You sit in a windowless bunker. The world exists only as whatever your machines can sense and survive to report. Perception isn't given to you here, you have to breed it.
Every strategy game hands you an all-seeing eye in the sky. Adapt or Die takes it away. You command from inside a sealed bunker, and the only knowledge that reaches you is what a machine detected in the field: a faint contact if it merely sensed something, a solid one only if it had the radio reach to transmit home while it was still standing. (Premise drawn from the game's design documents and checked against its simulation.)
The consequence is quietly radical. A robot that wins its fight but carries no sensor and no radio teaches you nothing, no map, no memory, no warning. A weaker machine that sees, remembers, and radios back is worth more, because it hands you the one thing the bunker cannot manufacture: a picture of the world. The game even scores it that way: each machine's fitness counts how many enemies it personally sensed and how many observations it relayed home, and you set how much those terms are worth. So perception stops being a given and becomes a trait, something you must evolve into your designs, right alongside armor and guns.
The game is dramatizing a real evolutionary pressure. Across the tree of life, light-sensing has been reinvented again and again: simple eyespots are estimated to have arisen independently between 40 and 65 times, and full image-forming eyes several times over.1 Cephalopods and vertebrates famously built nearly the same camera-style eye on entirely separate branches, a textbook case of convergent evolution reaching for the same answer twice.2
Evolution doesn't repeat itself out of habit. It repeats itself when a solution pays. Light is an unusually rich channel of information about the surroundings, and even a crude patch of photosensitive cells lets an organism steer toward food, flinch from a passing shadow, or read the hour of the day. Each intermediate step earns its keep, so the road from no eye to a working eye is a gentle uphill rather than a leap, which is exactly why so many lineages have climbed it.
Strip perception to its essence and it is the reduction of uncertainty about the world, information in the literal sense Claude Shannon gave the word. In his framework, information is measured precisely by how much uncertainty a message removes, and a growing body of work treats sensory perception as the same act: the nervous system inferring the most likely state of the world from noisy signals.3
Uncertainty is not an abstraction to a prey animal. To detect a predator's chemical trace, or a resource's scent, is to convert vague danger into a decision you can act on in time.4 The organism that resolves the world a fraction faster eats, mates, and escapes more often than the one that guesses, so the machinery of sensing is under relentless selection. Senses evolved because a clearer picture is a longer life.
An army you cannot see through is barely yours at all.
Here is the trade-off the game makes visible. Eyes, nerves, and the brain that reads them are metabolically expensive. In young Mexican surface fish, vision consumes roughly 15% of resting metabolism, close to the relative draw of the human brain, and where light is useless, evolution takes the discount: cave-dwelling populations lost their eyes entirely, cutting the energy spent on eyes and visual brain tissue by roughly a third, a saving the authors argue was freed for growth.5
That is the same ledger you keep in the bunker. Every watt a sensor or radio draws is a watt not spent on plating or firepower. Each chassis supplies a fixed power budget, and a design that overdraws it does not die outright: every module simply runs proportionally weaker, so bloat is punished in the field rather than forbidden on paper. The budget is finite, so each machine forces a choice between being deadly and being informative, and evolution, in nature and on your workbench alike, is the art of balancing the two rather than maxing either. (Sensing-as-evolvable-trait and the power trade-off drawn from the game's design documents and checked against its simulation.)
Adapt or Die makes an ordinarily invisible truth explicit. In a world of hidden information, the power to gather knowledge and carry it home is not a supporting act to force. It is force. The best killer that fights mute leaves you exactly where you started, staring at bunker walls.
So you breed eyes and radios as deliberately as you breed guns, and in doing so you re-run, in fast-forward, one of biology's oldest verdicts: that an organism able to detect food, mates, and threats out-lives one that cannot. Blindness, it turns out, is not the absence of a trait. It is a design you can choose to leave behind.
Facts specific to Adapt or Die, that you command blind from a bunker, learn only what a machine senses and transmits, and must evolve perception as a trait against a finite power budget, are drawn from the game's design documents and checked against its simulation code. Claims about biology and information theory are drawn from the references below.