The BYOND Cartridge Classic '05
In July 2005 I proposed a game contest: Create the best game in 8K of source
code. Officially published libraries were allowed, plus any files used by
DmiFonts would not count. The goal was to recreate some of the glory of old
cartridge games that had only 8K to work with. Because of that, I called it the
Cartridge Classic. The top five entries were to be awarded prizes, out of my
own cash hoard.
Prizes are as follows: $50 for 1st place, $40 for 2nd, and so on down to $10
for 5th. The top 5, plus any honorable mentions, receive a subscription to my
The original deadline for the contest was October 31, 2005. However when
BYOND started up a contest of its own that month, the Cartridge Classic got
rather buried. A deadline extension to December 31 helped somewhat, but because
the holidays Christmas and the new year, a lot of people
accidentally missed the chance to submit their entries. Finally this was pushed
back to January 15, 2006. By the end of that time there were 14 entries, a
healthy number. 11 of those entries qualified for judging.
Judges included myself, digitalmouse, IainPeregrine, splattergnome, and
Unknown Person. I asked judges to score their entries on four criteria:
- Gameplay: Smoothness or fun of play, and the general gaming
environment were the factors here.
- Presentation: Icons, sound, and even the look of statpanels and
help files applied to presentation.
- User-friendliness: Ease of control, ingame help, and so on
contributed to this score.
- Overall: The judges' overall impression, which could be
independent of the other factors but in practice was often related.
There was a lot of variety in the judges' results. Some judged more on
potential than on what was actually there. Some were very generous to some iffy
entries, while others were a little more demanding across the board.
Final scoring was done by taking a root-mean-square average of the
individual scores, which tends to weight scores slightly in favor of higher
marks vs. a direct linear average. It also has the nice effect of making ties a
little harder to come by.
I was rather impressed that a lot of entries remembered to take care of some
important basics. Ingame help was present in most of the entries, and a couple
of board game entries included player queues. The very different styles of the
games made judging interesting, and competition for the top 5 was strong. The
layout of the final winners was a bit different than I expected.
A few minor notes: Some games were mispackaged as libraries or as games
rather than DM demos (pure source code), but this didn't impact their results.
Polaris8920's entries had some trouble compiling, so splattergnome was unable
to get results for those.
Now, let's get to the results!
Concept: A TV show contestant is put through a jiggly gelatin obstacle course.
I really dug this game. Gello Jumper was a combination of fun gameplay with
great use of icons and music. Control could be a little frustrating at times,
and there were a couple of minor glitches that could use some work, like that
there's no way to reset to a checkpoint after getting stuck. I wasn't too
thrilled with the alert() boxes at the beginning, but they were easy enough to
plow through. Mostly though, the judges' major complaint was only that it was
too short. More levels would have been nice.
Interestingly, this game can be played multiplayer. It's actually rather a
blast that way.
Since no compression was done on the code, there'd actually be room to add a
whole lot more features, perhaps new types of obstacles. Picture this game with
100 levels, a little smoother movement, and some more creatures to avoid, and
you can really see the potential of an expanded version.
Gello Jumper had a good shot at 4th place but was just barely edged out; the
early scores made it look like a strong 3rd place contender. More than any
other entry, though, this one delivered the retro feel I was hoping to see. As
sponsor of the contest I awarded an additional $5.
Total prize: $10 + $5
Concept: 2-player strategy moving a knight on a 7x4 board from a random starting point until one player is blocked.
Knight is a simple concept, executed well. Overall I was impressed by how
much thought went into the game interface, including the use of a queue system
so players could spectate in between matches but still be sure to get a chance
to play. A game log records each game.
Judges mostly agreed that the board needed some more work. I thought in
particular it could use some edges, and labels for the columns and rows (which
are used in the game log). Alternating colors on the board might have been
nice. This entry, like Fallout, also suffered from the "the name" effect with
In general Knight did so well with what it had that it just barely beat
Gello Jumper to land in 4th place. It was clear from the outset of judging that
this was one of the better entries in the competition. Well done.
