Prior to making any other comment I would like to say that I have worked extensively with excellent off-shore development organizations and without exception they are uniformly excellent with extremely high skill sets and quality development work. So unlike other similar prognostications, mine has nothing whatsoever to do with skill, quality, time-zones or complexity of management processes. I believe these have all been completely mastered in off-shore development.
The reason that off-shore development is popular is cost. A very high quality product can be delivered at a lower cost. The problems that off-shore development organizations will face in the next 5 years are two-fold.
- The cost of doing business; paying employees, finding appropriate offices, powering datacenters is rising fast. I good friend of mine runs a small off-shore development company. His payroll costs have gone up 50% in two years, his office costs have gone up 100% and his off-shore billable rates have to go up to make it profitable. The IT market is so hot, he must provide project completion bonuses just to keep his employees through a full release. His company is not alone, so the unthinkable is already happening. Off-shore rates are rising as any good economist would predict. Well you say… they could double and we would still be ahead. Yes you are right and so am I. They doubled over the last 7 years already and will likely double again over the next 5, bringing the small off-shore organizations into the $70 / hour range and the big SI’s nearing the $100 mark for Senior IT resources.
- The amount of labour and skill level required to deliver software functionality is going to drop dramatically. Yes, I’ve been there. I lived through the promises of 4GL’s, 5 GL’s, CASE tools, Object Relational Modeling tools, Business Rule engines, OOA/OOD Software Development Environments and yes each and every one was to have promised to make software development simple, fast, high quality and low cost. They all failed to deliver on the promise. However this time it is true and my cynicism has been wiped out. (A very tough thing to do). If you have not checked out OSLO. Take a look. https://connect.microsoft.com/oslo OSLO changes the game completely. (new and misleading name is SQL Server Modeling CTP) It is conservatively estimated that it will bring a 10 times productivity gain to the app dev lifecycle. I think it will go beyond that. Why? Because the 10X number was for the hot shots. The super-programmers. Think 20X for for your average developer. It adds speed but more importantly it simplifies the entire process of software development.
The overall impact is a net reduction of 62% to 6% of the hours for development centric hours. 56% of the project time.
Now take the two factors above and do the math for a 2,000 Function Point system. In my previous post I used 2000 x 12 hrs/FP = 24,000 hours for the system. The rate for OSLO would be approximately 5.5 HRs/ FP.
- Off-shore – Today’s big SI’s will bill this at $55-$70 / hour off-shore $1.3M – $1.7M cost to the end customer
- On-Shore with OSLO at today’s rates $100-$150 / hour on-shore $1.1M -$1.65M cost to the end customer
OSLO is not architected for a waterfall Requirements, Specification , Analysis, Design , Code , test cycle. It is so fast and so modular that literally you can build on the fly with absolute quality, tear it down and build again in moments. It works with your customer in the room. and will change the way that everyone thinks about software development.
There is a finite IT budget in the world. ($1/2 Trillion perhaps for development?) and subsequently a finite number of projects. When you graduate 110,000 – 220,000 programmers every year from universities, it is unclear if there are enough IT projects to sustain that.
It will be 5 years before this paradigm shift in development takes root and starts to become broadly adopted. But when it does, don’t be heavily invested in off-shore development companies for your retirement fund.
I think its important to note that developing code is only a given percentage of project effort. That is, the complete scope of the project contains many items needing labour and management over and above the SW dev piece. Technologists often have a product centered view of project delivery, and do not necessarily take into consideration fully all sorts of other elements required to take a SW vision and turn it into a production solution. So for the SW development aspect I will fully grant you this argument. I would also argue that the price / effort of the rest of the scope will remain reasonably constant.