• Home
• Paper(pdf)
• 2004 Slides(pdf)
• 2009 Slides(pdf)
• Author
• Site Map

Steering the future of multicellular computing

Without question the evolution from single-cell to multicellular computing is well underway. Four Architectural Principles can help smooth the continued transition.

The complex dynamic interactions that drive the evolution of computing toward multicellularity have a life of their own. The next billion personal computing devices will be mobile.  They will collaborate with increasingly complex multicellular server-farms like Apple's Siri, Google, Facebook, Twitter, and various eCommerce sites, ...and with each other.  The next million multicellular server entities will increasingly be focused on the kind of "Big Data" to be found in the trails left by users of the billions of personal devices. It'll be a digital jungle out there in which every new species quickly gives rise to malware specialized to prey on its weaknesses.

It is fundamentally impossible to foresee the details of the future evolution of computing. Yet the general principles of interaction and collaboration (or competition) are amenable to some steering via judicious choices of basic architectural priniples. The question is: what roadmap should we steer by?

This site argues that the four principles discussed here serve as biological metaphors to help us make multicellular computing systems more manageable.  The complexities faced by multicellular computing are not unlike those faced by multicellular life. We are already seeing evolution in advanced multicellular computing environments toward all four biological principles. That fact demonstrates that such principles are indeed general enough and useful enough to provide us with useful insights.

If we are more deliberate and disciplined about making the transition toward multicellular computing, we stand to make the transition smoother and perhaps even quicker. To that end, the computing community needs to foster, in our education and in our cultural transmission of received wisdom, multicellular attitudes such as:

• What goes on within any given computer is far less important than what goes on between computers.
• Software that is inside a computer may be valuable, but when it is moving between computers it must be assumed to be dangerous.
• General purpose computers will become ever more troublesome -- specialize them
• The meaning of messages should be determined by the receiver, not the sender
• Individual CPUs are, or should be, expendable and be prepared to commit suicide neatly and safely when compromised.
• The multicellular computing organism is paramount. When a single computer offends, cut it off. Better yet, empower the individual computers to sense problems within themselves that cause them to cut themselves off.
• If you want to create a new multicellular computing organism, think about what its stigmergy structures might be. Conversely, if you see the emergence of a new kind of stigmergy structure, e.g., a public data base wrapped in a Web Service interface, think about what multicellular organism(s) will emerge around it.

The stakes are high. We are just at the beginning of the computing equivalent of the Cambrian explosion - the period of roughly forty million years ending about 500 million years ago  - in which a fantastic flourishing of novel multicellular biological organisms occurred. Most of these novel species remain, if at all, only as fossils preserved in the Cambrian mud. Given how rapidly computing has recapitulated a couple billion years of single cell evolution, perhaps we can recapitulate the forty million year Cambrian period in the next couple of decades. If so, it behooves us to pay close attention to how multicellular computing is progressing. Forward looking businesses and technologists will have much to do in the next few years. Great opportunities will be available for those who anticipate (or lead) the successful architectures and great peril for those who stick their heads in the sand or back the wrong horse. The choice is between winning and being a fossil in the mud.

It is already clear that there are competing organizational strategies, e.g., completely decentralized P2P communities such as bot-nets at one extreme versus centrally controlled hierarchical IT structures at the other extreme, with Map/Reduce server farms and Grid architectures emerging in between. And there are competing notions of how messaging should work, e.g., heavyweight SOA (which may sink under its load of complexity) versus lighter weight AJAX (which may fail to scale well). We also have competing vendors, each pushing the idea that what they sell is just what is needed for the future. Don't believe them!

Most of all, we have competing visions of how the evolving computing infrastructure interfaces with evolving business and cultural systems that use the computing systems. That is, different visions of what happens where the silicon meets the flesh. The digerati imagine cyborg-like symbiosis in which people are festooned with wearable or implantable computers that aid them in all manner of normal social interactions (or, perhaps, insulates them from normal human interaction). There are neo-Luddites who imagine that we can “keep computers in their place” which, presumably, is somewhere far from those who hold such views. And there are neo-fascists (or those who fear them) who imagine that every move we make and thought we voice will be surveiled by ever-present cameras, microphones, and RFID tracking devices. What all of these ideas share is a lack of understanding that the evolution of computing is already beyond our control. The coevolution of complex computing systems interacting with human social and economic systems already has a life of its own.

We should also remember that the processes that give rise to emergent multi-level complex systems do not stop, or even pause, at one level of emergent behavior. Before one level is fully fleshed out, another emergent metalevel likely will be forming based upon collaborations between the units at the still immature prior level. And so it is with today's multicellular computing. Multicellular “Selves” made most visible by their stigmergy structures are collaborating to form even larger and more complex systems. We can think of some of the more public multicellular systems such as Google or eBay as organs in some larger “body” we have not quite envisioned yet. That is, some mashup may already be evolving in Web 2.0 that combines a Google plus a BitTorrent, plus a Skype, plus who knows what else acting as “organs” to achieve some larger goal. All that is certain is that each stage of multi-level emergence will surprise us.

Contact: sburbeck at mindspring.com
Last revised 5/16/2012