Total prize: $20
Concept: 1942, with dragons!
This was one of the best action entries I saw, recalling the overhead
skyscrollers of the '80s and early '90s. The icons looked great, and I
particularly liked the use of sound effects and music. The music did have a
minor bug, in that it's played on a channel used by a different effect, and on
newer BYOND versions that means the music will stop at that point. To
compensate to get the full experience I had to recompile with a change. There
was a little trouble for some people with the player's dragon disappearing
temporarily. Picking up items was a little difficult; gameplay would probably
be enhanced by allowing any touch (wings, etc.) to pick up an item.
Controls are decent but could be better, particularly with the attacks.
Actually there's no help file whatsoever included, which made the attack keys a
bit confusing to figure out; judges had to look into the source to figure out
the special attack key (down). Still, given the author's grammatical
proclivities, maybe the lack of a help file was a good thing. ;)
Surprisingly this game was also not slow; digitalmouse had the guts to do
his testing under an emulator, and SkyWurm 8k ran fine for him. I too had no
problems with performance. Some judges did point out that some more variation
in the levels, opponents, etc. would be nice, and that some concept of stages
or progress would add to the gameplay. I can but agree, but I still found this
enjoyable and addictive.
SkyWurm was really something special, definitely the most action-packed
entry. It's a great anytime sit-down-and-play sort of game.
Total prize: $30
Concept: Puzzle game based on the classic. Move cars around till the green car can escape. Includes editor.
This was a great BYOND version of the puzzle game, even including an editor
so you can make your own levels. The icons were good, although I felt the cars
came off a little 16-color and would have liked to see something with a wow
factor. The game makes decent use of audio at the end of each level, but more
sounds would have been nice.
The help file was a sticking point for some, since it is not immediately
apparent and kind of hard to find. The Help command is there, but to realize
that you basically have to hit the tab key to see a list of verbs. Once you get
past that, though, the game is pretty intuitive, straightforward, and fun.
Traffic Jam was the only true puzzle entry of the contest, but it's hard to
imagine it would have competition otherwise. It was pretty decently polished,
and as such was one of only two entries to pick up a 10 in any individual
Total prize: $40
Concept: Checkerboard game of strategy. Break through to the other side first.
A huge standout in terms of professionalism--and that with no compression
done on the code--Breakthrough shows just how effective a simple concept can
be. The board game may be familiar to some, since it's a chess/checkers
variant. Each piece is identical; they may move forward straight or diagonally
by 1 square, and capture diagonally.
Breakthrough was the judges' general favorite, earning rare 10s in some
categories. It's not hard to see why. The gameplay is simple and it's freaking
addictive; the use of a player queue was a good thought. In presentation
everything was well done, from the graphics (including woodgrain on the
board!) to the audio, and touches like a running game log were not overlooked.
User-friendliness was also far above average.
I discovered one minor runtime error that occurs in certain rare situations,
but it turns out not to affect the game in any way. Overall this is a solidly
made game and a great deal of fun.
From the moment judging began, Breakthrough had star quality. It was an
uphill battle for any of the other entries to catch it, and with three 10s on
its scoreboard none of them could. A perfect example that more is more when it
comes to presentation and features, fleshing out a basic easily-understood
game, this one made the most of its 8K allotment and the competition as a
Total prize: $50
Concept: You're one of the few non-zombie stick figures left. Gun the zombies down with your weapons and escape the notebook!
This was a fun simple game after my own heart. Rogue Stickz can be played in
two ways: A single-player escape from the notebook played on multiple pages, or
a multi-player deathmatch on a randomly-generated map. Players have multiple
weapons available. Good work was done with audio overall, and I thought the
graphics fit the theme really well.
There were a few places for improvement. Reloading took a long, long time.
