Zero to 95,688: How I wrote Game Programming Patterns

April 22, 2014 book game-dev game-patterns

Update 2014/04/28: In case you missed the link, the web site for my book is gameprogrammingpatterns.com.

About an hour ago, in the quiet of my living room, alone except for a sleeping dog next to me, I accomplished the biggest goal of my life. I finished writing Game Programming Patterns. It’s a book on game programming (it would be a strange title for a book on ornithology) that I started writing about four years and a lifetime ago.

What I see now when I run the script that converts my Markdown manuscript to HTML.

It feels weird writing a blog post that doesn’t have any real content beyond my own personal story, but what the hell. It’s not like I have anything better to do! I get some vicarious pleasure (mixed with heaps of envy) when I read about other people finishing their books, so I’ll try to add to the canon.

The Call to Adventure

Like most stories, it starts with the hero having something bad happen to him. (Did I just call myself the hero? Seriously? God, this is already going my head.)

About five years ago, I was a game programmer at Electronic Arts in sunny Orlando, Florida. That’s the studio that does Madden, NCAA Football, Madden, Tiger Woods Golf, Madden, and also this football game you’ve probably heard of. They did a few other one-off games too.

I’d been there seven years, which is an impressively long time to actively dislike football while working in an office that lived and breathed it. The last game I worked on, Henry Hatsworth in the Puzzling Adventure, was an absolute blast, the kind of dream project you imagine game development to be all about. Just seven dudes hanging out making a cool game they all loved.

My pixel art alter ego in the end credits.

After we shipped, EA decided to never make that kind of mistake again and refocused on suckling the withering teats of its aging cash cows the shareholder-friendly profitability of beloved annual franchises. My entire team quit except me. I ended up bouncing around onto a bunch of different projects.

I was so burned out chunks of char were falling off me, and I was really frustrated by how hard it was to ramp up every time I was airdropped into a new team and its disorganized, deeply coupled, inconsistent code. I’d find bits of real elegance sitting just a few source files away from some hairball that would have benefitted from the exact same structure. People just weren’t talking to each other about their craft and weren’t learning.

Game dev culture, at least at the studio I was at, is kind of weird. One quirk of it is that a lot of the programmers I worked with didn’t give credence to ideas from the larger world of software. Things like Design Patterns were for Nancy-boy enterprise programmers, not real game coders.

On top of this stress, I’d just had a kid, an eventuality I did not plan for when I purchased a 900’ house at the peak of the housing bubble. One frustrated drive home from work, I had an idea: I knew some basic software architecture. I liked writing stuff down. What if I wrote a book specifically targeted towards game developers about this? If I aimed it straight at them, maybe they wouldn’t dismiss it out of hand.

A few notes from when I first started thinking about the book. Note that even then I was pandering to the masses on reddit!

(Also, let’s be honest, it’s not hard to write a more enjoyable read than Design Patterns. I’m pretty sure the Addison-Wesley style manual explicitly demands crushingly boring prose.)

Now, I am a world class project starter. I’ve got hard drives full of stories, videogames, art, music, screenplays, photography projects, table-top games, hell, there’s probably some poetry in there somewhere. But virtually none of it’s done. About the only thing I’d been able to get out the door at that time is the occasional blog post.

I was fully aware that taking on this project was pretty much doomed to failure. But, after visiting the Pacific Northwest, my wife and I really wanted to move, and a book would make for a nice bit of résumé padding to help me find a new job.

How could you not want to live here?

I felt like the book could be our ticket out west. So I needed to figure out every psychological trick I could play on myself to actually get it done.

The structure

If blog posts were the only thing I could finish, why not take a cue from that? Design Patterns is a stack of mostly unconnected chapters. I could organize my book the same way. Then instead of writing a whole monolithic book, it would feel more like writing a few dozen separate articles. Less novel and more anthology.

Likewise, each chapter in Design Patterns has the same top level headings and organization. I could do the same thing, so I didn’t need to come up with a unique outline and narrative for each chapter. They’d just be recipes with different ingredients and instructions.

The workflow

I’m an inveterate tinkerer and I knew if I didn’t put the kibosh on that, I’d spend all day futzing with CSS or some other stupid thing that wasn’t writing. So I spent a day or two putting together a minimal script that would take a file of Markdown for each chapter and convert it to HTML. Once I got it working, I swore to myself that I wouldn’t monkey with it (much).

