Tower Balance: Designing a Seesaw Decision Game
The premise of Tower Balance is about as simple as it gets: a seesaw on a pivot, blocks with weights labeled 1, 2, or 3, and two buttons — Left and Right. Each block goes somewhere. The beam tilts toward whatever side has more total weight. Tilt it past 60 degrees and the game ends. I thought this would be quick to build and tune. It took three full iterations before the decision loop felt right, and the reasons why are the interesting part.
The First Version Was Too Easy to Cheese
My first pass used a binary split: alternating every block strictly between left and right would keep the scale at exactly zero tilt forever, assuming equal weights. In practice, even with random weights from 1 to 3, a strict alternation strategy kept most players going indefinitely. The expected value of a random block is 2.0, so both sides accumulate at the same rate on average. A player who simply alternates will hover around zero tilt with small fluctuations.
I added a weight-3 block that appeared more frequently in runs of two or three to break the alternation pattern. That helped, but the real fix was something different: I realized the interesting decisions only happen when the current imbalance is non-trivial and the incoming block weight is large relative to the gap. I restructured the scoring to make clear what you had to do — show the running totals on screen, not just the beam angle.
Making the Imbalance Visible
The original design showed only the beam tilt. Players had no way to know exactly how far the beam was from the 60-degree limit without guessing from the visual angle. Once I added numeric weight displays under each end of the beam, two things happened. First, players started making better decisions immediately — the numbers let them do the arithmetic. Second, and more importantly, it gave skilled players a way to demonstrate skill: they could correct an imbalance intentionally rather than by instinct, placing a weight-3 block on the heavy side even though that side was already higher, because they knew the specific deficit required it.
Hiding information doesn't make a decision game more interesting — it makes it more random. The numbers added depth because they gave players more to act on, not less.
The Countdown Timer and the Auto-Place Rule
Each block has a five-second countdown before it auto-places on the lighter side. I added this initially just to prevent players from pausing indefinitely before every block, but the auto-place rule turned out to be a meaningful design element. A player who lets the timer run out does not lose immediately — the block goes somewhere sensible. But they miss the opportunity to over-correct a growing imbalance. If the left side is 4 heavier and an incoming weight-3 arrives, a thoughtful player places it right to cut the deficit to 1. An auto-placed weight-3 also goes right, but the player wasted the decision — and there will be a block shortly where the auto-placement is actively harmful.
The timer created a low-stakes cost for inattention without a punishing penalty for a single slow response. That balance — the timer can hurt you eventually but doesn't end the run immediately — felt like the right kind of pressure for a puzzle game.
The Beam as Feedback
A lot of work went into making the beam angle feel honest. The tilt is computed directly from the weight difference — there's no animation easing, no momentum, no gradual approach. When you place a block, the beam jumps to its new angle immediately. I tried a smoothed transition first and it felt worse: the delay between action and feedback made it harder to judge where the beam was going to settle. Instant feedback is right for a game where the decision is the product, not the execution of the decision.
The beam color changes from white to red when the tilt exceeds 75% of the maximum angle. This is the only warning the game gives. It doesn't slow down, add extra time, or give a grace period — it just tells you that the next few blocks are critical. That warning window is short enough that it doesn't trivialize the recovery but long enough that a player who is paying attention can still save the run.
Play Tower Balance