Things I've tried: Becoming an Indie Gamedev

Work area (1/10)

Back around the time of the competition, I was still living in my childhood home with my mother. My 108sqft bedroom was where I built Mad Kaz. I prepared it as a work area by duct taping my mother's thick faux-fur blanket to the walls bordering the room's only window. The intention was to minimize distractions by blacking out the room. There was a queen sized box frame bed with its head pushed up against the wall with the window. Butted up against the middle of the wall to the window's right was an IKEA HEMNES desk. On top of the desk was my PC tower and a single flatscreen monitor.

Except for food and washroom breaks, I spent the entire four months locked in this room building Mad Kaz.

Test devices (2/10)

The competition launched in late February but I didn't enter until early April. Given the time crunch, it didn't make sense to place an online order for an unlocked KRZR and sit around waiting for it to arrive. Instead, I went to a local store selling imported mobile phones and purchased the closest device I could find: The Motorola SLVR. The SLVR and the KRZR had similar features, but the SLVR was under-specced. My thinking at the time was that if the game ran fine on the SLVR, it would run fine on the KRZR.

Alongside the SLVR, I also purchased a Samsung SGH-D520, just so I could have another device to test with. The shop owner cut the tax off the price because I bought two phones and paid cash.

	KRZR	SLVR
Java version	MIDP 2.0 / CLDC 1.1	MIDP 2.0 / CLDC 1.1
Java heap size	939kb	800kb
Java record store max	512kb	512kb
Persistent memory	20mb	11mb
Display	1.9" TFT, 176x220, 256K colors	1.9" TFT, 176x220, 256K colors
Bluetooth version	2.0	1.2

Origins of the engine (3/10)

Before coming up with Mad Kaz, I was planning to make an entirely different game. My original game idea was a social top-down 2D MMO with mini-games. I created documents mocking out how it would look and how I would go about building it.

I ditched the MMO idea early on because, even if it all went according to plan, development would have stretched well past the competition's deadline of early August. What I didn't ditch were the game engine components highlighted in the work plan. I figured that many of them would still be needed regardless of the game idea I finally settled on.

Before ultimately deciding on Mad Kaz as the game to build, I spent a week building a font renderer, half a week on a Huffman compressor, three weeks on a virtual machine and assembler, and one week on image transformations and real-time audio mixing.

Origins of Mad Kaz (4/10)

Mad Kaz was intended to be a reimagining of the games NS-SHAFT and NS-TOWER, blended into a unified narrative experience with levels. Each level was to be unique in terms of graphics, audio, and gameplay mechanics. I came up with a story arc along with a list of levels. The title character, Mad Kaz, would climb up from inside the Earth and eventually out into space, before encountering an alien boss battle and falling back down to Earth.

Level Num	Name	Type	Description
1	Internal Core	Climb	Hot wavy screen. Stepping tiles look like they're so hot that they're almost molten. Background includes a cavernous view of the earth with magma shooting out of the edges, etc.. etc..
2	Below The Crust	No-Drop Climb	Stepping on clay pieces. Nothing to exciting about this zone, covered in a repetitive dirt view.
3	The Surface	No-Drop Climb	Earth's surface. Building windows, etc... Shake every now and then to symbol a missile hit and flash screen white. Possible rain as well?
4	In The Sky	No-Drop Climb	Zoomed out view of city and buildings in ruins, and missiles piling into buildings. Lots of Wind fucking with your movements. Hard time going left/right depending on direction of the wind.
5	Above The Chaos	No-Drop Climb	Dusk sun going down. Dark black clouds underneath. Nothing too special.
6	The Horizon	Low-Grav No-Drop Climb	Watching sun go down on earth's horizon Nothing special about this one, just an introduction to low-grav jumping.
7	Approaching The Ring	Low-Grav No-Drop Climb	Everythings really dark so it's hard to see where you're jumping. At some point earlier heat vision googles might have been picked up which could be used here.
8	Inside The Ring	No-Drop Climb	Alien ships flying back and forth and general alien ship look.
9	The Boss	Fast Drop	Hard drop down a corridor. The boss does a lot to mess you up, including blanking out your screen for seconds at a time. There is an electric beam coming down that you must avoid in this hard drop. Must press action button on specific tiles to damage the enemy
10	The Escape	Fast Drop	The laser beam is still following you even after the boss is dead, you must outrun it.
11	It All Comes Down	Fast Drop	Service tunnel you came up in is collapsing. Stay ahead of collapsing debris.
12	The Aftermath	Free Fall	Free falling back to Earth. You'll see pieces of the tubes used to draw magma from the earth falling down in the background in a hail of fire. Avoid hitting debris when falling.

