Marketing: The marketing function (which in the broad sense encompasses everything from positioning and competitive strategy to market research, not just selling) could not be more different. The high-touch/low-touch distinction is an obvious one, as is the difference in the length of the sales cycle and the revenue balance between initial buy and post-sale. But there are more subtle ones. Complex systems architecture companies must orchestrate a large coalition of companies to deliver a solution, so reputation and collective positioning matter more. Volume operations work primarily by pull rather than push, so branding and stand-alone segment positioning is key (rather than coalition positioning). Marketers who are good at one are rarely good at the other.

Innovation: Moore's book is based on the technology adoption lifecycle and the category maturity cycle. He has a thorough discussion of 15 different types of innovation, from initial disruption to end-of-life innovations and harvesting. The big point, which he elaborates painstakingly, is that all types play out very differently under the two architectures. For example, in the first kind, disruptive innovation, which comes at the birth of a category, complex businesses disrupt via technlogy innovations (example, Oracle in abstracting databases from mainframes), while volume businesses typically disrupt through business model innotavtions (for example, Kindle and free Wi-Fi downloads).

Profitability Constraint: Each business architecture also has a very different gross margin model (set by Wall Street expectations for that architecture). Roughly, "messy cost structure, large margins, low velocity" versus "lean cost structure, tight margins, high velocity."

The Free-Premium Stack: Every module vendor has grand visions of a stack where a free (possibly open source supported) base props up an SMB volume operations "premium" version, and a "super-premium" one for large enterprises. Unlike the integration challenges of 1.0 technology (like setting up special customizations to copies of Windows sold to a Fortune 100 company), there is a secret hope at work that SaaS, WebTops and XML-based SOA integration standards like RSS, integrating into a complex system will require no extra customer-specific work. Even if you are selling a 10,000 seat license. Holy grail, yes, but conceivable in a way perpetual motion machines are not.

The "Open Platform": If module vendors are dreaming low-touch integration dreams, the platform vendors have similar hopes. With the evolution of open APIs and SOA as legitimate technology strategies, there is the hope that integration work will not be demanding on the platform end either. Perhaps it can be as simple as the CIO's staff checking off a bunch of boxes for supported modules, and IT ecosystems simply culling out module species that don't talk the main integration languages.

The "Sip" Sales Cycle vs. the Mega-Deal: Another big reason to believe the Second Moore Law can be broken is that it is now possible to engineer sales models where even very complex offerings can be sold with an entry-purchase in the volume regimes. Cloud-metered by-the-sip sales in fact. Dell sells to large enterprises by the truck-load. Amazon EC2 can sell by the box-hour. The goal is still millions of dollars from a single customer, but you don't get that with a starter sale of hundreds of thousands. You can get started with one unit. Stowe Boyd just pointed out a great Niall Kennedy piece on the power of flexible pricing, and the implications for business model architectures are seriously significant. Google actually charges by the type of database access command.

Open Source as an Integration-Operations Game Changer: Moore has a quick treatment of open source, but I believe he fails to realize its true significance, which is this: much of the operational complexity of complex systems business architecture comes about because it is attacked top-down within a single umbrella sale or engagement. When your offerings are based on open source, much of this complexity is simply shunted out to the open marketplace, where it is solved by market/crowdsourced mechanisms rather than complex operations. If you think about it, the idea that complex systems integrations offer Quality of Service (QoS) guarantees within the SLA (Service Level Agreement) is a specious one. Well-organized open source communities beat formal QoS and SLA models hollow on specific fronts like bug fixing, security threat resolution and so forth. Slowly, open-source coordination models are taking on more and more of the integration challenges through their market-like models, and winning.

The Moving Target Effect: The customer isn't standing still in all this. If "bazaar" models are taking over the open economy of free agents via open source, the insides of companies are also going from cathedral to bazaar architectures. This makes the complex-systems architecture slightly less necessary. Thomas Malone's "Future of Work" (2004) chronicles many of these internal changes, and things are evolving so quickly, his book is is already outdated.

The Invisible Hand and the Single Wringable Neck: Perhaps the biggest reason complex systems architectures have been required in the past is that CEOs of Fortune 100 companies have demanded a "single wringable neck." IT systems are now so large and complex, and so critical to operations, that if something bad happens on a large scale, the CEO needs to be able to call up a single powerful VP level account manager (or even the vendor CEO) and yell. What happens when there is no single wringable neck? Everybody expecting the mix of automated plumbing systems and market mechanisms inside and outside the firewall to resolve everything? This is perhaps the biggest unknown. In a sense, the single wringable neck is replaced by the invisible hand. As with Obama and the financial crisis, there are people to blame, but no single person who can be called and yelled at to get things fixed when a complex system breaks. My opinion is simple: it is evolution. We are now at a point of complexity in IT systems, where limiting things to levels where "single wringable necks" can exist is to limit business itself. Individual large businesses must be willing to operate as mini-economies rather than controlled systems. You get the threat of meltdowns and messy recoveries, but you also get the promise of a brave new age of next-generation business model species. The dinosaurs are dying. The mammals are coming.