Calculated risks are often necessary. Considering the application will get typically 30 hits per day, I would balance the time spent vs. the return.
I often work on projects and think of ways to improve the design once I'm elbow deep in the project. This time invariably adds up and when a dollar ammount (or pound/yen/euro/DM/whatever) is attached to this time, someone is going to loose.
If you think about it, after the time implmenting the locking, write the status checking code, and the handful of other tasks that would certainly pop up is added together, you have to ask yourself, "Why didn't I use a database?"
Someone once told me an anacronym, K.I.S.S. which stands for, Keep It Simple Stupid! I think this ceratinly applies here.
By all means, I do not encourage sloppy coding, I do encourage setting realistic limitations on the functionality of an application. This requires balance between functionality and cost (in time).
On a philisophical note, what technique is used in log files to prevent these types of problems?