Let’s Study The Witness

This past March, at SCMS, I walked out on a paper being delivered by Oscar Moralde on The Witness (Thekla, Inc, 2016). I did so not out of disinterest. (I’ve enjoyed Moralde’s papers in the past.) Nor did I do so out of rudeness. Rather, I did it because of spoilers. Moralde was kind enough to warn ahead of time that his paper would spoil a small portion of the joy of teasing out the behaviors of The Witness’ world, and advised those who hadn’t played it to leave, lest they deny themselves a rich intellectual—and some would even say emotional—experience of personal discovery. And, in my eternal shame, as of March of 2017, I still had not played The Witness. Even though it had been sitting right there in my Steam library for months. (Ashlyn Sparrow and Whitney Pow can attest to the truth of this story.)

Moralde’s paper was a wake-up call to me that I needed to get better about my gaming backlog, if for none other than purely academic reasons. And I think I’ve done a pretty good job of keeping up on things in real-time since that moment. (I played Tacoma, already!) I offer this story, though, not (strictly) as a chance to to advertise my newfound dedication to keeping up with recent releases, but also as a warning. Basically, the heads-up Moralde offered in front of his talk also applies here. The pleasures of The Witness are the pleasures of discovering puzzle mechanics, and you will deny yourself a small portion of those if you watch this new video essay I’ve whipped together.

That said, if you don’t mind spoiling such things, or if you’ve played The Witness already, go ahead and dash right in. This video is considerably shorter and more focused than my previous experiments in the “Let’s Study” format. It focuses on the pedagogical aspects of the game’s puzzle design, in particular its fondness for safe failure. Whether it’s encouraging assumptions about its mechanics that quickly get proved wrong, or setting up perceptual bad habits only to nip them in the bud, Jonathan Blow’s puzzle design in the best portions of The Witness front-load failure, so as to hammer home lessons. I hope you enjoy my short tour through this technique!

As before, a full transcript of my narration is below the fold. (I’d love to eventually add these as subtitles to the YouTube upload for accessibility reasons, but that is beyond my abilities, at the moment.)

Hello, and welcome to “Let’s Study The Witness.” This is a little piece of video criticism & video analysis, in which I’ll be looking at some of the design techniques in Jonathan Blow’s The Witness, released in 2016. Specifically, I’ll be looking at The Witness’s approach to pedagogy—that is, the methods by which it teaches its players the fundamental rules of its puzzles.

Because my interest is in pedagogy, you would be right to assume that this video will primarily focus on the tutorial sections of The Witness. But I’m not going to focus on any and all tutorial sections. The first tutorial section of the game, for instance, in this castle courtyard area, is quite successful, but I’m going to forgo analyzing it. What I am mostly interested in are the multiple-screen sequence panels that dot the landscape of The Witness, and are used to teach players its most intricate puzzle mechanics.

We encounter the first of these panels not inside the castle courtyard that opens the game, but here in the farmland area just down the path from it.

The principle that this panel is trying to teach us is that we should separate out squares marked with a white dot from squares marked with a black dot. It’s a fairly simple principle, and, as a result, this tutorial panel is fairly simple in its design.

The positioning of the start point and end point of the first screen make it almost impossible not to make this connection immediately and intuitively. We’d have to do more work—with both our eyes and our hands—to try and not separate these dots out.

The second screen gives us the exact same principle, while shuffling the start and end point, to make us actually think about what we’re doing a bit more.

And, gradually, we expand things. First along one dimension …

… And then into two dimensions.

… then we get more elaborate shapes to trace …

… and our start and end points shift, forcing us to use the outside borders of the puzzle in more deliberate ways …

And so on, an so forth. I don’t think this initial panel is revolutionary as a tutorial, but I was very, very impressed when I saw it, because it illustrated to me that Jonathan Blow had grown from a developer from his previous game, Braid.

Now, Braid is justly famous, and I’m not going to be too iconoclastic here. But I think it’s fair to point out that its puzzle design was not as elegant as that of The Witness. Sometimes it threw too much at the player, too fast. There are two levels that really aggravated me the first time I played Braid, that still stick in my head to this day.

