Laiserin's Lemma—Power to the People
(lemma: a short theorem used in proving a larger theorem)
Jerry Laiserin

During the first third of the last century, a now-forgotten techno-pundit predicted that telephone operators would soon comprise a majority of the industrialized workforce—to manually route the burgeoning volume of local, long-distance and international phone calls. However, the innovation of direct dialing—with its international access codes, country codes and area codes accessed through the simple phone keypad interface—proved that prediction false by moving control over telephone connectivity behavior out of the hands of trained operators and into the reach of end users. Today, a comparable transfer of control is shaping up in design and analytic software—shifting power from software developers' pre-programmed or custom-coded functionality to end-user interaction and flexibility. Whether identified as multi-agent systems, expert systems, genetic algorithms or adaptive hybrids of these techniques, new and emerging software tools will empower users to manipulate or guide the behavior of their software—directly through simple interfaces, yet without programming.

Software with behavior that is modifiable by end-users is more likely to suit the constantly shifting nature of the work at hand than is fixed-function software in which programmers (by definition closer to the code than the tasks for which they're coding) try to out-guess what end-users really need. Early commercial and research examples of this trend range from: the drag-and-drop workflow design capability in OpenText's LiveLink for Construction; the work of the SmartGeometry Group with Bentley Systems' infelicitously named "Generative Components"; diverse research on multi-agent systems for design and construction at universities such as Stanford (PDF) and Sydney; and the knowledge-based adaptive program modeling in Cyntergy Technology's Thumbprint Collaborative Program Management (CPM). The bottom line? Workers in AEC/FM, plant/process and infrastructure businesses will soon enjoy software that adapts itself to the way the users work, rather than users changing their work to fit the hard-coded constraints of their software.

At the Key Centre of Design Computing and Cognition, University of Sydney, Mary Lou Maher, Pak-San Liew and John S. Gero have explored "An Agent Approach to Data Sharing in Virtual Worlds and CAD"—work published in Construction IT: Bridging the Distance, Proceedings of the CIB W78's 20th International Conference on Information Technology for Construction, edited by Robert Amor at the University of Auckland, NZ, and published by the University of Auckland as CIB Publication 284. As CIB itself decodes that breathtaking mouthful: "CIB is the acronym of the abbreviated French (former) name: 'Conseil International du Bâtiment' (in English this is: International Council for Building)." Founded in 1953, CIB in 1998 expanded its name to International Council for Research and Innovation in Building and Construction, but kept the CIB acronym. Truly international in scope, CIB is headquartered in the Netherlands. "W78" is the CIB working group that addresses information technology (IT) issues in construction. CIB W78's 20th annual conference was held in April 2003 in New Zealand, to which many attendees' travel (including my own, as a scheduled keynote speaker) was curtailed due to concerns about the SARS virus then active elsewhere in the Pacific region. The conference chair, Professor Robert Amor, who had invited me to speak about industry trends re construction IT, was kind enough to provide me a copy of the proceedings, despite my travel cancellation. The quality of the papers is outstanding, fully justifying the uniformly positive feedback I've heard from folks fortunate enough to have attended. Even in such a distinguished field, I found the Maher/Liew/Gero paper especially appealing.

What the Key Centre researchers have proposed is a system for real-time, interactive design collaboration consisting of:

> an intuitive, web-based, multi-user interface based on a videogame engine
> a parametric CAD back end for building information modeling (BIM)
> linked by a system of multiple agents (I think of agents as a kind of lightweight, distributed and adaptable middleware).

Many folks around the world are exploring the use of videogame technology and virtual reality (VR) for visualization of and navigation through design models. Other papers at the CIB W78 conference addressed this technique, and I reported in IssueTwentyOne's LLemma on an IFC-to-Game proof-of-concept project by Paul Richens and Simon Ruffle at the Martin Centre CAD Lab in Cambridge (however, see note below about the imperiled future of IFC2game). Multi-agent systems let users manipulate and edit the building model without the difficulty-of-learning and difficulty-of-use involved in programming a heavy-duty CAD system. This can be especially valuable in challenging design and construction projects that involve complex geometry and difficult coordination and interference-checking problems (as demonstrated in the research of John Haymaker (PDF) reported at last June's annual installment of the Stanford/CIFE Summer Program.

