3 ms·
It's quantifying that value that is the difficult part. Delivery of x amount of lines of code would be a terrible metric for example...
by davewasthere 12y ago
It's quantifying that value that is the difficult part. Delivery of x amount of lines of code would be a terrible metric for example...
- crdb 12y agoI'm not saying it was the only predictor, but we did have a candidate who solved the application task with a quarter (1/4!) of the code submitted by the next best, whose submission was already beautifully succinct. So, number of lines of code CAN be a meaningful predictor ;) More seriously, I agree with you. I don't think performance is quantifiable with parametric methods, you just have to allocate a fair amount of mental effort to the problem. For example, a "maintenance guy" who enjoys breaking other people's code, fixing holes and doing tests is extremely valuable, but not someone that is easily sold to non-technical management, nor that is very visible; at the same time, without some builders to work with, his salary is wasted. We get away with it because we have a carefully selected (around 3 man-month of just recruiting) small team (10) of experienced developers from a small but strong and mature community (Haskell) on a well defined problem (fix and improve a large e-commerce platform). I shudder to think of the magnitude of the problem for a much larger company with much larger teams, like Google or Microsoft.