Note: This is copied from my post on theserverside.net:

http://www.theserverside.net/news/thread.tss?thread_id=29949

Mr. Cook. BLOODY GOOD JOB!!!!!! Damon...


Since I was first briefed on this by Microsoft, anxiously awaited the build, and now have been using this for what seems like weeks, I am convinced that this is it.

A Silver Bullet? No. This will bring a different set of challenges, but economically they will be far less daunting then what we face today. I believe the next 12-24 months is the time for mass adoption of 'Software Factories' and ‘Software Product Lines’ in corporate America and abroad. It will fundamentally help fix an industry that is in incredibly bad shape (which you should know if you read any studies about our industry – We are a sick horse that in many companies has been shot ala outsourced to countries that do not even know what a bank account is). There are other products out there doing ‘Model Driven Development’, but there is nothing even close to this. I think my friends at Sparx are doing amazing things with their flagship Sparx EA Corporate Edition in Australia. However this is something different.

Microsoft Research perhaps is behind some of the innovations here (as the incredible book 'Software Factories' by Wiley - ISBN: 0471202843 shows).

The fact that 4 Microsoft Senior people wrote this, yet the intro is by John Crupi! (of SUN!) should tell you this is not a game.

It is authored by 4 incredible talents:

Jack Greenfield: Key Contributor to VS Team Edition and much, much more.

Keith Short: Another key Enterprise player for Microsoft Team Edition (both he and Jack are in Redmond).

Steve Cook: Coming out of a massive UML contributor role, he is in Canterbury, UK.

Stuart Ken: MAJOR UML expert and Program Manager for CS Team Edition.

Why am I taking my time to write this? Because I am strategically devoting a hell of a lot to this effort because it is it makes business sense, I enjoy working with it as a developer, and it works. My clients want it (yet many do not know how to ask for it yet), and my developers even want it. And I know we need it.

I have a book on our fundamentally improved (in terms of Agile family processes) Process coming out Q1 2005. What many people don't see is this is the logical step POST AGILE. If you have read the fantastic work of the Poppendiecks, and you are smart (and if you are on this site you almost certainly are), then you could make the connection. Not to spoil one of the major themes of my book (which is Technology Agnostic) but Software Factories are the ‘post agile’ place we will be if all works out well.

Think of how many people REALLY know Design Patterns for example? Unless you are in a software company, probably not many. As I always say (and it is totally true) : I can only hire 1 in 100 mostly because developers want ‘code and fix’ and will not devote any study time. It’s like ‘Hey I’m done with college, I don’t need to study’ I can only suppose. Now there are many INCREDIBLE people but I just don’t see all that many, and when I do I hire them . Even the most basic .NET questions are met with long silences like ‘what is boxing!’.

Think of component technology taken to a much higher level of power and abstraction, customized for YOUR needs, and then turned into a modeling environment that will generate implementations. I come from over 15 years of professional experience. I used the I.E.F. tool from Texas Instruments for example, and believed in I-Case in 1989. Now in 2004 we have possibly arrived to at least the generation of an implementation or simply one version of a collection of assets from a model. The best example (that is not even close to the power of this post is describing) is what my company is doing with SharePoint 2003. We are building hundreds of WebParts (and using the hundreds of external parts) to let SALES AND MARKETING decide how to package an application for the Financial Services vertical we are so involved in here in Manhattan.

Here is the analogy. The decision is made to pick say 40 ‘Web Parts’ and package them and brand them for the ‘Hedge Fund Industry’. Perhaps the decision is made that we need 5 new ‘Web Parts’ to make this happen. However this is mostly an assembly job (sound familiar?). So developers just think about a whole bunch of Web Parts. The product line managers are the assemblers now, not the old metaphor of ‘Component Builders and Assemblers’ as developers. The ‘Assemblers’ are non-technical. And a GRID component is not exactly a functional Application on its own so this is way beyond component technology.

Software Offering = A manifestation of the assembly of Software Assets, Customized to the domain

The Web Parts = Software Assets

Product Line = A series of Software Offering all made up of the core set of assets, differing by target market, and even individual customer. Customization, or Rapid Manufacturing (instead of Rapid Application Development) will complete shift the economics of software.

Are you with me at all here? Please email and tell me why you think this will not work if not. I’d love to know actually. I understand the human and political aspects but I also understand the economic and global competitive aspects. We live through cycles (I did Monolithic App, 2-Tier Client Server with stored procedures as the business & data tier, 3-Tier - where no business logic was put (or very little) in the database (where we still are today), N-Tier and now SOA N-Tier and beyond with all the WS- standards and what Indigo will bring).

The other dimension is process, where of course Structured Techniques, RAD, Information Engineering, early OO Innovators (Booch, Rumbaugh, etc.) , UML and the 3-Amigos / RUP, and now SCRUM, XP and agilefactor (the latest evolution we are releasing which is the next step in Agile and a step closer to Software Factories), and where we will go in process. Agilefactor addresses many of the issues people have addresses and is a process that is Agile to the core, yet can pass an audit and compliance group’s scrutiny.
Here is a common misconception I believe said with great clarity by Gordon Bell (with Microsoft Research now) - http://research.microsoft.com/~gbell/ in this snippet from his book “HIGH-TECH VENTURES : THE GUIDE FOR ENTREPRENEURIAL SUCCESS” : from 1991!
“Because nearly all software products are developed by a method that is subject to process control and quality standards, the Japanese appear to generate software that is a factor of 10 better in terms of quality, as measured in defects per thousands of lines of code, and at a productivity rate that is a factor of 2 to 3 times higher, as measured in thousands of lines of code per person per day. This development environment is often called a software factory, and as with any U.S. factory, no one wants to work in on ~ certainly not America’s creative software engineers.”
This has turned out to not be true, as much as I respect Gordon Bell and his amazing achievements (that are still coming). Now it is 2004 and his prediction has not come true, as others have experienced liked Ed Yourdon in the ‘Decline and Fall’ book (later reversed in the ‘Rise and Resurrection”.
Factories are nothing new as you can see.. Many were ahead of their time a decade ago. Now is the time and Microsoft appears to be the vendor that will make this real. Time will tell. But .NET is now multi-platform (BSD Unix, Linux, etc. as well as an
international ECMA Standard).

For example:

ECMA C# - http://www.ecma-international.org/publications/standards/Ecma-334.htm

ECMA CLI - http://www.ecma-international.org/publications/standards/Ecma-335.htm


The 2.0 version is around the corner as is VS.2005 Team Edition with DSL technology. Game On!

Damon Carr, CEO & Chief Technologist
agilefactor
www.agilefactor.com

P.S. For a MUCH more detailed treatment of these topics I will soon have a WIKI devoted to this technology. If you are interested please sign up for our newsletter here:

http://lb.bcentral.com/ex/manage/subscriberprefs.aspx?customerid=50925