Tutorials

In which I attempt to capitalise on my new-found Twitter followers and dig up an old post I meant to upload ages ago because it feels like it's relevant now.

Most games these days come with a tutorial. Tutorials have been around for ages. In the Good Old Days, they would be almost entirely hidden - the game would simply drip-feed features at you so as not to overwhelm a new player and not go into too much detail about how they all worked. It was largely up to the player to work that bit out. In fact, a lot of the games didn't actually explain anything at all.

That sort of stuff just doesn’t fly on mobile these days.

Have you ever tried something like Game Of War? The tutorial in that is insane! It takes an age to complete, carefully isolating and highlighting which button to press next*. But here's the kicker:

I bet you won't know how to play the game once you reach the end of it.

Strange, huh? So why does it do this?

Well, I'm really not sure, but I do have a theory. Games of this nature are driven almost entirely by analytics. I'm willing to bet that the initial tutorial was designed based on the list of features the player was going to need to play the game (A PvP multiplayer game of this nature also makes it hard to gradually ramp up those features as it means that new players will always be at a mechanical disadvantage and it won't feel fair). Anyway, having a 'soft' tutorial - one where it pops up information telling you what to press - is usually the first stage.

The most common approach to this is to display a dialogue box telling you what to press and why. Some games take a slightly less direct approach and do stuff like embed text in the environment, but the end result is largely the same - read the text to find out how to do the thing you need to do to progress.

Press the green button. The glowing one. With the arrow.

Of course, nobody likes reading text and so it's likely that the average gamer will simply skip or ignore the dialogue without really knowing what it is they have to do next. Then they get stuck, frustrated and quit. Analytics will pick up on the fact that the game hasn't finished going through the tutorial yet and alarm bells will ring. Someone will draw a nice big funnel diagram with an alarming amount of people falling out at the first hurdle.

So the next stage is to accompany the dialogue box with yet more information. Perhaps include an icon that shows you what the button looks like? Or even a nice, big arrow that points at the button itself? But people might still get it wrong and press something else. Analytics will happily show you that people are still failing to get past the tutorial.

At this point you might decide to move to a 'hard' tutorial - one that it's impossible for the player to deviate from - so you disable all other inputs that aren't the one you want the player to press. To further ram home the point being made by the text, icon and big, pointy arrow, you might even fade the rest of the screen out so that all the player sees is the one thing you want them to press.

Job done. Your analytics will show that 100% of the people will make it through the tutorial successfully**.

Except it really isn't done. Chances are all you've done is show players the quickest way to skip through the tutorial without actually learning anything. All they'll really have picked up on is how to pass the tutorial rather than picking up actual useful information. It's like the teacher that teaches you how to pass the exam itself rather than moulding you into someone who is properly capable of passing the exam with their own knowledge.

See, humans are a stubborn bunch. They don't like being told what to think or do and will actively rebel against that sort of thing.

As my old man used to say, "The best way of teaching something to someone is for them not to realise they're being taught."

Along similar lines, one of the first things Peter ever told me was that a player should "never fail the first level." I totally didn't understand that at the time, but it rapidly became something I firmly believed in.

With those things in mind, the best tutorials are ones that don't feel like tutorials at all. In fact, the very best ones don't exist - the game just lets you get on with it and, by the time you reach a point that requires a certain skill or control, you've already learned it in a safe environment earlier on.

Everything you need to know + a safe level to practice in

Sometimes it's quite easy - all you have to do is tell the players what each control does and the rest of the game is entirely about them working out the correct control or combination thereof to use at the correct time to progress.

Other times... not so much. The difference between, say Super Meat Boy (one button jumps, another one makes you run, don't touch spiky things) and Game Of War (even now I can't tell you how to play it) is staggering.

Focus

In all of this, the most important thing is to actually try it out. That is, try it out on some fresh meat - people who have never played before. In the days before analytics were a thing, we used to refer to this as Focus Testing.

Running a Focus Test is a skill in itself. Imagine you're watching someone playing your game and they can't even clear the first hurdle. They're stuck. They keep playing it 'wrong'. If you're a developer, watch the Cuphead footage that's doing the rounds and tell me what it does to your stomach. It takes a very strong will not to intervene. To explain what it is they have to do and where they're going wrong. To not just snatch the controller from their incompetent hands and do it yourself.

But that's the thing. You can't ship you with the game. You're not going to be there to show or tell everyone how it's supposed to be played. That's what your tutorial is for. So you test it. You Focus Test it.

That means you sit there, quietly, and observe everything the player does. Take notes. Lots of notes. But, and this is important - say nothing. Bite your tongue. Afterwards you can ask the player what they were thinking***. This is something that analytics won't show you. They'll show you how many people got passed a particular point and even what they did but they won't tell you why.

Looking at the Cuphead example there are several mechanical things all working against each other. It starts out okay - the controls are pretty simply laid out and hard to miss. "Press A to jump", right at the point where you're going to need to jump.

Then the wheels fall off. "Press Y to dash" is all well and good but the obstacle requires more than that. Not only do you have to dash at the apex of the jump, but you have to backtrack, jump on the previous obstacle that you thought you were done with and then do a jump / dash combo with a very small window for success. Your analytics will show you that the player was pressing dash and pressing dash and jump. It might even tell you how long it took them to go from behind the first obstacle to behind the second. But unless you can actually observe them playing it or, better yet, find out what they were thinking whilst doing so, it's going to be hard to fix from just raw data.

The graphical style is simply stunning

There are probably 2 or 3 missing tutorial steps between "Press A to jump" and the tall pillar puzzle: Dashing on the ground (possibly over a small, non-fatal hole), Dashing in the air (over a thing that pushes you back if you touch it) then, finally, Dashing in the air after jumping up on something (although that something should be in front of you and not associated with a previous stage in the tutorial).

Maybe you don't even include that final step in the tutorial, but introduce it as a puzzle element in one of the levels once the player has gotten the hang of the basic controls? Reinforce the mid-air dash early on with the use of collectibles placed along the desired flight path even though there is no obstacle. Then one with the obstacle but the collectibles still in place. Then you can rock the finished puzzle with no guides.

All of this sounds like I'm against analytics. That's not the case - far from it. Focus Testing is incredibly resource intensive and it doesn't scale very well. That is to say that it works best 1 on 1 which, assuming you want more than just a handful of people playing your game, isn't going to work later on.

As with so many things, it's about balance.

* Spoiler alert: it's the purple one.
** Assuming they haven't died of boredom before then.
*** A well run Focus Test will try to get the testers to vocalise their thoughts as they play, but you can't always rely on that.