There was a post today over at 37signals about sunk costs that I’ve been thinking about. It’s been rolling around in this skull of mine all day.
If you aren’t familiar with the idea of sunk costs, Wikipedia states that they are costs that have already been incurred and which cannot be recovered to any significant degree. The Economics Web Institute has a good definition, too (search on that page). I was first introduced to this concept by an old boss who was studying for his MBA. “You have to be able to walk away from sunk costs,” he said. So many people consider walking away a waste of time, money, and effort, though. In order to not waste their “investments” they cling to them, instead of expending effort on projects, software, and systems that do a better job from the ground up.
For system administrators, much of our time and effort ends up being a sunk cost, spent towards the overall investment in whatever we’re building. Software licensing is a sunk cost. Server hardware is a sunk cost. By themselves, all of the components of a system are sunk costs. The assets gained just aren’t usually worth anything in resale value.
Things are a bit different for programmers. Code can be considered an investment if your business derives revenue from it. It certainly isn’t a sunk cost. Joel Spolsky has talked about this in his posts about rewriting code from scratch. This may be true of tools sysadmins write, too. Maybe. Is it ever a good idea to throw a tool out and start over?
When should you scrap a project or a system and start over? When isn’t it a waste to keep going? Can we do something to limit the “waste?” How ready should we be to switch to something new? Can we do anything to turn sysadmin sunk costs into saleable assets?
I have some ideas, but they need some more time to soak. Hang tight.
You are spot-on with sunk costs and theory of sys mgmt. But you have to remember that sunk costs often (in the case of IT most always) generate variable future costs, operational costs (or OPEX) once the system has been installed.
These variable costs include on-going maintenance, license fees, admin labor, floor space, electricity, etc. This concept of sunk cost paralysis as one of your links describes, is compounded by ongoing OPEX costs tied with capital assets. Although we cannot recover from sunk costs, we can objectively look at the OPEX run rates that these sunk costs are generating and make management (technical and financial) decisions on how we can proactively reduce OPEX costs.
Dave Merrill