In single-player mode the missile weapons were dominant but in multi-player
they were basically useless. I also would have liked for the missile's mouse
cursor to allow you to select any place on the map, not just a stick, because
sometimes hitting the stick exactly was a pain. It was sometimes difficult to
tell if you had correctly initiated a rocket launch or not. Missiles could also
still be dropped after a player's death. AI was a little simplistic, but I did
like how stickz would jump over your bullets to dodge them.
Overall this felt like it might need a tad more development to really finish
it off. Still, it's pretty fun as-is.
A contender for the top 5, Rogue Stickz fell behind in the stiff competition
between Gello Jumper and Knight. However, it's a pretty well done game and
earned some recognition for that. Per my own discretion, I'm awarding a $5
prize for this entry.
Total prize: $5
Concept: Collect guns and ammo; be the first to earn 25 points by shooting your opponents.
A potentially awesome but half-finished game, Airsoft was a case study in
potential. (For my part, my ratings had more to do with that potential than
what was there.) The ultimate goal of this game can still be realized: a fun,
fast-paced deathmatch between a large number of players. It's clearly designed
for a big group.
So where did it go wrong? Well, there were runtime errors when non-gun items
tried to respawn; as a result they did not reappear. The movement system is a
bit hard to get used to, between turning/running with the keyboard and shooting
with the mouse. (The pixel-based shooting, however, is quite cool.) A bigger
problem was switching weapons, which required clicking on the weapon in the
statpanel. Some parts of the map also didn't work right, like a bridge in the
southwest corner. Audio was nonexistent. Final compression was never really
done on the game, so a lot of this could have been addressed.
The help file also could have used some more detail, including spelling out
various weapon strengths and effects. Some random level generation could have
been cool, or perhaps just a bigger map. Another gameplay issue that should
be looked at is the fact that players can simply stand in one place and click
until somebody dies; given the movement system that's actually the easiest way
What I liked most about Airsoft was the HUD. This was shockingly
well-designed. A lot of attention went into this part of the interface; it's
just unfortunate some other portions suffered as a result. The icons were also
very good, though as ever I could find room for improvement: The stream could
use some autojoining icons and some more tree types would be good.
I'd really like to see this game brought up to its full potential and
hosted regularly on the hub. It had the makings of a contender, but it ended up
kind of half-finished.
Concept: Fall without running into blocks. Get powerups, survive longest. Primarily designed for multiplayer.
A sort of upside-down version of The Sky Is Falling, Fallout is based on the
idea that you're falling from a great height and want to outlast the other
players. There are powerups you can use to do so. The game overall is decent,
but needs work. In multiplayer, some judges found that players didn't interact
very much after the beginning stage, when they would become more and more
separated--kind of the opposite of TSIF where obstruction is the dominant
The icons were okay, but probably should have been better, especially the
black space to either side of the screen where we could have had the sides of
other buildings or something colorful. Heck, awnings and clotheslines and the
like might not only have added to the visuals, but might have made interesting
game elements as well. Multiple player icons were also a must, I think. And
splattergnome pointed out that this game fell victim to a common mistake new
authors make, in that they use "[src]" or "[usr]" to display
a name so it ends up saying something like "the splattergnome" for lowercase
There was universal agreement that the startup sequence of hitting the left
arrow, right arrow, left arrow again, etc. was ridiculously complicated for no
reason. Even though it was documented it still confounded gamers.
There was a lot of room for additional features, since Fallout wasn't
compressed at all. The source code was mispackaged as a game entry on the
Concept: Machines and clones battle for dominance.
Combat Proliferation had a good concept but failed in execution. The most
widely noted problems were that maps took forever to load, and when they
did, no map icons appeared. The latter problem was no doubt the result of a
packaging error. The former was caused by reliance on Polaris's .Txt Parser
library, which apparently is horrifically slow. No visible map made gameplay
difficult, but there were also several runtime errors to contend with.
