Rediscovering the Telos of Systems
The first questions must not be forgotten as we optimize the world around us
I had an interesting exchange with a friend once. I asked an open-ended question: “what is the most efficient transportation system”?
His one-word answer came straight from the current common-sense consensus: “trains”.
My perspective differed, informed in part by the following:
I have experienced trains as part of a daily commute in a large US metropolis, and also as a part of days-long trips in developing countries
I have also experienced various other forms of pubic transport, including buses, subways, light rail and pedicabs
I had purposely left the term “efficient” undefined
When my interlocutor heard the word “efficient”, it flipped a switch in his mind that placed the context of the question into one of raw energy consumption per pound/mile. However, that was not what I had asked. I simply asked what is the most efficient.
A thinking person’s response to that question should be, “when you say efficient, what resources are you optimizing efficiency for?” You see, efficient doesn’t describe the resource being conserved, only that some resource should be prioritized in the equation.
A transportation system, for instance, is a network of moving pieces carrying some payload or other to its destination. Viewed in this way, efficiency must take various factors into account, account for resources to be variously conserved or consumed, and make prioritizations among them to make in order to select an ‘optimal’ solution.
For instance:
No one can tell me that grocery shopping via city bus in Indonesia is ‘efficient’ in any terms except in personal cost or fuel consumption. The bus ticket is cheap, but once you load up the bags, walk to the bus stop, ride somewhere near your destination, walk to market, back to bust stop, ride close to home, etc you have spent half your day obtaining basic provisions. Forget the idea of picking up anything frozen, and someone might have kicked your eggs on the crowded bus. From a air quality perspective, the local buses (in my experience) were not well maintained and you needed a shower after picking up your food.
A train can be highly time and fuel efficient when moving objects or people between two predetermined points. The problem becomes when neither of those points are the intended start and end points of those objects. Other transportation networks are required at each end to supply the final mile of every transit. Thus any valid efficiency calculations must take those networks and the interchange at each point into account. The closer together the 2 points linked by the train line, the less the peak efficiency of a train affects the total efficiency of the system from start to finish.
Returning to the train example, there’s a further effect that must be taken into account: that of a circuit-switched network. When the Internet first arose we all had dial-up internet access, which ran a packet-switched technology (ethernet) over a circuit-switched network (telephone wires). There’s a reason no one does this anymore: it was exceedingly slow at moving large numbers of distinct objects as each object had to be filed into a queue and transmitted via audio bits to the other static endpoint. The line couldn’t be used by another communication while the one already started was still working on its task; 2 computers in the same house had to take turns at dialing up. Similarly, a train line is mostly a single line of bandwidth that dominates the line of its transport of objects until it has cleared its destination and released the line for re-assignment to another train. A collision or breakdown on such a network can cause the entire system to stop moving for hours (I have experienced this more often than a car crash or breakdown, despite 1000x more time in a car). High-level controllers with complicated schedules and sections of side-rail tracking mitigate the impact of this somewhat, but because of this core limitation, trains remain at their most efficient (taking all resources and costs into account) when moving large objects a really long way from point A to B.
Now consider cars: cars operate similar to packet-switched Ethernet in that a single object can be inserted into a stream of other objects, and provided there are not collisions along the way, each ‘packet’ (or automobile) can arrive at its intended destination with minimal marginal impact to the entire body of other objects. A collision represents a slight cost to the system at large, a marginal one to each participant in the system except for the people in the crash itself. To be sure, there are other costs and resources to be considered, such as number of lanes in a highway, the likelihood of collisions causing traffic congestion, and the sum of resource consumption of the system as a whole.
My ultimate objective with this article is not to champion cars over trains or make any specific judgment call, but I am using the discussion about this topic to highlight identification of a system’s purpose.
OK, now that we’ve reframed what ‘efficiency’ is, and asserted that it’s not a simple emissions calculation or mindless talking point, what is the correct way to prioritize and judge efficiency of any human-created system?
I assert that you cannot even begin to approach the question without identifying the ‘telos’ of the system. Telos is a term popularized by Greek philosopher Aristotle, and means ‘the ultimate end (or goal) of a goal-oriented process’. If you don’t know the goal, how do you reach it, or even know when you’ve reached it?
Returning to transportation systems as our example, what is the ultimate purpose of a transportation system? Is it to save fuel? Is it to limit impact on the environment? Is it to save time? Is it to deliver goods? These are crucial questions to ask when considering a system.
To be clear: none of these questions are unimportant. Yet there is clearly a hierarchy to them. Among competing benefits and resources, which ones are to be optimized?
My answer in the case of a transportation system is a firmly humanitarian one: it is to serve human needs and desires.
Are we transporting people to work? If so, what system can move the most people to their ultimate destination minimizing time and complexity while maximizing reliability?
Are we transporting cars across the ocean? If so, what system can move them with maximum speed, minimizing space and damage to product along the way?
Are we transporting information? If so, what systems bring the information to the people who need to see it most reliably, securely and with clarity?
Are we transporting market value across systems of currency? If so, how do the various systems of currency interact, and what solutions best transit those interactions with a minimum of time, cost and slippage? In currency transactions, pricing signals are a critical data point that must return along the path of transport to the consumer. How efficiently and accurately does pricing transit the system?
Are we transporting fuel from extraction to refinement? Asking better questions when discussing energy transport, we are transporting some form of potential or actual energy to those requiring or requesting it: What is the nature of their need? Which form of fuel best meets that need, and after that is identified, which means of transport is best suited for that fuel in terms of safety, cost, speed and reliability, minimizing loss?
As you can see, my questions prioritize the human experience. After all, if we’re not building systems in service of humanity, why are we even building them? If we’re building a slaughterhouse for the benefit of the cows, the optimal building is a grassy field.
The questions framed above are what I would call ‘primary questions’ for systems engineering. In the case of a transportation system, I believe that if you evaluate based on its ‘telos’, its ‘raison d’etre’, its purpose, you’ll find that the fuel efficiency is always a secondary concern; it’s one of the costs, but never the primary cost—if it were, the optimal transportation system would be not to transport at all. We would call that ‘stasis’, root of the word ‘stagnant’, which I’ve never heard used in a positive way.
So that’s why I disagreed with my friend’s bland assertion about train efficiency.
If we are to think for ourselves and evaluate available information and our options with our own minds, we need to first identify the first questions and ask those. Failing that we will remain in a self-perpetuating loop of mistake, mismanagement and failure.
While this is true for us generally, it is especially true for systems we create or use. If we are to discuss these systems, their outputs, their effectiveness and their optimization, we have to focus that discussion dispassionately around the purpose for which the system exists. After all, people exist in their own right as individual, autonomous beings. Human systems, in contrast exist only at the behest of people and only in the service of a goal on behalf of people. Progress can only be made by first recognizing the telos.