One was “Fickle Companion.” On paper, it’s a brilliant use of World 4’s “move backwards and time moves backwards” mechanic. But, in practice, the behavior of the key is so finicky as to appear arbitrary, which disrupts the player’s ability to learn its rules. Especially when … you’ve got to be kidding me … no … just please … come on … oh thank god.

The other was “Fragile Companion,” a puzzle that required you to realize that if you rewound time while running forward, your momentum would be conserved for your shadow-clone, who continues running a few paces past where you had hit the rewind button.

The problem is, World 5 never gave players the slightest indication of this clone behavior. And we used clones to do a lot of things in World 5!

I used my shadow-clone as bait for a psychotic rabbit, so that I could steal a key from it!

I used my shadow-clone to pull levers for me!

I bounced a goomba off of my shadow-clone’s head, so that I could bounce off of its head!

And I even, yes, used my shadow-clone to unlock doors for me before. But I never used it in any way that would remotely suggest that it conserved running momentum when I hit the rewind button.

These two levels became, in my mind, emblematic of the failures of Braid to introduce mechanics and, crucially, test if the player has understood them before more on to more elaborate uses of them.

So when I finished this first multiple-screen sequence panel in The Witness, I had a big smile on my face. Because it is emblematic of how much Jonathan Blow had grown as, if not a puzzle designer per se, than as a teacher of puzzle mechanics.

There are over half a dozen places I could go next to show you examples of multiple-screen sequence panels in The Witness. Going to all of them would result in a very long video, and I don’t want to stretch this video out too much. So, instead, I’m going to spend an extended amount of time looking at one area: the “marsh” area. This area is notable for two reasons. One, it just has a lot of multiple-screen sequence panels. But secondly, and more importantly, I find the marsh to be one of the most concentrated and successful example of the pedagogical philosophy in The Witness. I certainly wouldn’t call the other tutorial areas of the game failures, by any stretch. But the marsh, for me, stands not just as a pinnacle of the game’s tutorial design, but also as its clearest statement of pedagogical philosophy. So I’ll be spending more time with its puzzles in this video than with any other puzzles in the game.

Here’s is the tutorial panel for the marsh section. It’s notable for having two separate sides, with fourteen puzzle screens, overall. We’ll begin at the beginning.

While it’s technically possible to get this first screen wrong—by going straight up the sides, rather than moving across the middle—it would be pretty difficult. The general shape of this puzzle shares a lot in common with the second screen of the “black and white squares” tutorial panel. Recalling it, we are encouraged to assume that what we want to do is separate out the square with the yellow mark in it from the square without the mark.

The second screen demonstrates that this wasn’t a fluke: Again, what we want to do is to separate out the square with the mark from the squares without one.

In its similarity to the second screen, the third screen invites an assumption: that, again, what we want to do is separate out the “different” square. This assumption, though, proves to be erroneous. Even if players don’t immediately grasp what’s going on here, though, there are a limited number of solutions—eight, if I’m not wrong, two of which are correct—enough for players to brute force their way through.

Upon doing so, we can revise our understanding of how these things work. What actually seems to be happening is a sort of “expansion.” If we turn back to the first two, we realize that if we “expand” the little yellow square, it fills the squares we’re cordoning off from our empty squares. The same thing is true of the third—only, this time, the “expansion” fills up the adjacent empty square, as well.

So far, we’ve learned how to do this expansion one-dimensionally. The fourth screen shows us how to do it in two dimensions. Now we have an x- and y-axis, and the shape is expanding along both.

Although we’re dealing with two dimensions now, we’re dealing with them in a very constrained way: there are no sorts of decisions that are being made about how and where to expand these yellow shapes. The next two screens ease us into that.

So, on the fifth screen, we get a square. There’s actually only one way to expand this, completely prescribed by the breaks in the lines. But at least now we are vaguely aware of the possibility that it could be expanded in different directions, if the breaks weren’t there.

