Enemy scaling is the spine of every tower defense game. Get it wrong and the game is either trivial or impossible. This devlog walks through how we approached scaling in Lost Tower, what we tried, what failed, and what we shipped.
On this page
Enemy scaling in Lost Tower uses a tiered approach: HP and damage grow on staircase curves with explicit cliff points at boss waves. This replaces pure exponential scaling, which we tried first and abandoned because it made every late wave feel identical.
The problem
A tower defense game has to keep escalating challenge across potentially hundreds of waves. Linear scaling means late waves are trivial. Pure exponential scaling means late waves are impossible. We needed something between.
We also needed the scaling to interact with multiple progression layers (parchments, god ascension, base tower levels) without breaking. A player at base tower level 5 cannot face the same wave 30 as a player at base tower level 50.
Exponential scaling: what almost killed us
Our first implementation used pure exponential scaling. Every wave's HP was a fixed multiple of the previous wave. This had two problems.
First, late waves felt identical. The difference between wave 50 and wave 60 was unnoticeable when both were absurd numbers. Players stopped caring.
Second, build differences became invisible. Whether you had a damage-focused build or a survivability build, late waves erased the distinction. The strategy layer disappeared at high levels.
Tiered scaling: what we settled on
We replaced exponential with staircase tiers. Every 10 waves is a tier. Within a tier, scaling is gentle. Between tiers, scaling jumps. This means wave 19 is meaningfully different from wave 20.
The tier breaks also map to boss waves. Wave 10, 20, 30, etc., are bosses. This gives players a sense of "I'm finishing tier 3" rather than "I'm at some arbitrary wave."
The math: within a tier, HP scales at roughly 1.18x per wave. At tier boundaries, HP jumps by 1.55x. Across the run, this produces a curve that feels challenging without being absurd.
The boss wave problem
Boss waves were where the original exponential scaling broke first. A wave 100 boss had so much HP that the only way to kill it was specific damage types. We rebalanced bosses to test multiple aspects of a build: phase 1 is a damage check, phase 2 is a survivability check, phase 3 is a positioning check. See our advanced guide for boss-specific strategy.
What we are still tuning
Tiered scaling works, but tier boundaries are still rough. Players sometimes hit a tier and feel the difficulty cliff is unfair. We are smoothing this in B0.6 and B0.7. Balancing scaling is iterative work. It is also one of the most rewarding parts of designing a tower defense.
Next devlog: designing the idle progression layer.