Now I couldn’t get distracted by design, style, editors, or anything. Just me, Sublime, and a handful of markdown syntax.

The shame

One thing I’d heard was that telling your friends and family that you intend to do something is a good way to keep yourself honest. So I totally made myself look like an ass and told everyone, “Hey, I’m gonna write a book. Because I’m so awesome. Don’t you wish you were as erudite as me?” Maybe I didn’t word it exactly like that, but I’m pretty sure that’s how it came across.

After working at EA for seven years, they give you seven weeks off. I used some for my honeymoon and then started writing.

The incentive

I’d been blogging for a year or so and one thing I’d learned is that the possibility of a good reddit discussion about my writing was incredibly effective at getting me to complete something and put it out there. So I planned to put each chapter online as I finished it instead of waiting for the entire book to be done.

One of the first chapters posted.

Crossing the Threshold

I rolled this all by my wife and, unbelievably, she agreed to it. Even though it meant she’d have to spend even more time taking care of our infant daughter while I wrote. Even though she knew I’d never finished much of anything before.

I’d like to think she believed in me, but maybe she just really wanted to move out of that tiny bungalow. Seriously, that house was so small we had to move the stroller into the living room every time we did laundry.

With her OK, I started working on a plan. Since padding my résumé was a primary motivation, I needed a real publisher. I figured putting my self-published book on my résumé would sound about as impressive as that “World’s Best Son” mug I got from Mom.

I spent a bunch of time reading submission guidelines. They usually want an outline (check) and some sample chapters (oops). So I spent a couple of weeks writing and revising a couple of chapters. This was a nice trial to see if I could actually write something that looked like a book chapter. Miracle of miracles, I could! For the record, the first chapter I wrote was Object Pool.

I carefully looked at a range of different publishers and their compensation packages before making an educated choice about which ones to submit to. And by that I mean I sent it to O'Reilly because OMG having a book with an animal on it like Perl’s camel book would be SO RAD.

I emailed them the submission with butterflies in my stomach which, when you think about it, is an absolutely pointless physiological reaction given that I was in my house all alone staring at my computer and not being attacked by a sabre-toothed tiger. A heightened sympathetic nervous system does not actually make email arrive faster.

Unbelievably, they got back in touch. They were interested! I had an editor! He had feedback! He talked to me like I was an actual writer and not some jackass who decided he was an author because he thought it sounded cool. There was talk of a (small) advance. I felt like hot shit.

And then… somehow… it sort of fell through. I don’t remember the exact details, but at some point they decided to not move forward after all. I reached out to a few other publishers and found another one. I signed a contract with Apress and was super pumped to be back in business. We had a writing schedule and everything.

During all of this, I was still writing and managed to get a couple more chapters done. Unfortunately (but unsurprisingly and understandably) my publisher was a little hesitant to have me put them all online, so I didn’t get that jolt of dopamine every time someone on the Internet acknowledged my existence.

The Ordeal

I thought I was back on track but then something funny happened. It turned out I didn’t need to pad my résumé after all. I got a new job. Outside of the game industry and in the city we were desperately hoping to move to.

All of a sudden, we were packing up our belongings, kid, and pets; saying goodbye to friends and family; and moving along what is practically the longest straight line move you can make in the continental US. I was plunged headlong into learning all sorts of new stuff at work that had nothing to do with games. We got another dog and had another kid.

That writing scheduled was looking a little, uh, unrealistic. Eventually, I realized my heart just wasn’t in the book anymore. I called my editor, apologized profusely, and backed out of my deal. Without the pressure to write, I forgot about the book almost completely. I hadn’t abandoned it, really. I told myself I’d get back to it some day. It languished for a year. Then another.

It would have gone into my overflowing bucket of unfinished projects if it wasn’t for one thing: people kept emailing me about it. You see, I’d put the entire table of contents online along with the handful of chapters I had written. Every now and then I would get a long, wonderful email from someone telling me how a chapter really helped some abstract concept click in their mind and how they couldn’t wait for the next chapter. Good God, their enthusiasm made me feel terrible.

