The Fallacy of the Artificial
Two unrelated incidents prompted this post.
First, the State of New Mexico – once again – announced that it had wasted $5 million on a software project, blaming the vendor (of course) and citing some Chaos Report derived numbers about 25% of all projects being abandoned before completion and 40-50% more delivered late, with cost over-runs, and diminished capability. As if this made it OK somehow.
Second, I had occasion to look at Herbert Simon’s Sciences of the Artificial again, and again was struck by his insistence that thought was nothing more than the operation of a “physical symbol system.”
These incidents led to thinking about the why of software failures and the failures of software successes. By the latter I mean applications like SAP that are overwhelmingly successful in the market place while generating equally overwhelming levels of dissatisfaction. [Note: I use SAP as an exemplar because of its name recognition – not because it is any better or worse than all the other database grounded, centralized control, integrated, systems in the marketplace.]
I believe the roots of the problem arise from a hidden assumption, an unexamined belief and worldview. In very broad and general terms, it is the belief that it is possible to replicate any “natural” system with a computer/software-based “artificial” system.
In Simon’s case the computer/software artifact (artificial system) can literally “think” in exactly the same way a human (natural system) “thinks.” In the case of SAP, the computer/software artifact (artificial system) replicates the structure, processes, rules, and activities of the enterprise (natural system).
This belief has deep roots – in nineteenth century Physics and the belief in that discipline that the entire Universe and everything in it was a machine – a complicated, but deterministic and Law driven system. Academic Business and academic Computer Science disciplines assumed this ‘Truth’ and derived all their other theories from this basic assumption.
The Practice – people actually writing software and developing systems – are not as susceptible as academics to this belief system. Again, for at least two reasons: pragmatics and tacit knowledge.
To illustrate the pragmatic issue, consider something as simple and direct as digitizing a sound. A “natural” sound is analog – visualized as a waveform. To digitize a sound, you sample this waveform – take a series of measurements of the height or depth of the waveform at specific time intervals.
The result is lost information – the area under or above the waveform as bounded by the vertical and horizontal lines of the sample rate. If we wanted to truly replicate the “natural” sound we would have to have an infinite sampling rate – something that is theoretically (remember your calculus) possible but not practical. (Audiophiles still insist that the best digital music is inferior to analog music precisely because of this sampling/lost information problem.)
If Simon’s physical symbol system were capable of replicating human thought, it would require an infinite “sampling rate” to account for all the analog (think chemistry and hormones, not to mention the idea of an ’embodied mind’) aspects of “thinking.” This might be doable in theory – but not in practice.
Tacit knowledge is everything you have learned, and learned to communicate, while practicing your profession that never made it to the textbooks. It is left out of the textbooks because even though you “know it” and can even communicate “it,” you cannot articulate it in the artificial, formal (logical, mathematical), propositional form necessary to get past academicians – acting as editors or reviewers of journal articles and textbooks.
(There is a body of research, mostly from studies in the area of knowledge management, suggesting that tacit knowledge outweighs “book knowledge” by a factor of 3-5 and a scope (breadth of knowledge) by an even greater amount.)
Champions of the artificial must repudiate both the pragmatic and the tacit arguments. They generally do so by claiming irrelevance, but more often the issue is just ignored or papered over with marketing. We see this in the way that the “mainstream” responds to anything that appears to be a challenge to the received wisdom from academia and vendors.
“Agile is only good for small, trivial, projects!” “Objects are nothing new, we have been modularizing our software since the sixties.” “RUP is Agile.” “We do not need to reinvent Computer Science, just make it ‘sexier’ and introduce it in grade school.” This is but a small sample of the multitude of comments about innovations arising from the practice. All reflect rejection and denial of validity, or cooption and dilution as a means of eliminating the threat.
What to do?
I would begin with a new metaphor. Consider Data from the Star Trek franchise. Advocates of the artificial claim that it is possible to replicate a human being with a machine and Data would be a proof of concept, despite his “flaws” – like a lack of emotion which is in any case irrelevant. It is, they would insist, possible to eventually emulate the success of the Tyrell Corporation and create replicants that are absolutely indistinguishable from humans – just like Rachel and Deckard.
Instead think of a composite system – like the Six Million Dollar Man, or Iron Man. At the core they are a natural system – a human being – that is simply augmented with the occasional artificial component.
Elaborate this metaphor with a little bit of General Systems (everything is a system and a system is nothing more than a set of elements and the relationships among them); a little bit of Objects (objects are differentiated base on their behavior); a little bit of Design (design is the informed choice to add, delete, or alter components or relationships in a system); and then connect various agile practices, and you will have a very real alternative to the conventional wisdom of software development.
This alternative ground will provide an approach to software/systems development that leaves the artificial ideal to the academics and provide a vastly superior alternative for clients like the State of New Mexico to actually achieve their objectives.
Dave West:
David M. West Ph.D
I can help your organization realize its vision and attain its strategic goals. I specialize in innovation, adaptation, sustainability, IT integration, and organizational design/culture.
An eclectic background with expertise in computer science, business, design, complex adaptive systems, cultural anthropology, and Asian philosophy provides the foundation necessary to successfully analyze, model, and design solutions for almost any enterprise problem.
With thirty years experience as a software professional and consultant to large and small enterprises, including members of the Fortune 500, in ten different countries, I have the knowledge, skills, and experience to help your organization achieve its goals.
It is essential for everyone in your organization to understand the concepts, values, principles, and practices of any new approach. Rote adoption is insufficient for success.
I can help you devise a comprehensive – yet simple – model and understanding of your enterprise. Help you use that model to realize your innovation, adaptation, and sustainability goals; while rethinking and integrating IT services in a way that will provide direct and obvious benefit while drastically cutting costs.
I believe in people, not technology or methodology. My goal is to assure your employees will become knowledgable and proficient. Proficiency for solving the demands of today. Knowledgeable so they can adapt, extend, and create the practices your organization will need tomorrow.
My consulting work is done under the auspices of Transcendence Corporation.
This is my personal site and the blog is a collection of occasionally posted opinion pieces. The articles page provides access to selected papers or presentations that may be downloaded. They are all subject to the creative commons license.
I am a lifelong reader – almost anything – and collector of books. The bookshelf page provides access to the list of over 3400 titles currently on my shelves. This number will grow by about 3-5 per week. If you browse the list and are inspired to recommend something that it looks like I might enjoy or have a comment about a particular book, I would like to hear from you. In the future this page will be interactive, but for now, please email your comments and suggestions.