In addition to the fully-loaded cost of programmers being substantial, you also have huge budgets for asset creation: art/animation/music/etc. Asset creation is expensive. It also depreciates straight off a cliff: the core audience will see it as novel for a period measured in weeks, days, or even hours in some pathological cases.

Many non-game applications have an adoption curve which looks something like the sales graph on BCC (100x now what it was at launch) or the famous YC "gradual adoption, spike, trough of sorrow, plateau, steady growth, traction" narrative. Games almost universally have a big spike at lunch, declining sales through the launch window, and then they fall off a cliff and never recover. The exceptions to this rule are so rare that even non-gamers know their names. My mom remembers Starcraft. You will not be Starcraft.

This greatly complicates post-launch marketing, iterating towards a game design which achieves fit with what customers will pay for, etc. (The ability to do that is one reason why social games work so well... that and fixing "our industry has no effing clue how to do customer acquisition" with copious helpings of spam spam spam your friends.)