On a good day estimating the scope of a software development project is problematic. Based on success rates it appears intractable. Without arguing whether or not setting up a web site constitutes software development it is interesting to note that our newly sworn in County Commissioner is "already working to create a web site that will show real-time information about her office's spending." So sayeth the AJC. The interesting part is that despite the preliminary work "she said she hopes to have the website ready in about a month, but it could take longer."
No joking.
This could just be adept politician-speak, where every apparent commitment comes with a caveat you could drive an epic fail thru or it could be a voice of experience that understands no software effort ever comes in on time and under budget. Not with all the requested features anyway. Or maybe she's thought better of hiring friends or family for this project as that would not be a good start towards good governance. But it would help her fit in with the rest of her colleagues on and beyond the Commission. Or perhaps she is seriously considering hiring the yahoos responsible for Dunwoody's CAD-to-CAD (non)interface and is certain that will make the project open ended if it ever works at all. Deliberately choosing to fail on the first and only known transparency project should give everyone pause.
Given the decades of software development experience in the community of The Other Dunwoody, we would like to offer a time tested software estimation technique:
No joking.
This could just be adept politician-speak, where every apparent commitment comes with a caveat you could drive an epic fail thru or it could be a voice of experience that understands no software effort ever comes in on time and under budget. Not with all the requested features anyway. Or maybe she's thought better of hiring friends or family for this project as that would not be a good start towards good governance. But it would help her fit in with the rest of her colleagues on and beyond the Commission. Or perhaps she is seriously considering hiring the yahoos responsible for Dunwoody's CAD-to-CAD (non)interface and is certain that will make the project open ended if it ever works at all. Deliberately choosing to fail on the first and only known transparency project should give everyone pause.
Given the decades of software development experience in the community of The Other Dunwoody, we would like to offer a time tested software estimation technique:
- ask the programmer how much time it will take to develop the feature or product, making sure you get the estimate they honestly believe in and not the inflated estimate they know their managers expect
- take that estimate, double the number and go up to the next time unit (e.g., 1 day -> 2 weeks)
We are guessing that the CAD-to-CAD bozoids told our previous Most Eminent City Manager that this simple interface project would take two months at the outside. Seems like it has been a bit closer to four years, doesn't it?
Posts