Resurrection and Rebirth

About a year ago, I had an unpleasant realization: as I got farther and farther from my days as a professional game developer, my expertise was getting out of date. I was losing the ability to finish the book.

While all my other unfinished projects didn’t bother me that much, the book was special. I’d only ever tried to write this one, and to have it die would break my heart. I figured it was now or never.

Around this time, I’d stumbled onto that article about Jerry Seinfeld’s trick for staying productive. It’s pretty simple: work on it every single day. Don’t break the chain of sequential days. I also met Chris Strom who’s been blogging this way for something like a million years.

Without putting much thought into it, I figured I’d give it a try. I hadn’t written a word on the book in well over a year, but on June 7th, I started a new chapter on Game Loops. I wrote 777 words of first draft, less than an hour of work.

The next day, I wrote 489 words. Then a meager 178. But it wasn’t zero. This weird desire not to break the chain made me drag myself off my ass and at least write something, and once those 178 words were done, I was that much closer to the end. The next day, I did 889 words. The day after that, I finished the first draft of the chapter.

Interlude: How a Chapter is Made!

I got so wrapped up in my own little narcissistic narrative I forgot to tell you anything useful about the actual mechanics of how I write! Since I decided to post chapters online as I completed them, I knew I couldn’t just do a first draft of the whole book before revising anything. Trust me, no one wants to read a first draft. Instead, I treat each chapter like a little standalone piece of writing. I do it like this:

1. Outline the chapter

Most of the chapters have the same top level structure, but within those headings, I do a pretty detailed outline of what points I want to cover and how they flow together. By the end, I have basically all of the material for the chapter, just with no actual punctuation or grammar. It’s everything I want to say, said really poorly.

Here’s a chunk of outline from the Game Loop chapter:

- old programs used to be batch: ran and quit
- then came interactive programs. these ran forever, which meant loop
- if you write gui app, the os has event loop
  - you receive ui event, handle it, and return to event loop
  - js is a great example
  - you don't see loop, but it's there
  - when app isn't handle user input, it isn't doing anything, just sitting
    there
  - this can work for simple games (minesweeper, solitaire)

I don’t think those bullet points or indentation even mean anything. It’s just word salad.

2. Write the code

The outline will hint at the example code so I know what blocks of code need to be written. Then I come back and write the actual code. Sometimes I’ll interleave this with writing the text, other times I’ll do all of the code up front.

I write the code in separate C++ source files so that I can compile it and make sure it’s free of errors. For some chapters, I even wrote unit tests. The script that converts the Markdown to HTML pulls in the code snippets directly from those files.

Here’s a bit of markdown for the prototype chapter:

To create a ghost spawner, we just create a prototypical ghost
instance, and then create a spawner holding that prototype:

^code spawn-ghost-clone

One neat part about this pattern is that it doesn't just clone
the *class* of the prototype, it clones its *state* too...

The little ^code tells my script to hunt in the C++ code until it finds:

void test()
{
  //^spawn-ghost-clone
  Monster* ghostPrototype = new Ghost(15, 3);
  Spawner* ghostSpawner = new Spawner(ghostPrototype);
  //^spawn-ghost-clone
  use(ghostSpawner);
}

It stitches in those two lines of code and we’re good!

3. Write the first draft

I hate first drafts. I hate the blank page. I do a detailed outline up front to try to make this easier. And then I rush through this as quick as I can to get it over with. Even so, this part takes the longest and I usually only get about 500 words a day on it.

I try to silence my internal critic during this phase so I can keep making forward progress. I rely heavily on later revisions. Knowing I will edit later gives me the freedom to not edit now.

4. Do the second draft

As soon as the first draft is done, I circle back and do the second. This involves fixing all of the mistakes, bad grammar, and poor flow of the first draft. Sometimes, there’s major surgery here when I decide something I crammed in just isn’t working. I make a lot of changes:

A chunk of the diff from the first to second draft of the last chapter I wrote.

The second draft is shorter than the first. Writing to me is like pottery. I slap all the clay on the wheel in a big blob and then gradually work it down to the final piece. I can get about 1,000 words of this done a day.

I love this part. Feeling the prose get tigher and clearer is exactly as satisfying as refactoring messy code. It’s less stressful because I feel like anything I do now just makes it better but the chapter is safely complete regardless. There’s no more scary blank page, so editing is just pure goodness.

