The Witness

Jonathan Blow, creator of Braid – one of the most critically acclaimed games of recent years – returns to the GameCity Ballroom to give you a sneak preview of his forthcoming project, The Witness. Exclusive in the UK to GameCity!

Jonathan Blow.

My Thoughts

A Myst-like game with non-linearity and less stringent puzzle solving mechanics as in Braid – sounds interesting, even though I’ve never played Myst, but have played a lot of Professor Layton.

One thing that came to mind after it however; yes, I do like how the tutorial guides you up through complexity without words, but I wonder if there will be too much hand holding to guide the game; we’ll see when it’s done no doubt and I am sure he knows what he’s done, even if it is sort of “I don’t playtest, I test my theories about how people will play! muahahahha!” *strokes white cat*.


Want to do an adventure game; but wanted clarity. Adventure games are the least clear games in existence (see Penny Arcade Myst strip!).

The gameplay is also not how the designer intended. The designer wants you to have an experience figuring things out well, but the players experience is usually more the confusion then finding out the answer online on how to solve the problem.

Standards of playability has changed, so no lament for the adventure game genre. A lot more playtesting; adventure games have tried to evolved, but unable to advance to keep pace since how do you deal with the player getting stuck with a puzzle. People today are going to choose things that are more playable today.

Problem is still want to make an adventure game. How do I do that?

Don’t want to exactly make an adventure game; so something in the spirit and in the tradition of the adventure game but removing all the things that hold it back. Similar to Braid; taking a platformer and adding a rewind. Other designers say why add something that removes all the failure? Braid added something in since this was taken out.

Removing the adventure game problems; No inventory, since they need a massive amount of space and clicking to combine things. Things will be clearly depicted where all switches are exactly the same looking; rather then old adventure games where they are all different. Puzzles are clearly marked – no question they are a puzzle, and no question on how to solve it in a general sense. The game is structured for non-linear progress, where you don’t get stuck.

Adventure games have had more and more easy puzzles so players don’t get stuck on the linear story. Nonlinearity means the player can leave it, even possibly ignore it and still finish the game.

Epiphany can’t be demoed – it’s a player experience that the player has to have.

Want the game to take place in a 3D world like Myst and other graphical adventures. There are a lot of things in 3D games that are problems for me. Some are indicative that we’re missing some things as designers. In first person shooters, today they’re very expensive – players want richer and richer graphics, with unique maps – no backtracking. Lots of scripted events, and illusion of progress in the game. The problem is being good at a FPS is killing things quickly and getting past things. It means getting through the painstaking content very quickly; if you’re going slowly you are not playing it well are you?

Assassins Creed – sort of trying to be a spy in a big city, and being very perceptive and executing your plan. But they give you a radar so don’t need to engage with the world since you just look at the radar to find where to go and do the mission. Graphical effort helped sell the game but it is superfluous to the gameplay and even actively ignored by the gameplay. I’d love to play a Assassins Creed version designed without the radar.

Fable 2 – beautiful world, no radar – but a lienar world. No dream dog – need to learn to interpret the dog. There are icons over his head though so you know when to dig. A missed opportunity.

There is something, some magic of existing in a world that is 3D, that a lot that can be done with it and it seems to be ignored a lot of the time. With The Witness can focus on the sublties.


Not going to show all the cool things in the game, but just the intro to the game. Showcased it a month ago at Pax, it worked exactly as I wanted it to, so I guess the game is working. All pretty much placeholder art. Don’t infer anything about how it’ll look. Prototype code is also running slower then I’d like.

In many adventure games you’re a protagonist with amnesia and later find your past – The Witness plays with this in a self-referential way.

The initial puzzle is in tutorial mode where it shows you want to press and do. Some following the dot along a path puzzles (mazes), on a blue panel. It opens up somewhat after a few doors, and there is some branching paths and non-linearity.

The initial panels help learn and get used to the panels. A wordless tutorial on how the game works. It is important for this to happen effectively – it has to be at least a bit interesting, the first ones are not really but only last for about 5 puzzles.

Before couldn’t verbalise 3D space; now with some choice of areas to explore there is some surprise and mystery (not great but enough for the beginning).

Seeing a lever shows you how it works with the hoses. The last hose has a button and some hidden hoses; this leads to button pressing. There is a long string of non-verbal communication that the designer has to say to the player. It is how the semantic payload of the game is delivered.

There are some puzzles right after the tutorial where it is complex; so much so you might thing to come back later to it. Going on you see an area with a set of these puzzles in a simpler way. Got after a few puzzles an evolving concept; knowing you can’t go over your own path, that you need to go between white and black. Having the puzzles next to each other it is like a set of sentences, and all of them together like a paragraph. Can have very sophisticated and subtle experiences from the game.