Mad Kaz's art (5/10)

To help piece Mad Kaz's levels together, I contracted out to several artists. A chiptunes musician by the name of Xonic the Fox produced three MIDI tracks. A few basic concept art sketches were created by a high school classmate, William. Those sketches then went to Mykola, a pixel artist who agreed to do the important work of reimagining them as pixel art.

Closest surviving archive of Mykola's website (July 2007): Myk31.

Mykola's Mad Kaz uploads to PixelJoint: Upload 1 and Upload 2

- Danger Zone, original MIDI

- Road Warrior, original MIDI

- Solitude, original MIDI

Closest surviving archive of Xonic's website (September 2007): Technomancer.

Scaling down Mad Kaz (6/10)

There wasn't enough time to both complete the game engine and build the game's twelve levels. To try and make things work, I scaled down the levels into six zones. Each zone was to have multiple levels that re-used assets and had similar game logic.

Of those six zones, two were writeoffs. The first zone was a tutorial and the last zone was nothing more than an in-game credit sequence cycling through names and photos of friends and people who worked on the game. The four zones in the middle were the meat of the game.

Zone breakdown and analysis

Old emails from that time show that I wasn't keeping it together. I hadn't properly scoped out the game's development and dumped too much time into the engine's development. My blacked out bedroom also wasn't helping. It was impossible for me to tell if it was day or night. I was sleeping and waking at random times.

From: Mykola <m***@g***.com>
To: Kasra F. <k***@c***.com>
Date: Wed, 13 Jun 2007 02:49:06 PDT
Subject: Re[2]: Progress?

Man what the hell, don't get pissed.
look I told you I got sick with flu last week, I know you're pissed by it but
so am I - cuz there's nothing I could do about it.
I'm still not 100% but now I can at least use my computer normally so
things should be ok.

I'm absolutely not gonna drop out, that's for sure. Don't panick man!
Please.

NOW. on to the game itself.
I like both the music and the concept art. music is a good fit to the
game I believe.

I attached some gfx you requested first. some floor tiles + the player
sprite.

notes:

[...]

let me know what you think bout these :) and then I'm moving on to
more gfx.

Mad Kaz submission (7/10)

July 26th was when the last zone was completed. Bugfixes and tweaks ran through July 27th and 28th. I submitted Mad Kaz to the competition on July 29th. There were still 2 weeks remaining until the competition's deadline, but money was running tight and I was burning out.

Mad Kaz's fundamental flaw was that it just wasn't fun to play. It ultimately ended up becoming a game engine tech demo rather than a game, consisting of a mishmash of unbalanced and narratively disjointed levels that tried to shove in as many engine features as possible. In a perfect world, I would have spent those remaining 2 weeks play testing and polishing the game play as much as possible.

Mad Kaz source code and assets.

Mad Kaz game, converted by AI to be playable inside a browser.

I showed Mad Kaz to a few friends. Many of them didn't understand or care about the technical merits of the engine. It simply came off as a middling game. I also wrote my father, living in California at the time, letting him know what I had been up to during the four months I'd been unemployed. I don't think he fully grasped the situation. His response was that I would always be a winner to him.