5. Do the third draft

This is the home stretch. At this point, the chapter is almost in its final form. For this last pass, what I’m looking for is mistakes and rhythm.

I read the entire chapter out loud. I can’t stress enough how helpful it is. You can read something ten times and think it’s fine but the first time you run it through your lips you’ll find all of the wrinkles. This fixes awkward repetition and bad cadence. If the prose reads naturally and easily, that’s not because it was easy and natural to write. It’s because I edited the hell out of it.

6. Illustrations

The last thing I do is draw a few illustrations for the chapter. I think visuals are hugely important for making abstract concepts concrete in the user’s mind.

I spent a lot of time trying to figure out how I wanted to illustrate the book. I’d done a few illustrations in Photoshop, but didn’t like how the text looked pixellated. It also made it a real pain to change the design of the site, and looked terrible when retina displays came on the scene.

I thought about using SVG, but that seemed like a huge time sink to get the level of quality I wanted. Eventually, I hit upon the idea of going in the other direction. I decided to hand-draw and scan the illustrations. I could scan them at a high enough resolution to look sharp on a retina screen. More importantly, their intentional imperfection would help turn off the OCD part of my brain that would be unsatisfied with anything less than perfect Edward Tufte-quality diagrams.

Some of the 66 hand-drawn illustrations.

7. Publish and fix bugs

At this point, I think the chapter is done. I put it online and tell the world, and immediately they point out a bunch of glaringly obvious bugs that somehow survived several rounds of editing. I fix those, and now the chapter is mostly solid.

The Long Road

OK, where were we? I started logging each day’s work in a little text file. Less than two weeks later, the chapter was done:

2013-06-18 - Finish third draft of Game Loop
2013-06-17 - 1,280 words revised in third draft of Game Loop
2013-06-16 - ~1,000 words revised in third draft of Game Loop
2013-06-15 - Finish second draft of Game Loop
2013-06-14 - Revise couple of paragraphs of Game Loop
2013-06-13 - Revise ~500 words of second draft of Game Loop
2013-06-12 - Revise ~900 words of second draft of Game Loop
2013-06-11 - Finish first draft of Game Loop
2013-06-10 - 489 words on first draft of Game Loop
2013-06-09 - 178 words on first draft of Game Loop
2013-06-08 - 889 words on first draft of Game Loop
2013-06-07 - 777 words on first draft of Game Loop

I spent some time redoing the design of the site not to be totally busted on mobile devices and then started the next chapter. Now, if this was a movie, this is when the eighties instrumental rock would start and the montage would kick in. Before you know it, the book would be done.

But, I tell you what, I lived that montage, and it did not pass quickly. Working on this damned book was a pain in the ass every single day. I work full time, and have two small kids. We bought a house and all of the work that that entails and it turns out that, holy crap, it’s pretty easy to fill an entire day with stuff that’s not writing a book.

I started getting up early in the morning before the kids so I could have some quiet time. Ostensibly, I would write then, and sometimes I did. But, more often than not, I’d just drink coffee and stare at the Internet while I slowly roused myself. Then, after I’d worked, made dinner, gotten the kids cleaned up and ready for bed, read them books, gotten them milk, cleaned up a bit and gotten them in bed, then, I’d still have to write.

Yet, strangely, I would. I’ve never demonstrated an ounce of self-discipline in my entire life, but for some reason I just didn’t want to skip a day. Even though I was often dog tired, I would scrape up just enough energy to put in a barely half an hour of writing. I would look at the pathetic word count and feel bad before going to sleep exhausted.

But here’s the thing. Even tiny amounts of effort add up if you keep at it. I’d have entire weeks where I felt like I didn’t have a single good writing day, but after two of those weeks… that’s a whole chapter.

Because I was writing every day, it made it easier to build it into the routine. I never had a chance to get used to not writing. It was just this thing that must be in the schedule, like brushing my teeth or eating, even when I traveled or did outings with the kids. I gave a talk at Strange Loop last year, and every evening after dinner, while others were out at bars socializing, I went back to my hotel room so I could write. When I flew to another office for all-day meetings, I wrote on the plane.

