.
Despite all this subtle teaching, do you know how many actual instructions are given?
Two.
One, telling you how to climb.
Two, telling you how to dash.
Celeste never tells you what to do. Instead, it creates an environment where you need to learn how to do what’s needed to progress.
If you don’t, no pressure—you’ll die, respawn, and you’ll learn something then, too, which is that you can’t do it that way—so it’s time to try another way!
This cycle will repeat until you have mastered a concept, at which point you are naturally rewarded by being allowed to progress.
Just imagine if instead of the aforementioned process, Celeste tried to give a tutorial the way most games do it, and the first screen you saw looked like this:
It would give you tooltips at every turn telling you how to handle the platforming, but you’d still be bewildered, and you wouldn’t have that smooth, natural experience that is core to Celeste.
In contrast, let’s look at a different tutorial, to one of the most complex and massive strategy games ever created: Civilization VI.
Civilization is a long-running franchise about building cities, armies, and eventually, taking over the world. In Civilization VI, the player manages dozens of cities and, at times, nearly 100 units in a turn-by-turn system whose intricacies require many dozens of hours to learn.
The Civilization tutorial starts in a completely regular instance of a 1v1 game against an AI, whereas Celeste starts off with the basics, which is, I think, a much better approach.
At every turn, the Civilization tutorial tells you exactly what button to click. There is no choice. No agency. And because of this, my “learning” experience left me in a situation where although I performed the actions in creating a moderately successful tutorial city, I couldn’t possibly remember how to replicate such actions in the future.
In fact, I would later need to consult several hours’ worth of YouTube videos in order to learn the game enough to wipe my friends off the map — and isn’t that the whole point of Civilization, really?
Compare that to Celeste, where I would never have dreamed of needing to consult YouTube to figure out how to perform any of the movement, basic or advanced, in the game.
To be fair, Civilization is one of the most complicated modern games, while Celeste is one of the most simple. Some might write this tutorial off as a necessary mediocrity.
But I don’t think that means Civilization’s tutorial is doomed to be bad.
If Civilization’s tutorial were designed like Celeste’s, it would start focusing on one, small mechanic, let’s say, a single unit.
It would give the player the opportunity to explore with that unit, finding out some of the important rules of the game — mountains are impassable, Tribal Villages are good, and how combat works — without having to force them through it with tooltips.
Once that stage had been completed, the mode of the game could shift, and the player could be placed in control of a single city, forced to defend its borders from encroaching Barbarians.
If the player failed to make their city efficient enough to succeed, they would fail and restart, having learned a valuable lesson about how not to design their cities, which, in Civilization, is just as important as knowing how to do it right.
Slowly introducing new concepts could vastly cut down on the amount of text, and allowing the user agency would make the tutorial far more engaging. Allowing failure gives the user a deeper understanding of the proper technique, and short segments would ensure that if (when) the user failed, they wouldn’t become angry and give up.
All of these are design concepts pulled straight from Celeste, and, once they’ve been generalized in this fashion, you can see that they would work beautifully for other games, including Civilization, as well.
More games could, and certainly should, take a page out of Celeste’s book and give players a chance to learn by doing rather being force-fed mechanics by the spoonful. Even the most complicated games (hello, Civilization) out there can achieve this by introducing things one concept at a time and allowing progression once that concept has been mastered.
For this to be possible, we need to allow failure. Not only will this cut down on those cookie-cutter opening sequences with boring, intimidating walls of text that are impossible to fail, but it will allow players to build muscle memory from the very start.
Tutorials, too often, suck right now, but they can be better if we follow the example of Celeste, and let players learn by simply doing.
