Should really we measure developer efficiency?

About 10 yrs in the past, I wrote a web site put up called “Can we measure developer productivity?” In it, I mentioned the a lot of objective attempts that had been made to do it — lines of code, purpose details, and many others. I also proposed some subjective steps. However, the summary was that regardless of the dreams of KPI-loving managers, there was no feasible way to evaluate the productiveness of an particular person computer software developer.

I mention this article released 10 a long time back mainly because points have transformed significantly in the many years considering the fact that. When I wrote it, Git and Mercurial were being both of those outstanding and popular program resource control programs. I was a software program manager at the time, migrating my crew off of Visible Source Risk-free from Microsoft, and we made a decision to go with Mercurial simply because it was a lot more Windows-friendly.

We picked the improper horse for the reason that, in the many years to occur, Git would come to be the de facto typical for variation management. As a consequence, a cottage sector has arisen all around Git repositories. GitHub is a substantial business enterprise for which Microsoft compensated $7.5 billion. Quite a few businesses now supply metrics about your code in Git. And several of these corporations purport to evaluate the productiveness of application developers.

Gimme metrics

If we concede that it is achievable to evaluate developer productiveness (a proposition that I am not entirely marketed on), we then will have to ask irrespective of whether we should really do that.

The wish to do so is surely solid. Administrators want to know who their ideal builders are, and they want metrics that will assist them at efficiency evaluation time. HR wishes to be equipped to doc efficiency issues. CEOs want to know that the revenue they are investing is staying utilized correctly.

Even if you use new tools to evaluate personal developer productiveness, all those metrics will most likely be gamed. Lines of code is viewed as a joke metric these times. “You want lines of code? I’ll give you traces of code!” Is amount of commits for every working day or normal time to very first PR remark any distinctive? If you measure unique developers on these metrics, they will most definitely make improvements to them. But at what price? Probable at the price of workforce efficiency.

An old CEO of mine used to say that application development is a team activity. If specific builders are measured towards every single other on any metric, they will start off competing with each other, specifically if funds and promotions are on the line. And a team of people competing versus every other is not a team.

It is groups, not individual developers, that get points completed in the program business enterprise. Software program improvement is exciting in that regard. The real coding is generally finest carried out by persons in deep assumed, but the perform that comes about right before and following the code gets prepared contributes drastically toward building factors productive.

Measure the workforce

A progress staff discusses the structure and implementation of a offered venture ahead of any code is created. When the specific builders write the code, it is frequently with the assistance of teammates who solution thoughts and provide insight. All group associates critique and approve what is carried out during code opinions. Every person operates jointly to make issues materialize.

The power of the team is each and every personal member. The strength of every single member is the workforce. —Phil Jackson, NBA coach

That is why, as a substitute of measuring unique developer efficiency, it is staff productiveness that should really be measured. Developers doing the job alongside one another as a staff, pushing toward a frequent intention, is what administrators truly want.

Groups know that if they boost their team metrics, they make improvements to their results. Teams that can see the advantages of concentrating on the suitable factors and trying to keep an eye on the ideal metrics will strengthen. Teams want to increase their productivity. They want to get improved. They want to deliver. Measuring workforce-dependent metrics assists them do these points.

10 many years ago, I asked if we ought to or even could evaluate developer productiveness. But I was asking the wrong issue. Individual builders are only as potent as their teams. Correctly measuring group metrics prospects a workforce toward greater outcomes and better software program. As a substitute of measuring people today, we should motivate our teams to make program greater and a lot quicker by measuring what they do with each other.