Same thing goes for the sixth screen. Now, I don’t know about you, but just looking at this, I have a strong perceptual urge to put the yellow figure right at the corner of where the two legs intersect. This urge, though, is forcibly denied. I can’t help but think that we’re supposed to try this, and fail, just as a way of emphasizing all of the myriad ways that a form can expand that we might not immediately perceptually jump to. In this case, this is the solution.

Alright, that’s it for the first side of the panel. Now, onto the reverse side.

If the initial side of this panel taught us positioning, this one teaches us the way in which shape addition works in these puzzles.

So, the first panel on this side. Again, we’re in a situation quite like before: because of how these are shaped and where they’re placed, we can’t really re-position them at all. The only thing to do, then, is to trace the outline of them once we’ve mentally expanded them.

Then we move on to the second panel … and we hit a serious road block. Given what we’ve been doing so far until now, this panel doesn’t make sense. It breaks the rules we might have assumed we were operating under.

If we expand the square, we have a little bit of leeway as to where we’re positioning it. The three-square line, though, is a different matter. There seems to be no way we can expand it without it simultaneously occupying areas of the grid that the square is occupying. And the rules obviously forbid this type of simultaneous occupation. If we try it, it doesn’t work.

The game isn’t telling us much here beside “what you thought you knew before is wrong.” But, given that there’s only nine squares on this grid, there are not a lot of permutations to try. If we happen upon a correct one, we learn how to revise our previous assumption:

Unlike where there is a single yellow shape, the rules of multiple yellow shapes are different. The “adding” of multiple shapes allows us to treat them in different ways. Namely, we can re-position them quite thoroughly, as long as the squares with the original shapes in them fit within the contours of the larger shape that we’re drawing.

This rule didn’t apply when we were on the first screen, which seems deliberately engineered to briefly give us a false sense of security, before exposing us to something new, that messes with our previous assumptions. And the next few screens are going to bounce back and forth, using the knowledge we previously knew, and the knowledge we now know.

So, the third screen again requires us to re-position as we add the contents of the two shapes. The fourth screen doesn’t. The solution for this one is immediate and intuitive, and it’s included just to show us how there can be multiple types of positionings that result in the same “shape sum.” And, again: the same thing on the fifth screen.

The sixth screen tapers us down, back into the realm of simplicity, as does the seventh screen, which has one obvious solution. The eighth and final screen on this side of the panel throws a bit of a loop, teaching us that we don’t necessarily need to add yellow square shapes together. We can separate them, if we wish, even if they are adjacent to one another.

Now, we can get on this moving platform, which, in a nice detail, is a physical reification of the concept behind these puzzles. The control system for it follows the standard rules of the puzzle that we have learned, and it also functions as an exact analog of the device we’re standing on.

Now, it’s not my intention to do a complete run-through of any of the areas in The Witness. I mainly want to limit myself to tutorial panels. But there are several other panels in here that work as a sequence to introduce new ideas, and I want to turn to just one more, in particular: this sequence right here.

Until now, we have dealt with re-arranging shapes as we “add” them, but we have never rotated them. The solutions of the previous screens indicated that it couldn’t be done. That changes here, as we realize that some shapes can be rorated. Not all of them, but those that are at an angle.

The first screen on this panel makes this obvious. There’s really only one way of conceiving the solution to this puzzle, and it means that the shape must be rotated. So, mentally, we do.

The second screen asks a bit more of us, in that we can actually get it wrong. But not much more.

The third screen blends this new concept of rotation with the old concept of us being able to “add” overlapping shapes, moving them around inside the contours of a shape as we do so.

This fourth panel is, in the grand scheme of things, pretty simple. But I love it, because it forces us to break a bad habit we may have acquired. In all of the previous panels, the degree to which we had to mentally rotate the shaspes was less than 90 degrees. So, I think most people’s first reaction to this final screen will be to do the same thing, and mentally rotate this shape just a bit.

But if we do this, we have a problem. It won’t fit right. No matter where we try to put it, the gaps make it impossible to complete.

