General talk about software patterns

[mark AT graybill.com]

Having experience in two ULS programs (IBM Workplace OS and the Army's FCS), I have some confidence in saying the origin of problems in ULS programs are mostly non-technical.  That is to say, issues in psychology, namely social psychology, prevail over technical issues - even over the most carefully developed software engineering and management processes.

My observation includes experience with programs executing under CMM level 5 and ISO9001 certified processes and also rigid government-regulated processes (FAA, FDA, MIL-STD).  Furthermore, it has been my observation of 30+ organizations that as complexity and team size increases, the issues become much more people-originated.  (You might want to look at the work of Larry Putnam and Alistair Cockburn.)

But I'm just an average consultant working in the trenches and don't really know why I'm writing this (chuckle) other than to opine what I believe underpins dominant issues in ULS programs.  Unfortunately, I don't wield impressive credentials - only some experience and perhaps a slightly odd perspective, so take this for what its worth.

Cheers!
Mark

 

> ----- Original Message -----
> From: "Richard P. Gabriel" 
> To: members AT hillside.net, patterns-discussion AT cs.uiuc.edu
> Subject: [patterns-discussion] Ultra-Large-Scale Systems Workshop @ OOPSLA
> Date: Tue, 22 Aug 2006 14:45:07 -0700
>
>
> This workshop might be of interest to you or to people you know. At OOPSLA. 
> Please pass this along.
>
> http://dreamsongs.com/Feyerabend/Breathturn2006.html
>
> Short blurb:
>
> Scale changes everything.
>
> The trend in the design and development of software-intensive systems today 
> is toward scale that increases in every measurable way. Lines of code, 
> complexity, dependency, communication, bandwidth, memory, datasets, and 
> many other measures for our systems continue to reach and exceed the limits 
> of our ability to produce high-quality systems for all purposes.
>
> These systems will be unbounded, integrating internet-scale resources. They 
> will serve diverse stakeholders with competing objectives and at the same 
> time be constrained by policy, regulation, and the behaviors of their 
> users. The lines between development, acquisition, and operations will 
> blur: ULS systems will not die; they will be too large to be replaced and 
> will be inextricably connected to the day-to-day mission. Rather, they will 
> continue to evolve over time with behavior often more emergent than 
> planned. Because complete specifications will not be achievable, sufficient 
> assurance will have to do. ULS systems present "wicked problems," ones for 
> which each attempt to create a solution changes the problem. Some of these 
> characteristics appear in conventional systems, but in ULS systems they 
> will dominate.
>
> -rpg-
> _______________________________________________
> patterns-discussion mailing list
> patterns-discussion AT cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/patterns-discussion