Three features we built and took out again

Three things were built for Cosmic Starship, worked, and were removed anyway: a fleet command mode, the capture of enemy ships, and the first formula for the underdog bonus. Each one failed a different test, and the reasons are more useful than the features were.

Fleet Command

For two versions, 3.11 and 3.12, the game had a second way to play. You left the cockpit and commanded from above, in the manner of the classic fleet strategy games. Ships could fly on several flight decks stacked at different heights. A mothership sat behind your line and built things in the middle of a battle: salvage picked up on the field paid for extra squads and for warships called in as reinforcements.

It ran. It was also hard to read. With ships at several heights drawn on one flat screen, you could not tell at a glance who was above whom, and a strategy mode in which you cannot see the board is in trouble before any balance question comes up.

So we asked the balance question directly: does it help to be on a different deck from the enemy? The test flew the same engagements with fleets on the same deck and on separate ones. There was no clear difference in who won. The decks cost the player legibility and paid nothing back.

At that point the choice was to rebuild the mode around a camera that could make height readable, or to remove it. We removed it in 3.12.1, all of it: decks, mothership and yard. Half a strategy game bolted to an action game would have made both worse.

Capturing enemy ships

The idea came from survival games where you knock a creature out instead of killing it and then keep it. Here, you would shoot out a warship’s engines and turrets, leaving it crippled but alive, then send a squad of boarders across. A captured ship came home as a prize with a few random traits of its own, so no two were alike. It arrived in 3.9, together with a Boarders job for squads.

In a small test battle it worked as designed. The trouble is that Cosmic Starship does not have small battles. A large campaign mission can put more than 150 hostile craft on the field. To cripple one ship without killing it you must stop shooting at exactly the moment everything around it is still shooting at you, and then keep that ship alive, and your boarders alive, in the middle of the fight. The verdict after playing it in real missions was short, and the design should have predicted it: too many enemies to carry it out.

We could have made the battles quieter to suit the feature. That would have been the tail wagging the dog. Capture and the Boarders job came out in 3.14, and anyone holding a prize ship was refunded.

A formula that measured the wrong thing

This one never reached players, and it is the one we are least proud of having nearly shipped.

The underdog bonus pays more when a weak ship wins a hard mission. To do that, the game has to know what “weak” means. We already had a number for every hull: its battle value in the Who Would Win? simulator, measured over thousands of automated fleet fights. By that table a Morrow Hive is worth 1,600 and a Starship 6,000. Dividing one by the other gave a tidy multiplier, the Hive came out at three times the reward, and it all looked finished.

Then the test pilot flew The Iron Colossus in six different hulls. The Morrow Hive, supposedly the weakest ship in the game, won three flights of four. So did the Starship. The Obsidian Crown, worth 8,000, won two.

The simulator values were not wrong. They answer a different question. In the simulator a ship fights by itself, on its own base numbers. In a flown mission the main gun is yours whichever hull you sit in, and most of your toughness comes from upgrades, mods and rank. The hull matters far less than what is bolted to it. A bonus built on hull values would have paid triple for flying a perfectly good ship.

The version that shipped measures the ship as the battle actually builds it, and it was checked the same way before release: a bare Starship won two of five flights against the Colossus, the same ship with level 3 upgrades won all five. The underdog guide has the details.

What the three have in common

Each of these looked right from the inside. Fleet Command had depth on paper, capture worked on a nearly empty field, and the first formula was built on real, measured numbers. None of them survived contact with an ordinary mission.

Two habits came out of it. Measure in the conditions players actually meet, not the ones that flatter the feature. And when something goes, take all of it: no leftover menu entry, no orphaned currency, and a refund for anyone who had spent on it. A removed feature that leaves pieces behind keeps costing players attention long after it has stopped giving them anything.