Key takeaways
- An upgrade offer is only interesting if taking one means not taking the others — an offer with a strictly best option is a reward, not a decision.
- Build identity comes from synergies: upgrades that are individually modest and jointly transformative.
- Survivability upgrades are undervalued because their benefit is invisible, while damage upgrades produce visible numbers.
- The correct answer to 'damage or survival' depends on where the run is failing, which is information you have and the game does not.
- A run that ends because you had no answer to a specific threat is usually a build problem rather than an execution problem.
An upgrade offer is only a decision if taking one means not taking the others. That single property separates systems that create builds from systems that hand out rewards with extra steps.
Opportunity cost is the whole mechanism
A level-up that offers three options and lets you keep all three is not a choice. Neither is one where a single option is clearly best.
The design requirement is that the options be different in kind and exclusive. Then the player has to decide what this run is going to be about — and different decisions produce different runs, which is the entire replayability argument.
The roguelike tradition made this explicit; the Berlin Interpretation is an attempt to define the genre whose loop depends on exactly this, and what makes a roguelike covers where those definitions hold up.
Synergies create identity
Individual upgrades are usually modest. What makes a build feel like a build is combinations whose joint effect exceeds their parts.
Common shapes:
| Synergy type | How it works |
|---|---|
| Enabler + payoff | One upgrade makes a condition happen; another rewards that condition |
| Multiplier stacking | Two upgrades that multiply the same quantity compound |
| Threshold crossing | Enough of one stat unlocks a qualitatively different behaviour |
| Coverage | Two upgrades that individually cover half a problem and jointly cover all of it |
The best of these are the ones a player discovers rather than reads. A synergy you worked out yourself is the moment that makes a run memorable, which is why upgrade descriptions that state their combinations explicitly deflate the system slightly.
The survivability problem
Players systematically undervalue survivability upgrades, and the reason is about feedback rather than about judgement.
A damage upgrade shows you bigger numbers immediately. You take it, and the next enemy dies faster. The benefit is visible.
A survivability upgrade prevents a death. The death then does not happen, so you never see what you avoided. The benefit is real and invisible.
This is the same asymmetry that makes prevention undervalued generally. The practical correction: judge survivability by whether your runs are ending, not by how it feels.
Choosing between them
There is no universal answer, and the useful question is diagnostic:
Why did your last run end?
- Overwhelmed while enemies were still arriving → you need survivability or crowd control. More damage would not have saved you.
- Comfortable, then killed by something you could not kill fast enough → you need damage. More health would have delayed the same ending.
- Killed by one specific threat you had no answer to → you need coverage, not more of what you already have.
That third case is the one worth naming, because it looks like an execution failure and is not. A run that ends because nothing in your build answers what is in front of you is a build problem, and the fix is at the next upgrade offer rather than in your reflexes.
Build problem or execution problem?
A useful test:
Ask whether a better player with your build would have survived.
If yes, it was execution — you were in the wrong place, or reacted late. Boss pattern recognition covers improving that.
If no, it was the build. The fix is earlier in the run, in the choices you made.
Players default to blaming execution, because effort feels more actionable than planning. In build-driven games, the build is usually the larger factor.
Persistent versus run-only upgrades
A design question that changes the experience substantially:
Run-only. Everything resets. Each run assembles a build from nothing, and what you take is what you have. Maximum decision weight; nothing accumulates.
Persistent. Progress carries between runs — permanent unlocks, ship upgrades, starting bonuses. Lower per-run stakes; a sense of accumulation across a session.
Neither is better. They produce different relationships with failure: run-only makes a loss a clean reset, persistent makes it a partial gain.
Our two arcade games sit on opposite sides of this, which makes them a useful comparison. Zombie Survival: Last Survivor assembles its build inside the run: experience collected during a stage levels you up mid-fight and offers a choice between weapon boosts, damage increases, survival upgrades or a special ability. Space Shooter - Galaxy Arcade does the opposite — rewards earned during a run go into weapons, unlocked upgrades, ship systems and bonuses you choose, so the ship you launch with next time is a different ship.
Same shelf, opposite answers to "what does a loss leave behind".
Where our games fit
Both are free to download on Google Play, supported by ads, with optional in-app purchases and Everyone content ratings.
The zombie survival build guide covers making those in-run choices well, and horde survival versus wave shooter covers how the two structures differ. Difficulty curves in arcade games covers the escalation your build is racing against.
Everything we make here is under action and arcade games.
Good to know
Frequently asked questions
What makes an upgrade choice interesting?
What is a synergy?
Should I take damage or survivability?
Why do survivability upgrades feel worse?
How do you tell a build problem from an execution problem?
Sources
- Berlin Interpretation (opens in a new tab)RogueBasin — accessed
- Space Invaders (opens in a new tab)The Strong National Museum of Play — accessed
