- Games are written on a prospective basis - it is hoped that people will pay to play.
- Games don't make money until they hit the shelves - a delay in shipping can kill (and has killed) development companies or publishing companies.
- Games are surprisingly low-margin - the retail price is split between the retailer, the software company, the publishing company, and in the case of consoles, licence fees. There's advertising, shipping, salaries, and rent to pay for - plus stockholders want some profit.
Most other software (not all, but most) is written on a contractual basis - there is a business user who wants it, or a number of business users who will pay some money up front for it (like how MS and Apple got starter money in the 70s and 80s). This means that the client can be charged for milestone deliveries - unlike games.
That said, the abundance of programmers who want to get into the industry means that games software companies can suffer surprisingly high rates of churn in their staff.
Games developers themselves can take some of the blame - I know a few, and at the start of a one or two year game project, they're playing around, trying cool stuff, and then end up in 3 months of crunch development. That's more of a cultural thing - it's rare to see this in business software companies that run on fixed shipping dates - they tend to have rigid schedules for research and development. Other companies that have a "ship it when it's done" culture, like Google or Valve, are different again. If they're good, and get the balance right, all the cool stuff makes it into the final game. It doesn't always work out, though.