The only insurmountable problem I ran into was a camping trip. I couldn’t imagine bringing a laptop to the woods and charging it in my car or something. Instead, I did what comic strip artists do: I banked some days. For two days leading up to the trip, I wrote two sessions a day. Then I wrote early in the morning before leaving, so I only ended up skipping one day. I basically loaned myself the time, with interest.

Refusal of the Return

Once I started putting chapters online again, people started noticing. I got that feedback and encouragement that I thrive on and I was delighted to have stuff to talk about on reddit again.

Around this time, an editor at another publisher reached out to me. Now that I was working on the book again, would I be interested in a publishing deal? She was super nice and was very willing to work with me. I could keep the book online, I could be involved in the design and layout. It could be everything I wanted and more.

I kind of dragged my feet for a while before I finally realized my reluctance was trying to tell me something. When I’d had publishers before, the power relationship felt really strange. O'Reilly and Apress were both great, but I felt like they were in charge, which is bizarre when you think about how much effort the author puts into it compared to the publisher.

What I was finding was that I really liked it being my book. I’m not doing this for the money, which means I’m doing it for my personal satisfaction. And what’s most satisfying to me is feeling like I got to put as much of my own creativity into it as possible without someone else calling the shots.

That’s not to say I don’t like collaboration. I’ve gotten about 130 bug reports along with a bunch of pull requests, not to mention lots of comments and email. I absolutely treasure all of that, and the book is way better than it would be without that.

I really like working with readers and contributors, but I’m not really interested in working for a publisher. About half-way through the book, I decided to self-publish. That meant I needed to start doing the things a publisher does. In my mind that’s:

  1. Developmental editing. When you get a book deal, this is what the editor you work with does. It’s their job to guide the overall direction of the book to make sure it lines up with what audiences want.

  2. Marketing, advertising, and distribution. Basically, getting the book in front of people and in stores. Making sure people know it exists.

  3. Copy-editing, proof-reading, illustrations, and design. The sort of blue collar stuff that turns a manuscript into a book.

I’d been putting chapters online and interacting directly with my audience since I started the book, so I felt like I had a better handle on what they wanted than most editors would. Having an inbox with hundreds of emails from people is pretty helpful when you’re trying to figure out what they want.

All of the manual labor side of things—editing and design—are some of my favorite parts. Before I was a programmer, I was a graphic designer, and I’ve dreamed of typesetting and laying out a book. Bug reports from readers helped do some of the work of copy-editing. (Special shout-out to mystery superhero colms who basically line-edited the entire book free of charge.) Finding a freelance proof-reader didn’t seem too hard.

Of course, distribution is a mostly solved problem now. Especially with technical books, people buy them from Amazon or get eBooks. I can do those just as well as any New York publisher. You don’t need deals or shelf space any more (though it certainly doesn’t hurt).

That left marketing and advertising—making sure people knew about the book. So I created a mailing list. I added a little blurb to the top of each page saying the book was a work-in-progress and to sign up if you want to know when chapters come out. As of this morning, I have a little over 3,000 subscribers, which is probably more reach than a “real” publisher would have given me.

Finishing the Damn Thing

Once I had that decided and squared away, all that was left was to write the rest of the chapters. So I wrote. And wrote. And wrote. Every single day. I wrote while my wife gave the kids a bath. I wrote at five in the morning before we went on day trips. I wrote when I was sick. I wrote on Halloween, Thanksgiving, Christmas, and New Year’s.

When I finished a chapter, I’d spend a few days where “writing” consisted of fixing bugs, responding to email and other minutia. But then I’d get to the next chapter.

After a while, writing the book receded into the back of my mind. It wasn’t something I thought much about during the day. When people asked what I did outside of work, I’d forget to include it sometimes. It was just ever-present background noise. Exactly 322 days after I started the first draft of Game Loop, all of these little slices of low thread priority background work on the book added up.

95,688 words. 21 chapters. 66 illustrations. 133 fixed issues. Hundreds of commits. Yesterday, I committed the third draft of the last chapter and the manuscript was complete.

I’ve wanted to be able to claim that I’ve written a book for so long that it feels weird to finally be on the other side of finish line. I wrote a book. Not “want to write”. Not “will write”. Not even “am writing”. Wrote.