I heard a presentation by Pegasystems founder and CEO Alan Trefler on Oct. 20 that ties to my research along the lines of "What Happened to the Software Factory?" (See also http://www.itbusinessedge.com/blogs/den//?p=108here.) In front of an overflow crowd of what I think were mostly systems (now called business process) analysts at the PegaWorld user conference, he laid into the 1970s way of software development.
He even said he thought the still-continuing problem was older than that, but that he wasn't old enough to remember. I am. He's right.
Trefler was talking about the methodology of the systems analyst talking to the user, coming up with a systems spec, taking it back to the user some time (months?) later, getting approval, taking it to the developers, getting code back from the developers some time (months?) later, etc. He bemoaned the fact that the industry has not made much progress in 30 years and wants to see software development that is software-enabled.
He sees three steps in emerging software development tools:
- Automate the discovery of the "objects" that the user needs (what's the data look like, how does the process flow?)
- Automate and compress the "programming" by getting all the players in one room, at least virtually (getting rid of a person coding thousands of lines of error-prone HTML code while you're at it)
- Automate the result, "dynamically bound."
Step 2 is common sense and just requires cultural change more than technology. Steps 1 and 3 are the challenge. Pegasystems and others are beginning to offer such capabilities within their own environments, but the longtime goal of IT is to enable what Trefler calls discovery/programming/binding no matter whose objects, no matter where the bound result will run.
It reminds me: I have to circle around and see how the Open Service Oriented Architecture Collaborative is doing.