From: MOE F <m***@s***.ca>
To: Kaz F <k***@h***.com>
Date: Tue, 31 Jul 2007 19:12:07 PDT
Subject: Re:

Hi Kasra,

I saw your game, You did a fantastic work, I am very proud of you. With
all hope and good luck to you you are the best, anyway you are a
winner to me.

Cheers!!!!

    Kambiz

----- Original Message -----
From: Kaz F <k***@h***.com>
Date: Tuesday, July 31, 2007 4:49 pm
Subject:
To: m***@s***.ca

> Hey Kambiz,
>  
> Here's what I've been working on for the past 3 or so months. I
> entered it into the Motorola competition a couple of days ago. It
> was designed to run specific Motorola phones and some other
> phones (Motorola Vxxx, Motorola SLVR, Motorola K1, Samsung SGH-
> D520, etc.. etc..), so you can't play it on your phone.
>  
> http://www.codeyielddevelopment.com/madkaz.php

Post-submission job hunt (8/10)

Back around that time, I was using a spreadsheet to track my bank balance. Money was beginning to run tight. The records show that I started the competition with around $7,500 to my name. By the time I submitted Mad Kaz in late July, I had around $2,500 remaining and was on a steep downward trajectory.

Date	Chequing	Savings	Combined
Mar 15	$436.93	$7,001.25	$7,438.18
Mar 30	$366.81	$7,001.25	$7,368.06
Apr 6	$366.81	$5,781.31	$6,148.12
Apr 9	$516.81	$5,631.31	$6,148.12
Apr 13	$29.65	$5,631.31	$5,660.96
Apr 15	$160.96	$5,500.00	$5,660.96
Apr 27	$1,708.62	$5,500.00	$7,208.62
May 6	$1,185.26	$5,500.00	$6,685.26
May 10	$876.76	$5,500.00	$6,376.76
May 30	$876.76	$4,698.41	$5,575.17
Jun 28	$1,876.76	$3,699.28	$5,576.04
Jul 11 16:17	$2,576.61	$3,000.00	$5,576.61
Jul 11 16:20	$539.04	$3,000.00	$3,539.04
Jul 17	$539.04	$2,000.00	$2,539.04
Aug 7	$1,039.04	$1,500.26	$2,539.30
Aug 14	$231.86	$1,500.26	$1,732.12
Aug 19	$631.86	$1,098.26	$1,730.12
Sep 8	$311.58	$1,098.26	$1,409.84
Sep 20 17:15	$311.58	$798.38	$1,109.96
Sep 20 17:16	$611.58	$498.38	$1,109.96
Oct 6 16:27	$611.58	$197.45	$809.03
Oct 6 16:29	$761.58	$47.45	$809.03
Oct 27	$592.09	$47.45	$639.54

I began looking for work again almost immediately after submitting Mad Kaz. I applied anywhere that was hiring, but I was also trying to leverage Mad Kaz to get hired at a real game studio. I ended up open-sourcing Mad Kaz in the hopes that a technically-inclined reviewer would look through it and come to the conclusion that I was skilled enough. I sent applications to four now-defunct studios: Radical Entertainment, Threewave, Blue Castle Games, and IUGO.

The application to Radical Entertainment went nowhere. Threewave had me fill out a questionnaire and scheduled a phone screen afterwards, but ultimately it didn't move forward either.

Blue Castle Games had me come in for an interview. They were hiring a junior network programmer for Dead Rising 2. It didn't go well. An interviewer had gone through Mad Kaz's code and made derisive comments in what felt like an unsuccessful attempt at shit testing me. Another interviewer had me whiteboarding how to model zombie movements, and when he jumped in with a question a competing thought entered my head and I inadvertently told him to shut up. The interview ended with a brain teaser that had me fumbling and freezing up (three-prisoners hat riddle).

From: Melissa S <m***@bluecastlegames.com>
To: 'Kasra F' <k***@***.ca>
Date: Mon, 10 Sep 2007 18:28:08 PDT
Subject: RE: Interview Next Steps from Blue Castle Games

Hi Kaz,

There will be two interviews - one from 10-11 and another from 11-12.D***
will be there, as will J*** (Technical Director), V***
(Programmer), and T*** (Sr. Programmer). I will meet you when you
arrive and get you settled in.