This kind of end-user programmability and adaptive program behavior is rapidly spreading from labs and academia into the commercial realm. For example, form*Z, many architects' front-end design tool of choice, now offers a STEP Translator plugin to efficiently import and export parametric objects and their controls. Over at CAD stalwart and BIM innovator Bentley Systems, the adaptive buzz revolves around their so-called "Generative Components"—user-programmable parametric controls for integrated and interactive geometric design explorations. Popularized by the UK-based SmartGeometry Group, whose kick-off summer school I attended in Cambridge last year, these adaptive design modeling technques started life as a "Custom Objects" research program by Bentley Chief Scientist Robert Aish (a "been-there, done-that" kind of guy, with design software experience dating back to the beginnings of ABACUS at the University of Strathclyde and the legendary RUCAPS—erstwhile contender for ur-BIM laurels).

Adaptive program behavior need not be limited to design geometry and modeling. One of the most intractable problems facing large AEC organizations today, especially multi-location, multi-building owner-operator organizations, is the efficient management of large-scale building programs across multiple collections or portfolios of projects. Traditional approaches apply canned scheduling tools on a project-by-project basis. Rules-based, process management approaches offer considerable improvement by applying project templates across multiple, rolled-up projects. However, for many program management situations, template-based processes merely shift the user's problem from project-by-project custom scheduling to template management. A promising adaptive modeling approach to this problem that is gaining early commercial acceptance is ThumbprintCPM from Cyntergy Technology in Tulsa, Oklahoma.

I've written in some depth about Thumbprint/Cyntergy, Buzzsaw, Constructware, Meridian Project Systems' Proliance and OpenText's LiveLink for Construction (in the context of emerging trends in process modeling generally) in my recent article entitled Work in Process. This originally appeared in Cyon Research's paid-subscription, print-based A-E-C Automation Newsletter, edited by the estimable CAD journalist Randall Newton (although there is an online component to A-E-C Automation, most articles such as my "Expert's View" column are—understandably—accessible online only to paid-up print subscribers; which is why I'm hosting a version of my article here on Laiserin.com).

Whether considering game/VR design navigation interfaces fronting adaptive agent middleware (as in Cambridge CAD Lab's IFC2Game), user-programmable, adaptive geometries (as in Bentley's Generative Components), or adaptive process modeling for large programs of projects (as in Cyntergy's ThumbprintCPM), I think that adaptive software techniques have the potential to transform the use of software as profoundly and rapidly as DTMF-based direct dialing transformed the use of telephones. Direct dialing shifted attention from the logistics of the telephone medium to the content or message of the individual phone call. In so doing, this started shifting the control of telecommunications power to the people. Adaptive programming has the potential to repeat this power shift in software. The question we in the AEC/FM, plant/process and infrastructure industries need to ask ourselves now is this: "will we embrace adaptive software technology to help us shift our focus away from the technology medium and back to the design/construct/operate message?"

Let me know what you think.


Editor and Publisher, The LaiserinLetter
Analysis, Strategy and Opinion for Technology Leaders in Design Business

NOTE RE IFC-TO-GAME: The previously reported proof-of-concept for "one-click" export from a CAD model to a video game (via IFCXML and an IFC model server) relied on the IFC Connector available in Microsoft Visio (2002). Sadly, Microsoft inexplicably chose to remove IFC support from current and future versions of Visio. I've suggested—with no success to date—that Redmond outsource Visio's IFC support the same way that Autodesk outsourced IFC compliance development for Architectural Desktop to Germany's GEM Team Solutions. I would urge anyone with an interest in interoperability among AEC design and analysis tools to express support for restoring the IFC Connector to Visio by writing directly to Richard See, Program Manager - Advanced Technology Teams, Business Tools Division, Microsoft.

JL