The Rules of Attraction: Location by Mark Seemann
Widen your hiring pool to the entire world; you won't believe what happens next.
Every week in my Twitter stream, there's some enticing offer:
NoWare is looking for strong F# or Haskell developers with experience in TDD, property-based testing, object-oriented design, REST, and software architecture.For half a second I go: "Oh, man, that's so much a description of me! This sounds terrific!" But then, with small letters, it says: "Must be willing to relocate to Sauda."
If your company is based in London, New York, Silicon Valley, Seattle, and perhaps a few other places, I can even understand that attitude. The local 'talent pool' is already big in those places, and more people are willing to move there for work.
In most other areas, I find that attitude towards hiring programmers lazy and self-centred.
As an example, consider Copenhagen, where I live. The greater metropolitan area is home to some 3.5 million people, so you'd think it was possible to find programmers to hire. Yet, every employer here whines that it's too hard to find developers.
While Copenhagen is nice enough, I can understand if the majority of the world's programmers aren't itching to move here. Apparently, this place is only attractive if you live in Afghanistan, Somalia, Nigeria, Syria, or some other horrible place.
All digital #
The outcome of programming is software, a 100% immaterial, digital product. Source code, as well, is entirely digital. Documentation, if it exists, tends to be digital. Source control systems are networked. Developers shouldn't have physical access to production servers.
Why do programmers need to be physically present?
Tacit knowledge #
I think I know why programmers need to be physically present in many organisations. It's because all the knowledge of 'how things are done' is in the heads of people.
- How do I configure my development machine? Ask Joe; he can tell you.
- What's the password to the CI server? Call Kathy in the satellite office. Maybe she knows.
- How do I find the logs related to this bug I was asked to fix? Go and talk to Jane. She can show you.
- Why did we build our own ORM, instead of using Entity Framework? Uhm, yeah, there was this lead developer, Rick... He's no longer with the company...
- Would anyone mind if I add an extra argument to the BloobedlyBloo method? Sure, go right ahead!
- How are we going to address this new feature request? I don't know. Let's have a meeting!
- What's the status of the Foo component? I don't know. Let's have a meeting!
- Is there any way to prevent the back-office developers from accessing our database directly? Probably not, but let's have a meeting!
Perhaps organisations that work that way are effective, but it doesn't scale. Every time a new team member or stakeholder joins the project, someone needs to take time bringing him or her up to speed. The more people in the organisation, the worse it becomes.
At one time I worked in an organisation that operated like that. I'd have meeting upon meeting about particular topics. The next week, I'd have the same meetings again, because a new stakeholder wanted to get involved. Nothing new happened; nothing new was decided, but it was necessary, because nothing was written down. Lots of time was wasted.
That's when I started writing documentation and meeting minutes. Not that I enjoyed doing that, but it saved me time. The next time someone came and asked me to sit in on a meeting, I'd forward the minutes and documentation and ask people to read that instead.
Conway's law #
Imagine that you could hire programmers from all over the world. Not only within driving distance of your offices, but from the entire world. Just think of the supply of developers you'd be able to choose from. You can literally select from among the world's best programmers, if you're willing to pay the price. Some companies, like Stack Overflow, GitHub, Basecamp, and Basho, already do this, with great success.
It's only going to work, though, if your organisation is set up for it. Processes must be written down. Decisions must be documented. Work must be asynchronous, in order to accommodate people in different time zones. You can't easily have meetings, so you'll need to discuss strategy, architecture, design, etc. using collaborative online tools.
If your organisation doesn't already work like that, you'll have to transform it. Software development with distributed teams can work, but only if you enable it. The good news is that there's plenty of publicly available examples on how to do this: just find your favourite open source project, and study how it's organised. Most open source is written by programmers from all over the world, at various times, and yet it still works.
As a modest example, my own experience with AutoFixture has taught me a thing or two about running a distributed project. Contributors have submitted code from Australia, South Korea, USA, Greece, England, Denmark, Sweden, Russia, and lots of other countries. Discussions happen in the open, in GitHub issues.
You'll most likely find that if you can get distributed development to work, your software architecture, too, will become more and more distributed and asynchronous in nature. This is Conway's law: organizations which design systems ... are constrained to produce designs which are copies of the communication structures of these organizations.
If your organisation is based on ad hoc ephemeral communication and impromptu meetings, your software will have no clear communication structures either. There's a word for that kind of software architecture.
If, on the other hand, you set yourself up for distributed development teams, you'll be forced to make the communication channels clearer - both in your organisation, and in your software. This will enable you to scale your development organisation.
You can claim that you're a 'world-class' organisation that hires only the best of the best (within a radius of 100 km).
Or you can be a world-class organisation by literally having developers working all over the world. You still need to be able to find the best, though. Perhaps you should try to let them find you instead.