The interviews will cover a mix of technical and "soft" skill questions. I
can't tell you the exact questions they will ask, but I can tell you that
they will ask you to do some coding work on a white board in addition to
answering questions orally.

Hope that helps! I will send you the NDA and directions tomorrow morning.

Have a good night!

IUGO had me come in for an interview after I made a personal appeal to the CEO. It went fairly well. The atmosphere was relaxed. A college acquaintance was working there. One of the interviewers had me sit down with a stack of papers full of game coding and design questions. When I finished, he tossed them aside, saying something to the effect of "I don't need to look at it. I know you know your stuff."

The last part of the interview had me chatting with the CEO. He had either given me an offer or was on the verge of giving one, but I ended up walking away. I don't recall exactly why. It may have been because I was actively interviewing at Blue Castle Games, or I may have been feeling too full of myself and gunning for a more senior role, or it may have been because the office felt uncomfortably bright. It was unusually sunny that day. IUGO's office space let a lot of that sun in. I had spent most of my time building Mad Kaz in my blacked out bedroom.

From: Hong-Yee W <h***@iugome.com>
To: 'Kaz F' <k***@h***.com>
Date: Wed, 05 Sep 2007 13:57:30 PDT
Subject: RE: Application for J2ME/BREW Programmer

Hi Kazra,

Let's try this again. Please come in for an interview tomorrow (6 Sep) at
11am. Our office address remains unchanged - see below.

[...]

From: Kaz F [mailto:k***@h***.com]
Sent: Thursday, August 02, 2007 3:20 PM
To: h***@iugome.com
Subject: RE: Application for J2ME/BREW Programmer

Hey Hong,

I'm just getting in touch with you to give you an update on my status. I
gave in my resignation at AirG a little while after initially sending in my
resume to IUGO so that I could take part in the Motodev game developer
challenge. I made sure to incorporate before I made my submission (strictly
for liability reasons). You can find the game and source code (GPLv3) on my
website. If any other positions at IUGO open up that you think I'm qualified
for, don't hesitate to give me a shout.

I've attached my resume in case you want to glance it over again.

- Kaz

From: h***@iugome.com
To: k***@h***.com
Subject: RE: Application for J2ME/BREW Programmer
Date: Mon, 20 Nov 2006 13:37:58 PST

Hi Kasra,

Please come in for an interview tomorrow (20 Nov) at 3pm. Our office is in
Yaletown: [...]

-----Original Message-----
From: Kaz F [mailto:k***@h***.com]
Sent: Sunday, November 19, 2006 5:51 PM
To: h***@iugome.com
Subject: Application for J2ME/BREW Programmer

Dear IUGO HR Department,

This letter is to express my interest in discussing the mobile game developer position posted on the IUGO website. [...]

In the end, money was running dry. I had around $800 left in the bank and so I accepted a job at a company that made calendar software. The salary was around 15% lower than that of the job I resigned from to enter the competition. Weeks after starting, I revisited the MOTODEV Game Developer Challenge website. Mad Kaz hadn't won, nor was it one of the finalists. I never heard from the people running the competition about their thoughts on Mad Kaz or how it ranked against other games in the competition.

Engine postmortem (9/10)

Mad Kaz's engine was the culmination of my experiences with gaming and coding, starting from early childhood experiments on the first family PC my father purchased in the early 1990s all the way up to the job I resigned from to enter the competition.

As a game, Mad Kaz failed. As a game engine, it succeeded at doing something few at the time had attempted: It was a programmable UGC platform for the resource-constrained mobile devices of that era. The individual components of the engine weren't all that unique or original. Doom had already popularized pack files, Quake had already popularized scripting, QuakeWorld had already popularized in-game level downloads, countless games performed dynamic audio mixing and graphical post-processing, etc. It was the collection of those individual components together under a single umbrella, running successfully on a Java device with 800kb of heap, that made for something special.

