Farmsourcing: Part II
In my last post I stated the real failure of outsourcing firms is ignoring the fact that they are responsible for quality. Most offshore outsourcing firms view software development as a manufacturing problem. I don't think this way though. You don't think this way. In fact the midwest doesn't think this way about software. Cray doesn't manufacture it’s compilers, it crafts them. DowJones doesn't manufacture it's systems, it crafts them. They're crafted by craftsmen using tools they have often built themselves. Quality is a focus.
Additionally the cost of living here is reasonable compared to either coast. I travel to New York often and stay in a basic room at the Waldorf Astoria in Midtown. It costs around $350 - $400 a month. For that same price I can get a two room suite at The Saint Paul Hotel. I get twice as much hotel here in Saint Paul compared to New York. The same is true for software development. In fact with raising prices in India year after year, the price difference per hour can be as little as $20 and as high as $50. Though on average it's around $25 to $35 and hour. That price difference is even less once you take into account soft costs. For example having only one hour time difference, common language, ability to meet in person cheaply (thank you SunCountry). By far the best benefit is talent though. Developers in the midwest are using the best and latest technologies, have access to a number of mentors, conferences and user groups. Did I also mention developers in the Midwest can produce nearly twice as fast as those in India. If this speed is included, developers in the Midwest are actually cheaper and of course, developers in the Midwest focus on quality.
This is the essence of what I call Farmsourcing. This combination of reasonable developer costs, developers that focus on quality, increased productivity and proximity to fields of corn. So how does it work? After all good talent isn't cheap. My solution, use unpolished talent, pay them lower wages, mentor them for 1 to 2 years and then get them a better job somewhere else. This solution has problems though - some similar to those India faces.
First I need to find a good source of unpolished talent. For that we have new CS grads from universities like the U, St. Thomas, Macalester, Metro State and so on. Let me qualify what I mean by unpolished talent. I define it as someone who understands the concepts but doesn’t know how to implement them.
Second problem is the lower wages. I pay anywhere from $20 to $40 an hour and most times closer to $25. Why would someone work for these lower wages? Let me ask you this question. Why would someone pay to go to grad school? Most likely because they want the final education that will set them apart. Well that's what I’m offering: grad school expect they don’t pay - I pay them. Here’s what I tell prospective employees: come work for me for one to two years, I'll pay you a decent but lower wage during that time. But I will also mentor you on everything from version control systems and continuous integration to project management and contracting. At the end of the one to two years you'll not only be able to get a job worth twice as much but I'll help you find it.
The third problem is how to actually mentor barely functional into quite good? Or put another way, how do you scale your team? I'm not going to go into the details of this since this process is our secret to success. But I'll say it's a combination of three things: The mentors, the process tools and Ruby on Rails. The latter of which I'll talk more about.
Rails really makes this happen. First, I can explain Ruby to someone in just a couple days. There isn't even a book they have to buy to learn it. Try that with C++. Second, Rails is very opinionated. So long as you do things the Rails Way life is easy. When you don't, things get hard. It acts like a warning light for newbies when they’re doing something they shouldn’t. I tell them, if it's hard, ask a mentor. Third, Rails is Ikea. It has darn near everything you need already so there is no need to build it. This is good for a couple reasons: it makes apps smaller so statistically there should be less bugs, you can produce faster because you don't need to build it, and all the code in Rails acts as a nice reference when you do need to build something yourself. Fourth, Rails has testing built in. You need tests when working with barely functionals. Fifth, there is no compile step in Ruby and thus encourages what if exploration which results in deeper learning.
Because of Ruby on Rails and Farmsourcing I no longer worry about the enterprise and every manager and CIO I talk with about Farmsourcing with Ruby on Rails worries less too. They also quickly find a project to try it out on. How big can Farmsourcing get? I’m not sure but what I do know is outsourcing is a $20 billion business and I’m not quite there yet. So if you have capital to invest so we can faster and keep up with demand or are interested in a job, please contact me.
