Oftentimes, developers don’t even know their own game well enough to effectively finesse the load times until near the end of a development cycle. Game development is made up of countless moving pieces, each of which is on the verge of catching on fire at any given moment. It’s a series of trade-offs. Implementing one thing over here might cause longer loading times or crashing bugs over there. It depends on the game engine, too. Some engines handle the particularities of certain loading techniques better than others, and most general-purpose game engines are built before developers really know how their particular game will turn out. There’s guesswork involved, and those guesses don’t always pan out.
To avoid breaking stuff, developers typically like to have the content of their games locked in before they start surgically slicing away at things like load times. Problem is, that doesn’t usually happen until near the end of development, when there’s just a pinch of sand left in the hourglass. Developers can only address so many problems before it’s time to send a game off to certification and then release. And load times aren’t the highest thing on the priority list.
“When you enter the home stretch to ship a game, you take stock of all the problems it has and prioritize each based on the costs and benefits of fixing [or] improving it,” JP LeBreton, a designer who has worked on games like BioShock and Spacebase DF-9, said in an email. “For launch, devs often prioritize stability—making sure the game never crashes, and is fully playable/completable—over performance.”