Judging the sizes of the stacked pages from the last 60 days of code would probably be a pretty good indication of productivity. Perfect? No. Good enough? Maybe.
I never thought I would see someone on HN unironically advocate for counting LoC to measure productivity. Unbelievable how this guy's cult of personality poisons brains.
100% right. I just watched an OJ Simpson trial documentary. From that making the small point being famous brings perception then perception management... Which goes fast to nowhere. Fixing Twitter can't start with loc and the new owner reading code. I can guarantee Gerstner never did that at ibm.
I’ve worked in big companies and I’ve worked in small companies. The one constant has been that writing code is required to make product changes. No code likely means no changes. Maybe LOC != productivity makes sense when algorithms are of great importance, like in situations where a genius algorithm can unlock tons of value. Machine learning would be one example. However, most work is feature work, which typically has a straightforward path from start to finish. In product feature work, it’s unclear to me what activity would use time productively that didn’t result in lines of code.
People, even those at other big tech companies, unironically refer to working at Google as "retiring." Not sure 'big tech' is a great example to use when talking about productivity.
As in, the fewer pages, the more concise they are? What about clarity, and which languages are we talking about? Bonus points for Java? I think it’s been fairly well determined that LoC isn’t a good metric for judging productivity.
I think LoC has been determined to be a bad productivity metric because it’s gameable and incentivizes bad behavior. However, that doesn’t apply here as the engineers didn’t know they were going to be evaluated by LoC a priori. I struggle to think of cases were lines-changed wouldn’t be correlated with the productivity of engineers working in a consistent environment if they weren’t trying to game the system.
The problem is some languages are more verbose than others. Considering boilerplate (which is often automated these days) it extra has nothing to do with productivity. More lines doesn't mean more action - I mean, we're on a Lisp-derivative forum.
And sure, if I was getting judged by LoC I'd just paste in a library versus including it, which would generally be worse.