Audience: How do you know what is going on in peoples head?

Jon: Blind text at Pax, where it wasn’t said it was by myself. Most people went through exactly how I thought. I don’t know exactly what is going through players head though. You can go to things in any order, and are thinking different things, even in the intro experience. It helps carry them through the rest of the game however.

Audience: Talked about spacial arrangement of puzzles in Braid and fitting things in screens. With a 3D camera it is quite hard to control what the players see. How is that working at the moment?

Jon: That is true, and that is a large concern I had. With Half-Life 2 they go to great lengths in the commentary to do signposting in game design to call players attention. Even when you think you are going through a tremendous effort to do it, it might not be enough. The visuals are placeholder but will be minimalist even once the final art is done. It removes distractions so signposting is easier – think Mirrors Edge in a way. One part is not having a volcano even if the volcano is awesome since the players might get lost going to it. Control over the camera is another, the game is about discovery and epithination – unlike Half-Life where it is a linear rollercoaster ride. For this game it is taking into account people missing things and maybe just ignoring things people don’t want to see. There is also audio and other things to communicate rather then be a problem – it repurposes it into gameplay.

I feel bad for not showing the

Audience: When creating a consistent set of motifs do you worry about getting rid of the suspension of disbelief and so you starkly know what the designer intends?

Jon: Yes, I know, but no I don’t worry about it. It isn’t Lord of the Rings. There is constantly an awareness they are in a game but accept it early on. We don’t understand game designs well enough yet to do the fundamentals. Putting everything in fiction limits you a lot. No one thinks about the switches in Braid which are all the same, it’s a convention in there. In comparison Myst had a different thing in every puzzle, all switches were different.

Audience: Is it going to have Braid-like story or just all minimalist and abstract?

Jon: Yes, there is intention there, audiologs like Bioshock. The first room would have that kind of log on it. It is not a Bioshock like story, or a Braid like story. It is an envelope of ideas, that helps shape the game. It hopefully has a journey for the player going through it. If it doesn’t work the game stands on its own, but it is a good idea to try it. Just starting to experiment on it now.

Audience: Finding old puzzles in adenvture games is frustrating. You’ve removed the radar; and Braid had numbers for puzzle pieces. How do you deal with this?

Jon: There are different areas like a theme park, the puzzles inside each area have their own theme. “This is the black and white spot puzzle area”. On top of that, and I should show you this but I won’t. It can’t be an entirely flat game since it appears to be just random puzzles, but it needs to be non-linear. When you get to the end of an area you get a communication laser pointing to a building that you can’t get to yet. You can see the laser towers, and open the gate to the new area. You can see the turrets, and the challenge is getting to the turret and activate it. The plan is that takes care of the problem, and I believe it will.

Audience: You were talking about the magic of 3D. To younger developers you say “Do a 2D game, do a 2D game”. Discuss.

Jon: Doing a good 2D game which sells well on XBLA, make enough money to then go and do a weird abstract 3D game, with enough money to do this freedom is something I’m half way through and I don’t recommend it. I had a 2D engine for the Xbox already, and plans for a similar Braid-scope 2D game. However working on the 3D game, even though it is harder and more expensive if I don’t do it now. Recommend simple things in gamemaker for beginners. Don’t do this first.

Tried this with my first game and company with the hovertank pixel-specific graphics with texture shading and stuff learnt, and burnt me out. Don’t do what I’m doing unless you really want to have a difficult time with it; Super Meatboy came out and sold 50-90thousand copies on XBLA.

Audience: How do you take on board the criticism that some players with Braid had them complaining that they needed to figure out how the designer wants you to play it?

Jon: I disagree with that criticism. Some games allow you to find a way up to a location. Braid is not about player freedom. Crosswords and Soduku are the same. It’s simply a different genre. Some puzzles in Braid are pretty hard and so express frustration and say why can’t I do more, when in fact they simply got stuck on a puzzle. I sound like a dick, whatever!

The Witness is very different to Braid. There is more freedom as more complexity is added; sometimes the opposite problem – they solve it without realising what they did to finish the game. It’s not that I think one is better then the other – it is just a different genre of game.

Audience: Why did I not do it in 2d?

Jon: So yeah, these square puzzles could just be done cheaper in 2D right? Not talking about it today.

Audience: What is the Epiphany? You can’t really talk about it since it’d destroy the epiphany; how do you get it tested and get the press play it?

Jon: The way I’d prefer to deal with the press is respect them. Someone could easily spoil the game and that would suck. I’ll be coping with that problem as time goes on. As with feedback part, I don’t really like feedback. Reality check at Pax, blind playtesting; is what I think was happening actually happening. Could potentially turn parts of the game off to do this.

Leave a Reply

Your email address will not be published. Required fields are marked *

Website and journal of Andrew Armstrong