Charles Zedlewski's post on agile development made me think about long accepted adages in the software industry. We are shackled by"iron triangle" thinking - something's gotta give mentality- if you do not adjust software scope or the software budget or the software schedule, the software quality will suffer.
Which is why we seem to just accept Oracle 11i significant quality problems. Just shrug our shoulders when SAP says only half its customer base will by 2010 have implemented what it is marketing as its single biggest innovation - its web services architecture. Cynically listen to Microsoft talk about "scrum" development knowing it has spent $ 20 b in R&D in the last 3 years and does not have too much to show for it. Or when IBM extols its embrace of the Linux - "bazaar" type - development model while its large global services group still uses more traditional (and significantly pricier to clients) "cathedral" development models. Just because the scope of what each does is large, we accept their traditional quality, time and budget standards.
Several benchmarks in the last few years allow us to challenge their performance - and broadly prepare us for a revolution in software development and implementation.
Let's start with the Indian services industry. The major reason companies went there for software development was economic. But they are delighted with the quality improvement . The CMM and Six Sigma initiatives at a number of Indian firms have quantifiably improved software quality. 60% of all CMM`Level 5 certified shops are in India. Users are even more delighted with the time to market advantage that comes from "follow the sun" development teams handing code back and forth.
But India is just the beginning. As I wrote in Frederick Taylor and Technology Services, the software development and implementation ecosystem has so many inefficiencies, that Indian vendors have only scratched the surface. There are so many geographic, functional, capacity optimization and continuous improvement and automation opportunities in the software and services supply chain.
I was talking to someone recently that in the near future we could be seeing very good software talent at $ 5 an hour. How? A number of countries are aggressively emulating India's success. A number of US states are providing incentives for software factories. Communities like those around open source products can test software for near free, and often more robustly than internal testing teams. Communities being built by elance and amazon.com allow for chunks of development to be easily parceled out at affordable costs and rapid turn around.
Time for the software and technology services industry to break out of the old mentality and old cost, quality and time benchmarks. Melt the iron triangle. Oh and by the way, then we can turn to the golden calculator that makes sales and marketing costs in the industry even larger than software development costs.