How do you measure your productivity?

Question

Hi everybody,
I have troubles estimating my productivity as a Java programmer. As far as I know "a normal programmer" produces from 80 to 400 lines of code (LOC) per month (comments are not counted).
I now it is very unclear how to define and estimate programmer's productivity a and I know it depends on a lot of factors. But still, I would be extremely greateful if someone would share with me how many lines of Java/JSP code s/he produces per hour (or per month).
A lot of thanks in advance.

Answer 1

How about trying to measure the financial return on the code you produce.
If, for example, you are a reuse asset producer, perhaps you could measure costs on the basis of SLOCRBO (Source LOC Reused By Others)?
So if you write a 1KLOC module that is reused in 10 projects, that's
SLOCRBO = 10 * 1000 = 10,000
You could try to take this forward by gathering stats for the relative cost of reuse for modules produced in your company. Let's say it costs 20% of the cost of writing a module from scratch to use a 'reusuable' module instead. So you can try to work out the amount of money you are saving the company:
CPLOC = (Cost per line of code) = $100 (say)
RCR = (Relative cost of reuse) = 20%
DCA = (Developer cost avoidance) = 10,000 * 0.8 * $100 = $800,000
But you'd need to measure code quality, so you need stats on error rates, and how much it costs to fix errors. This might depend on when errors tend to caught, and how critical the apps the company develops are. So:
ER = (Error Rate) = 0.001 (errors/LOC)
EC = (Error cost) = $10,000 (say)
MCA = (Maintenance cost avoidance) = 10,000 * 0.001 * $10,000 = $100,000
So your Reuse Cost Avoidance would be DCA + MCA = $900,000
Then you'd have to take into account your labour costs, costs of maintaining the reuse program, asset administration and so on. You are also going to need to take into account the impact that your bugs, and any future incompatable changes, are going to have on systems that reuse your code. You could (I think) get a fair idea of what you might be worth to the company, and how much money you could save by increasing Reuse a bit. Obviously, this kind of calculation is going to need statistics from the company to make the calculation meaningful.
You could also look at your maintenance patterns: Do you have to keep making lots of small changes to your code? If so, maybe you could improve your productivity by performing some preventative maintenance - improving the maintainability of your modules by re-engineering them.
I do think it's important that you do try to measure productivity - producing metrics is the hard part - but without taking measurements, how can you improve a software process (either from the pov of a manager, or as a matter of personal improvement)?