So we have no choice but to rotate it significantly more than 90 degrees. This isn’t actually a “hard” puzzle, by any stretch of the imagination. But it is surprising the degree to which it is perceptually unintuitive to mentally rotate a shape this much. Obviously, there is an awareness here that we’re not going to get past this psychological hurdle unless we’re forced to, early on, in a simple puzzle that firmly puts the breaks on our bad habits.

I find these tutorial sequence panels in the marsh to be among The Witness‘s most pedagogically successful moments. I think that, in them, we can really find the core of Jonathan Blow’s design philosophy. To explain this a little bit more, I’m going to draw on the concept of “safe failure,” an idea that has floated around the writings of teachers and academics interested in applying lessons from play in matters of pedagogy.

The idea of failure as a necessary stepping-stone to success is today pervasive in the areas of management, software development, and startup culture, where one frequently hears the rallying cries of “fail forward,” “fail faster,” and “fail better.” In education, of course, “failure” has traditionally been laden with severe academic consequence. Challenge to this paradigm has been mounting, however, as pedagogical theorists have begun looking to videogames as possible sources of inspiration.

“Part of what makes play valuable as a mode of problem-solving and learning is that it lowers the emotional stakes of failing,” write Henry Jenkins, Katie Clinton, Ravi Purushotma, Alice J. Robinson, and Margaret Weigel in a MacArthur Foundation white paper. “[P]layers are encouraged to suspend some of the real world consequences of the represented actions, to take risks and learn through trial and error” (Jenkins et al, 23). Shira Chess and Paul Booth expand upon this basic idea, pointing out that “[o]ne expects to fail in a game, and the consequences for said failure are minor, as opposed to failure in the classroom” (Chess and Booth, 1006).

So that’s “safe failure,” in a nutshell. But we can also take things one step further. Jenkins and his collaborators specifically liken the safe failure offered by videogames to the failure inherent in the scientific method. “No sooner does a player enter a game,” they write, “than he or she begins by identifying core conditions and looking for problems that must be addressed. On the basis of the available information, the player poses a certain hypothesis about how the world works and the best ways of bringing its properties under their control. The player tests and refines that hypothesis through actions in the game, which either fail or succeed. The player refines the model of the world as he or she goes” (24).

For these authors, videogames offer up a general philosophy of safe failure and hypothesis testing that can serve as a broad pedagogical model. Of course, what I’m undertaking here in this video is a more specific case study. And what is interesting to me about The Witness is not simply that it offers a safe space for failure—nearly all videogames do that—but that its best-designed tutorial sequences seem specifically engineered to front-load failure. Based on just what we’ve seen here in the marsh, I would boil the pedagogical philosophy of The Witness‘ tutorial sequence panels down to three maxims:

  1. Players are more likely to remember rules that are learned via the modification or outright rejection of an initial hypothesis.
  2. Puzzle design should encourage the formation of faulty assumptions early on, so that these assumptions can be challenged through a process of fast failure.
  3. Applying good habits in puzzle-solving is foremost a matter of breaking bad habits. Therefore, you should force players to break their bad habits ASAP.

There’s enough similarity between the three points I’ve mentioned here that you could, if you really want to, compress everything down into the nutshell version: “GOOD LEARNING = CHALLENGING ASSUMPTIONS EARLY & OFTEN.” That the game is committed to this pedagogical philosophy in its puzzle tutorials should come as no surprise, as this general theme pops up multiple times in the philosophical musings recorded on the game’s audio logs—I am thinking particularly of the quote from Richard Feynman on the audio log here, on the outskirts of the jungle …

… as well as the two audio logs from William K. Clifford. One of those is found in here, by the statue of the bag-toting man in the keep …

… and the other is found in the audio log here, on the shipwreck. I’m going to play this Clifford quote on the shipwreck. It is a short parable, taken from the opening paragraphs of Clifford’s 1877 essay “The Ethics of Belief,” and it is a potent distillation of these themes.

