You can do some performance indicators, for example, cost of data centres and things along those lines, which are measurable based on the dollars. We need to drive down transactional costs, for example, volume on a wide area network. You drive the cost down and you measure per unit of cost. Over the years the number of units may go up because people are doing more, but the overall cost has gone down per unit. Those are the kinds of metrics that you need to establish up front: what does it cost the Government of Canada to delay; what kind of investment do we need to get to tomorrow; and what do we expect we'll be paying in the future?
As Kevin indicated, hardware costs are going down continuously already; cell costs are going down. Government has done a wonderful job of negotiating enterprise agreements for some of these things, for software as well. Lots of good work has been done doing that.
The message is, it's not always about cost. Cost tends to drive people down to the lowest common denominator. Sometimes you need to pay to get a value-added service for what you're delivering. On infrastructure, for example, can you drive down the cost because you need to maintain what systems need to do? Sometimes systems aren't the cheapest because they deliver things that you have to invest in to make that happen. There's always that trade-off.
The issue is you develop your performance indicators up front, and you measure against them. It's not just low cost, because overall, when you measure it from an end perspective, it needs to be best value. Inherently, from a business case perspective, you look at 45 departments delivering their individual things or one corporate enterprise that's bringing all that together. Inherently, the one corporate enterprise makes perfect sense, but you need to put the right governance around it, and there are indicators of how you manage that and measure that to move it forward.