> when you can stop at developing the specs and send it overseas to be coded for half the cost or less.
Good luck with that. Please, do try it and report back to tell us how it went. My prediction:
No matter what you do, the spec will be incomplete, and if it's large enough, it will contain contradictions. The culture of not questioning superiors in many of "overseas" will make it hard to notice and only after substantial time without progress someone will realize the problem. That person or group of people will start communicating with clients and overseas to work out the problems in the spec, accumulating additional overhead. The changes and additions to the spec will render a lot of work already done unusable, so the overseas team will have to start from scratch. Then, they will work on the code, while you will be wondering if they're working or not. If it turns out they do, sooner or later they will provide you with some results. The result is going to be pretty bad, because competent people don't want to work for $2/h, no matter where you go. But you will get some result, and will begin testing it. You will discover a lot of bugs, and then you will have to fight tooth and nail to have them fixed, because nobody will want to take responsibility for the failures. At this point, the project will be a year late, and will have flown past all reasonable estimates in terms of required funding. In the end, you're forced to contract consultants - you'll have to sell your kidney to pay them - who will make the product barely-usable some 2 years after anticipated launch date.