Everyone here is familiar with the sugar-soaked juggernaut that is King’s Candy Crunch Saga. This “silly little puzzler” has been holding steady for months at a time as a top grossing title on Android, iOS, as well as Facebook.
Miska Katkoff, wrote a great analysis about how CCS has some of the best Mobile/Facebook viral and monetization design of any game out there. But while the elegance of those features makes the game long-lived and spendworthy, what makes the game itself so purely addictive is its near-perfect level design.
And when I say near-perfect, I mean neurologically and psychologically, viscerally and logistically, brain-in and balls-out, the near-optimal blend of challenge vs. release.
For starters, there’s something elementally brilliant about the Match 3 mechanic, first invented by Eugene Alemzhin as the DOS game Shairiki: The Balls. As with Tetris (man, what is it about those Russians?), the match 3 ruleset combines simple color, geometry, and gravity giving players the ability to work from a base of pure randomness and triumph or fail across a wide spectrum of possibilities, all the while flexing but never mastering caveman-brain skills of pattern matching, thinking ahead, and quick reflex.
I know a thing or three about match 3. iWin’s Jewel Quest is probably the 2nd most successful match 3 franchise, after Popcap’s masterful Bejeweled. In fact, many of the mechanics found in Candy Crunch Saga were first prototyped and invented by iWin’s game design savant Warren Schwader:
- Turning background tiles gold to win so that where you match matters as much as how quick you are.
- Different shaped boards, with hard to reach nooks requiring the clever pre-positioning of jewels.
- Gaps in boards create narrow, unmatchable corridors and channels.
- Unmatchable squares.
- Moving elements from top to the bottom by removing jewels beneath them.
- Special bonuses for matching four or five jewels.
- Special bonuses for matching horizontally and vertically at once.
- Etc. Etc. Etc.
But while Jewel Quest‘s level balance and design relied on the singular brilliance of Warren and other designers (along with a bit of level-reordering and time-tweaking based on results from a beta test or two), the creators at King have truly crowdsourced their balancing act, using metrics to be sure each level is barely solvable but increasingly tough. Numbers they are obviously looking at and tweaking constantly are:
- Number of failed attempts at a level before success.
- Number of moves made before success.
- How much failure is too much, leading to game abandonment.
- The blend of failure/success leading to the highest percentages of players returning and, ultimately, purchasing.
Using these metrics, they scientifically balance the difficulty and layout of each level so that most people are just one or two matches away from a win, inspiring the purchase of a few more moves, lives, or “get me out of any tough spot” candy hammers. The trick is to bring players mere paces away from the gates of heaven before plummeting them back down to the fiery abyss.
But is there more that King could be doing?
What if, instead of counting on randomness and aggregate stats, they came up with algorithms to tune each level and distribution of candy on the fly, so that the game intelligently reforms itself to bring players to the Golden Almost, the inevitable “just one more game” twitch. Why not pre-plan the layouts and new candy drop-downs to match each individual player’s style, progressively making levels either a touch more difficult or easy, in a way that reacts to that specific player’s strengths (good at looking ahead) and weaknesses (bad at speed). Like having your very own personal Product Manager.
Diabolical? Hell yes.
Fun and yet profitable? You know it.
Possible to do? Definitely. At least for a simple rule set like match 3.
It’s impossible for an outsider to tell whether King is already doing some more advanced heuristics like this, but based on my own frustrations of being stuck on some levels for waaaay too long, I don’t think so.
Then again, I am still playing, ain’t I? So maybe data-driven intelligence doesn’t even need to be that nuanced for most suckers.