On 6/25/10, Greg Harris <[email protected]> wrote: > This is going to cause a few mad Friday tangents…. > > If you have ever worked on a sales performance recognition system that > allows for any flexibility beyond the total of who sold how much in dollar > terms when, you will have seen that there quickly starts to be manipulation > of the input data to affect the recognition outcome (rewards). > > e.g. look at all the sales people who are running around just now to meet > their end of financial year goals, where the sales will be quietly credited > back into the system at the beginning of the next financial year! > > If sales people (who have a far lower average IQ) than programmers can > manipulate the system to meet their personal goals, what are programmers > going to do?
FWIW, not only is this a pretty offensive comment, it also doesn't logically follow that someone who is "smarter" will be "more dishonest". But it is a fact that any system needs to consider how it can be 'gamed', and to what degree that is expected. That's why, IMHO, the most appropriate one is where the limits can be controlled by management trivially (i.e. time estimates) and the result is directly useful. More code isn't directly useful. Less code isn't directly useful. A project delivered on time *is* directly useful. > But I would also say, if you cannot measure it, you cannot manage it! > > But take care with your stats, statistics show that far more people die in > hospital than in the community as a whole, so if you are sick, you should > avoid going to hospital! WRONG! (for the sales people in the room, the > reason more people die in hospital is because it has a biased population, > sick people, who have a higher probability of dying) > > Have fun on Friday > > Greg H -- silky http://www.programmingbranch.com/