“A shipowner was about to send to sea an emigrant-ship. He knew that she was old, and not well built at the first; that she had seen many seas and climes, and often had needed repairs. Doubts had been suggested to him that possibly she was not seaworthy. These doubts preyed upon his mind, and made him unhappy; he thought that perhaps he ought to have her thoroughly overhauled and refitted, even though this should put him to great expense. Before the ship sailed, however, he succeeded in overcoming these melancholy reflections. He said to himself that she had gone safely through so many voyages and weathered so many storms that it was idle to suppose she would not come safely home from this trip also. He would put his trust in Providence, which could hardly fail to protect all these unhappy families that were leaving their fatherland to seek for better times elsewhere. He would dismiss from his mind all ungenerous suspicions about the honesty of builders and contractors. In such ways he acquired a sincere and comfortable conviction that his vessel was thoroughly safe and seaworthy; he watched her departure with a light heart, and benevolent wishes for the success of the exiles in their strange new home that was to be; and he got his insurance-money when she went down in mid-ocean and told no tales.

What shall we say of him? Surely this, that he was verily guilty of the death of those men. It is admitted that he did sincerely believe in the soundness of his ship; but the sincerity of his conviction can in no wise help him, because he had no right to believe on such evidence as was before him. He had acquired his belief not by honestly earning it in patient investigation, but by stifling his doubts. And although in the end he may have felt so sure about it that he could not think otherwise, yet inasmuch as he had knowingly and willingly worked himself into that frame of mind, he must be held responsible for it.”

There’s an overriding message here, also present the other Clifford and Feynman quote that I mentioned: It is necessary to refine one’s beliefs through doubt. This refinement process gets us closer to accurate observation of the world. It is also a uniquely human capacity, an expression of our best potentials. And Clifford would furthermore posit that there is an ethical dimension to this, as well: there is a fundamental sense in which we don’t deserve to have beliefs unless we have tested them, had them fail on us, and altered them accordingly.

In the game’s best puzzle sequences, we see a confluence between this message spouted on the game’s soundtrack, and its puzzle design.

As I hope I’ve demonstrated, we see it all over the place in the marsh.

We also see it in the treehouses area. The early puzzle screens in the tutorial for this area encourage us to assume that we’re dealing with a “like with like” logic. We want to keep the orange sun-shapes together, unseparated.

But then, after encouraging this assumption, the sequence forcibly breaks it, with this screen. It turns out what we want to do is keep not all sun shapes of the same color together, but instead to break them into sets of twos.

Next, it introduces multiple colors of sun shapes. In the introductory screen, encourages the assumption that we should separate out these colors, much as we did when we kept the black and white dots together with their own kind.

But this belief is immediately challenged on the next screen. We were encouraged to briefly adopt it for the sole reason of discovering early on that it was wrong. It’s impossible to separate out the colors in this screen. So it’s not something you have to do.

It turns out it doesn’t matter if different colors are together in these types of puzzles.

The important part is to keep the same-colored sun-shapes in pairs.

And we see it in what is perhaps my favorite puzzles in the game, in the climax area inside the mountain. We we first approach this sequence panel, we assume that it’s going to work like every other sequence panel in the game.

But that belief is immediately proven wrong. In fact, we’re not solving these screens one at a time: we have to find a single solution that works simultaneously for every screen.

What follows is a process of refinement, that I won’t spoil in its entirety. But it is very much like the process of constructing a scientific theory, and then refining it through experimentation. To continue along this sequence is to discover more an more things that you have to account for, that your initial proposed solution didn’t account for. You must refine your solution through failure, much in the same way that actual scientific models of our universe are refined, to successfully account for more and more phenomena.

I hope that you’ve enjoyed this tour through some of the most interesting tutorial sections of The Witness, and through its failure-driven approach to pedagogy.

Again and again, throughout its puzzle design, The Witness tells us: It is okay if, during experimentation, your hypothesis is disproven. This is what experimentation is for. True intelligence is always being open to your hypothesis being disproven—to in fact be excited when your beliefs are proven wrong, and new possibilities open up. I think it’s safe to say that this sentiment sits at the core of Jonathan Blow’s broad humanism, as expressed in the game.

Thanks for watching!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s