So 2 examples, the first one is really a chart that shows that the 2 lines are pretty much the same. One is based on generation, the other is based on installed capacity. But just to show the transmission -- I'm sorry, the transition in terms of how many megawatts or how many megawatt hours with a single operator in our fleet responsible for generating in 2005 where the number is about 10-megawatt hour by employee. There's probably a factor of 1,000 there. It's probably somewhere around 10,000-megawatt hours, but just forget it for now. So 10,000-megawatt hours per employee to move into around 20,000-megawatt hours per employee, if I haven't skipped in order of magnitude. And with employees or people being our single most expensive cost in operating plants, this has been a very dramatic change.
The other area which shows what we can do today when we have the scale and also the expertise is what we have done on oil field maintenance. As Dita mentioned, we currently own about 9 rigs. They're used across the fleet for different activities, but among the 9 rigs, we have 4 rigs that are used for oil field maintenance,
Whether it's pump replacement or for the repair of existing wells. And back in 2005 or 2006, all of this work was contracted, all of this work was third-party work. And as of 2011, all of this work is done in-house. Why is it important? First, there's an element of price. As the oil field, oil and gas business was booming, prices went up, although just because the supply and demand, not because the work became more complicated and we were paying a lot more money for the same service in 2009. And this is one element, which sometimes you can address with your suppliers. It doesn't have to be like this, but this is one element. But probably, a more important element is the fact that there are not too many geothermal rigs, and they tend to be -- their owners tend to try and keep them busy. And so when the well fails or if the well requires work, somewhere in one of our operating facilities, often, we had to wait for a rig to become available somewhere and be moved by road for 1,500 miles or more than that. And this is all money that is lost. And so, when you do the analysis, actually, owning equipment and keeping it idle in many cases made more sense than waiting for 3 or 4 or 5 weeks for a vendor to come in and do the job. So many, many examples, but this is how we have leveraged both scale and technology to drive our cost of operation down.