IainPeregrine summed this one up nicely: "Though I have to give the game this
score based on my experience, I would love to have the chance to play it as it
was meant to be. It looks like it could have been very fun as a multiplayer
This game also takes the cake for being hardest to compile. It was
mispackaged as a library, and also had no instructions regarding which
libraries would be needed for the final result. Some of Polaris's libraries
include each other indirectly, which confused issues further.
Concept: You're nauseas [sic] and need to eat pills to feel better. Eat the required number of pills, then move to the next level.
Most of the judges agreed this made a good first effort, but needed more
actual gameplay elements to make it interesting. The addition of crude enemies,
perhaps other types of challenges, would have helped a lot, but nothing was
invested in making this more fun. I liked the fact that the maps were randomly
generated, which was a plus. The game also had an interesting retro feel;
splattergnome said he liked the way the tall screen enhanced that effect.
As several judges pointed out, the game didn't give any indication of
elapsed or remaining time, which would have been helpful, nor did it say how
many pills remained to be eaten. A timer would have given the game more
much-needed suspense. Little details like this would have improved some of the
scores slightly, but for greatness this game needs more action.
Concept: Classic Pong
Sometimes simplicity can be its own enemy. So it was with Pong, which was
presented in a disastrous minigame format that made the screen difficult to
impossible to see. (The screen shot you see here does not expand to a larger
version. That's it.) Most judges scored it low on presentation as a result.
Sounds were also nonexistent.
No real gameplay was involved because there was no way to change the ball's
direction as in Pong. The computer opponent is therefore also perfect;
IainPeregrine said it may as well have been a wall, since it only gave the
illusion that it could miss. The scoring is also just to 1 point, which is
rather silly; whoever misses first loses the game. I had an even bigger problem
with the controls, which really highlight how BYOND isn't designed for arcade
games and this should have been handled with something else like a mouse. In
his review digitalmouse agreed: "Keydown/keyup events are needed for a game
like this. Controls were difficult to get used to."
In summary: It may or may not be possible to redo the minimalist classic
Pong in BYOND, but if it is, it isn't via keyboard controls or on a
4×1-tile screen. With no scoring system and no means of affecting
trajectory and speed, the game is just a challenge against yourself to see how
long you can go without missing. It actually has fewer features than the
I have such mixed feelings about this entry. Classic Zelda was disqualified
twice over for two important reasons. First, its total code size was 8200
bytes, which is 8 bytes over the legal limit. That was just tragic because the
code wasn't compressed at all; it contained a lot of useless whitespace. The
worse issue, however, was that this was just a playable demo; there are only
two screens, three if you count the cave. This wasn't a complete game.
What I did like about this entry, though, is that it raises two interesting
questions: Can Zelda be recreated more or less faithfully on BYOND? And if so,
can it be done in 8K? I suspect the answers are yes and probably, respectively.
It'd obviously take some mad technique to cram the code into 8K, but I think it
just might be feasible for someone with that skill.
Prehistoric Hunter was, like Classic Zelda, disqualified for being a demo
only. It's a sidescrolling game with just one screen to get through, which you
can do in under 2 minutes. After that it thanks you for playing and that's
Graphically this game had very interesting presentation, including some nice
icons and health bars over each creature including the player. Gameplay and
interface-wise, it had some issues. The jumping system, for instance, jumps
straight up but then falls diagonally forward, a rather more frustrating
departure from physics than the typical "control your jump mid-air" method.
Enemies were also very difficult to kill, and had little variety. Although
there were supposed to be some items to boost your health, I never found
What was Racer? Was it a demo like the author's other two entries, or a full
game? It never compiled, so I have no idea. The original submission was sent in
with all three entries packaged together, which was an obvious no-no. When
FinalFantasyFreak repackaged and resubmitted the entries individually at the
13th hour, Racer had no working download link, and even a manual link provided
in the hub entry didn't work. So while I know the source code is only about 2K
for this game, it was never submitted in an easily compiled form even though it
had two chances to do so, one of them after the submission deadline.