You're missing his point: Large or small parts of a codebase may be reused from project to project, but, especially in the days of Starcraft, games are often released as finished products, closed cases that the developers quickly move on from after shipment.
Digital distribution has made the subscription-based (MMO of your choice), "MVP"/"paid beta" (Minecraft, Terraria, and other indie games), and long term microcontent/paid update (TF2, various iOS games) approaches more viable than ever, but many game genres and corporate cultures are very much locked into the idea of shipping a finished product and moving on from it.
As an example of both the modern "continuous updating" and traditional/greedy "shipment oriented" approaches to development in a genre traditionally considered "ship and forget", you could imagine a company that makes adventure games (or interactive fiction or visual novels or whatever you want to call them) for smart phones that reuses and improves their game engine from release to release. Since a well-done adventure game should still be appealing even years after its original release, one approach would be to backport engine improvements to their older games in order to make them more attractive to long-tail customers. You might think of this as "we're selling stories, not engines." A more traditional approach (still employed by many Japanese developers) is to wait for a significant hardware or platform update and re-release the game, usually (but not always) updating the visuals or enhancing the interface, often several times, in hopes of milking fans dry with minimal effort (lazy ports) or risk (updated rereleases of games they already know people will buy). With the success of iOS and Android as continuously improving platforms in contrast to traditional static consoles, we may see less of the latter strategy in coming years.
On the other hand, Activision's Call of Duty series has been very successful sticking to an unquestionably franchise-oriented approach. Each game is basically the same as every other one before it, dressed up in a slightly different coat of paint, with slightly tweaked gameplay mechanics, etc, yet people enthusiastically plop down $60+ for every new release. Activision has no incentive to backport their engine improvements to their older titles because they want people to move on to the latest and greatest installment, and no one wants to buy the old installments anyways -- no one actually wants Modern Warfare 1, 2, or even the latest installment of today, specifically. They just want access to the latest iteration of the Call of Duty formula and all of its players, and they're willing to pay $60+ every year to do so.
