Hacker News Reader

Comment by danieltillett | Hacker News Reader

Do we really know what the average programmer actually produces? We are all biased when deciding what is average as we use those around us to decide what is average - if we are surrounded by great programmers then our average is much higher than if we are surrounded by poor programmers.

The other complication is the intellectual energy hurdle that each problem brings. Some problems can only be solved by a small percentage of programmers, while others can be solved by anyone who can fog a mirror. How do we measure difference between programmers when the complexity of problems vary so much?

View on HN

> Do we really know what the average programmer actually produces?

Industry average is 2,000 lines of code per programmer per year.

I've regularly produced 10x to 40x that amount. Others I've talked to have done the same.

The difference between "low volume" and "high volume" programmers is usually pretty stark. My experience is that it's not so much experience as problem solving skills.

If you have a good understanding of the problem, you can come up with the correct solution, and move on to another problem. If you don't understand the problem, you'll work for weeks and get nowhere.

There's also a methodical approach. As an anecdote, I once had to use a script from another developer in the same company. It just didn't work. Paths hard-coded to his home directory (for a script supposedly used in the test suite), etc. I spent 3 days fixing it, and told him it had problems because it had no error checking.

His response? "I don't have time to write error checks."

My response was "WTF? I don't have time to not write error checks."

The 10x programmer has a fundamentally different approach to programming than the bad programmer.

I've observed the same thing in different contexts, but IMO that's supervisory negligence that allowed an employee to go rogue.

The story you're telling isn't a bad programmer -- he's a liar! He probably copied the file into the test suite and watched it fail, but it allowed him to report progress to the PM. (Covering for the time spent fucking around on the Internet)

That clown needs adult supervision.

The LOC/month amount depends heavily on the problem at hand.

With new code and familiar domain I'd say I can hammer in a few thousand LOC a week.

With a horrible legacy stack I've spent weeks on work that in the end resulted in a single new function call at a particular location.

Lines of code isn't a good measuring stick.

Wow 2000 lines of code per year - that is around 8 lines a day! I consider myself a very average developer and I would write 20x this. Maybe I am not so average after all :)

I could delete more than that in a week :)

2000 Loc/year. Wow, that's almost hard to believe.

I know people who write that in a day when they start a new project.

Of course LoC isn't the best metric, but still.

The best way to solve a problem is not to have it in the first place. Great programmers aren't better at dealing with complexity, they are better at reducing non inherent complexity and finding ways to break inherent complexity down into pieces that are easy to reason about. That takes experience and dedication to the craft. The latter I think is a huge difference between 10x programmers and others. The simple desire to be the best you can is a powerful force.

Reply on news.ycombinator.com