Page 160:

Planning the Level:

“Grey-boxing”. This is a method employed by level and environment designers where early models of the level are created to get a sense of player movement, sizing of the space and placement of items from furniture to power-ups to health packs etc. The concept is to begin to build up an environment so that it ‘feels’ real and natural when of course it’s designed and it’s also maximized for player interaction. Knowing the sightlines of the player aids in placing interactive objects (primary) against flavor objects (secondary).

Grey-box.mp4 video

Once I'd refined the process a bit, I had testers fill in a Google Docs Survey about the build, asking which levels worked best, what the best and worst things were about the game, and what score they'd give it if it was released in this state. The biggest changes I made in response to testing were generally level design - I'd usually outright delete the two or three least popular levels and build new ones based on what people liked about their favorites. The average 'review score' from testers increased with every new test build, and that was one of the most satisfying and motivating measures of my progress.

Creating Levels that are Linear, but feel non-linear:

In these two videos I discuss how using a ‘hub and spoke’ model for an exploration environment, along with impossible architecture, creates the impression of player freedom. The architecture of the house relies heavily on the player’s perception of ‘houseness’ and preconceptions of how buildings work. The player is able to go in many directions (non-linear) but the level ultimately pushes the player where the designer wants them to go (linear).

Level-Design-01.mp4

Level-Design-02.mp4

Page 170:

Extra life!

The interview with game designer Tom Francis was too long for the book and so parts were edited out. Here’s the rest of the interview where Tom talks about editing and testing.

Trimming the fat:

There were two gadgets that I ended up scrapping because I felt they were hurting the game. The first was the Transfuser, which just consumed some battery power to let you link two things that are on different circuits. It was meant to open up new level solutions, which it did, but since separate circuits are my main tool in puzzle design, it felt more like 'pay to skip this puzzle'. I already give you the option to skip 2 levels in case you get stuck in a non-fun way, so this wasn't really serving a purpose and it was hurting the motivation to look for more elaborate solutions.

The other was the Smashbang, a glass device you could place in the level, then wire something to it to make it loudly explode. It was just a distraction device - guards would come running - and it was kind of a fun idea. But it complicated the controls - you needed to be able to pick it up or place a new one, and since you could do that near some stairs that you can take up or down, you would need four 'Use' buttons to be able to consistently do what you intended in every spot. I could have designed around that, but in the end I wasn't happy enough with its effect, which like the Transfuser often felt like you were simply skipping puzzles.

The concept of an environment you could alter led to a noir setting which also needed to be near-future to explain all the outlandish gadgets. I then sent out a call asking for artists to submit possible aesthetics. We got a huge variety of possible looks for the game from 34 different artists, and I spent a long time deciding between them. Fabian's gloomy neo-noir cityscape felt like a really exciting place, and John Roberts' lean, faceless, realistically-proportioned characters worked for every situation I wanted - from tragedy to slapstick - and expressed personality in the way they moved.

One constraint was not knowing how to build a game in Game-Maker or knowing what the software was capable of producing.  Most of the look and feel of the game comes from learning the software as we went along and having to produce artwork within the constraints of that game engine.

For example it’s not entirely intentional that the camera is fixed, instead that’s largely because I developed the game on a low-resolution screen: 1360x768. For me, the camera pans to track you. But some levels are only 1920x1200 big, so the experience is pretty different on screens running that. Adapting fixed-resolution pixel art to the massive differences in people's screen resolutions is a nightmare, one I know to sidestep in the future.

Testing the game:

To test the builds of the game asked anyone interested to sign up for testing right from the first week I started working on the game, and sent out new builds on a fairly regular basis over three years. The pool of testers was only about eight people for the first round, but grew to 15,000 by the end of development. Once it was over 1,000, I typically only sent a build to 1,000 people at a time, since the volume of feedback was about as much as I could process.