Rearrange, Don't Reinforce: Surviving a World Boss Wave

Diagnose the survivor before you spend

Three enemy properties fail your defence in three different ways, and the fixes do not substitute for one another.

Armoured units survive because your damage arrives in pieces too small to matter. Rate of fire stops being the useful stat and per-hit damage starts being it. Siege and Cannon answers are relevant here in a way that faster small hits are not.

Flying units survive because they ignore the route your coverage was built around. This one is binary rather than gradual: either something can hit them where they pass, or nothing can, and no amount of upgrading ground coverage changes it.

Elites survive because they outlast a short window of fire. Their weakness is time in range, which makes slowing effects and longer exposure more valuable than raw damage — a Frost tower that adds two seconds to a traverse multiplies every other tower's contribution across those seconds.

Countering armoured, flying and elite enemies works through each of these in detail, and tower types and what they counter maps the seven attacking types onto the problems they exist for. The point here is narrower: identify the survivor first. A wave watched without spending is worth more than a wave spent on guesses.

What to move, and in what order

When rearranging is the right call, order matters, because each change alters what the next problem is.

Move support before attackers. Aura Towers raise the damage, attack speed, range or elemental power of neighbours. Re-aiming one at the cluster that is actually carrying the level is the cheapest large change available, and it does not cost you the coverage that moving an attacker would. Aura towers and support stacking covers how those boosts combine.

Then consolidate, rather than spread. A mixed wave tempts you into covering everything thinly. Late in a level the opposite is usually right: one location where a great deal happens beats four where a little does, because elites and bosses need sustained fire rather than intermittent contact.

Then fix the binary gap. If flying units cross ground your towers cannot reach, that hole has to close regardless of what else is optimal. A binary failure outranks an efficiency one.

Leave the boss itself until last. Bosses carry destructible components in some games and pure health in others; either way, the boss arrives after the mixed wave, and a layout that survives the mixed wave has already done most of the work.

Why this is genuinely difficult

It is worth saying that this class of decision is hard in a technical sense, not just a subjective one. A study applying reinforcement learning to high-level strategic control in tower defence games reported that training an agent which generalises across levels is difficult, with a hybrid of scripted logic and learning outperforming heuristics alone while still leaving a broadly competent agent out of reach.

The reason is the branching: placement, upgrade timing and resource allocation interact, and the right move depends on a wave that has not arrived yet. Research on generating and balancing game levels treats difficulty tuning as its own problem for the same reason — a level is only "fair" relative to the set of strategies a player can actually construct within it.

Introductory game design teaching frames the productive response as iteration: change one rule, test, observe the interaction. That is the right posture for a boss wave too. Change one thing between attempts, watch the same enemy type, and let the level tell you whether the change mattered. More writing on the studio's arcade and strategy titles sits under action and arcade games.