Many of the individual components within Mad Kaz were born out of my experiences at my former employer. The R&D head at the time was forgiving enough to look past my attitude problem and let me work under him. He made several attempts at a scriptable platform for their mobile apps. Those scriptable platforms, along with Java's virtual machine and Quake's virtual machine, became the inspiration for Mad Kaz's virtual machine and assembler.

Mad Kaz's PNG palette editing module was inspired from overhearing a watercooler discussion one day at work. A senior engineer mentioned how someone they knew wrote code to go in and edit a PNG's palette at runtime. After hearing that, I visited a now-defunct website known for indexing file format specifications, pulled out the specifications for PNG, and spent a few days at work trying to come up with interesting things to do with palette modifications.

Mad Kaz's image manipulation functions were inspired by a college classmate. While sitting in a college computer lab one day, he wrote code to resize an image using bilinear scaling. We didn't know that it was called bilinear scaling at the time, but his idea of pixel-to-pixel gradients was the seed for Mad Kaz's image rotation, scaling, and post-processing effects.

Mad Kaz's image manipulation functions required rational numbers, but most mobile phones at that time didn't include a hardware floating point unit. Somewhere along the way, I had learned about fixed-point math and lookup tables. I used fixed-point math to implement a progress bar in one of my employer's mobile apps, and I also used it to perform the image manipulation math in Mad Kaz.

The list goes on. It was all prior art that, at some point or another, I had experienced and internalized in some way.

• Apr 3: Empty repository.
• Apr 11-14: Built Hershey font conversion and rendering, then added word wrapping, resizing, and Huffman compression.
• Apr 21-May 3: Added an assembler and VM instruction set, generalized the VM, and built J2ME benchmarks and runtime support.
• May 4-7: Added WAV mixing and image/PNG tools, palette replacement, graphics management, and tiling.
• May 8-12: Added game canvas, sprites, scenes, music, logo animation, and text handling; adjusted loading and key-repeat behavior for phones.
• May 13-31: Added menus, windows, dialogs, persistent storage, and image rotation; refined text layout and fixed device/input bugs.
• Jun 2-4: Added Micro3D-related tools and scrolling/physics code; comments note unfinished gravity and offset behavior.
• Jun 4-14: Added the first large Mad Kaz implementation, including its VM, gameplay, collisions, and image/resource handling. Later changes renamed the play canvas and optimized performance; one replaces many StringBuffer UI values with char[].
• Jun 14-19: Expanded game-specific VM operations for state, input, effects, and timers. Added custom-map UI and began network access.
• Jun 20-28: Added network map downloads, local storage and browsing, and tools to package downloadable levels from XML. Extended gameplay controls, sprites, and graphics parsing.
• Jul 1-3: Prototyped Bluetooth multiplayer and menus, then added a credits sequence and gameplay/timing fixes.
• Jul 5-10: Added an automated-scene engine and tests, then connected scene behavior to the game and SDK.
• Jul 12-21: Added configuration storage, level selection/unlocks, and substantial zone and level content.
• Jul 22-27: Removed the unplayable multiplayer path, added peer-to-peer custom-map transfer, and optimized level packaging with an indexed combiner. Reorganized zone 6 as zone 5.
• Jul 27-28: Revised setup, credits, copyright/license text, and packaging; removed map compression; reorganized into trunk/tags and created MOTODEV/alpha snapshots. Added a downloadable "Space Quest" map.

Incorporating (10/10)

One unimportant but mildly interesting fact about Mad Kaz and the competition was that I incorporated a company to enter the competition with: CodeYield Development Inc. I contracted out the company's website design to a high school friend and got myself a fax number. For some reason, I thought that people wouldn't take the company seriously unless it had a fax number.

Closest surviving archive of CodeYield Development's website (February 2011): CodeYield Development.

Corporate registry notice.