Speed is a very important thing in late modernity. It is one of the few things that the various industrial complexes we run our societies on tend to treat as intrinsically valuable (money and size are two others). Paul Virilio even theorizes contemporary technological society as based almost entirely on what he calls a dromology, an organizing logic of speed that reduces all space, all action, to a series of durations that are always too long.
Even if Virilio is correct on this point, speed, unlike money, is not quite a universal commodity-fetish capable of standing in for all other values. We late moderns have no problem conceptualizing a variety of costs of speed. By "costs" I mean tradeoffs between speed and other valuable things, tradeoffs that we must almost always accept if we want to make something faster. (By contrast, while we can conceptualize a "cost of money," we can only express it in terms of money, as lost interest or an amount in some competing currency.) However, we do not always think about whether or not we should pay these costs, or about what our high valuation of speed as a culture might be costing us in general.
For these reasons, I thought that it might be fun to play at developing a taxonomy of these costs. Here is a preliminary list:
- latency and smoothness (as of mathematical functions): Massaging data into a nice, meaningful form that looks like a nice, differentiable function often requires looking at more than the most recent point(s). The further back in time one looks, the greater the latency, by definition.
- latency and error rate / jitter: In a closely related point, demanding reduced latency in any signal makes it more vulnerable to disruption by noise.
- speed and mechanical failure: The faster you make a device go -- whether in terms of rotational / translational velocity or processor cycles per second -- the more stress it places on the device, through frictional force, pressure, heat, etc., and thus increases the number of instances of failure, on average, per unit time.
- speed and accident rate: Similarly, the faster a person is going -- whether in terms of velocity in a vehicle or task completion -- the more likely it is that they will screw up. There also tends to be a similar relation between speed and accident severity, in that these screwups become not only more frequent but also more consequential.
- development/design/construction speed and quality: As expressed in the old adage: "good, fast, cheap: pick two."
- development/design/construction speed and safety: A close corollary of the "speed and accident rate" relation. The faster you make something, the more likely it is that someone will get hurt doing it, or that it will fall down at an inopportune time after you are done.
- speed and accuracy: In a wide variety of senses of "accuracy," from accurate responses to posed questions to how quickly one can grab a small object. In human-computer interaction, this relation is often modeled using Fitts's Law, which predicts that time to complete a pointing task increases as error tolerance decreases (and distance to the target increases). In psychometrics, it is typically just called the "speed-accuracy tradeoff."
This group of costs could be conceptualized as representing an inverse relationship between speed and reliability.
- time complexity and space complexity: Doing stuff faster usually requires more space. This vocabulary comes from algorithms design; algorithms designed to work using very little memory generally run slower than algorithms that are allowed to scribble willy-nilly all over the place. However, this tradeoff often holds true outside of computing environments as well: organizing a 25-sq-ft. room full of boxes takes more time than organizing a 100-sq-ft. room containing the same number of boxes.
- velocity and fuel usage: Making a vehicle go faster requires more fuel, both from the increased fuel required to accelerate it to speed and from the increased fuel required to maintain that speed against increased frictional and drag forces.
- speed and heat: in a closely related note, making things go faster (again, even in terms of processor cycles per second), generally makes them waste a lot of energy as heat.
- rapidity and expense: although "time is money," and doing things quickly can be lucrative, doing things faster often, perversely, costs more. (Finding the profit-maximizing point in that relationship can be very important for many applications.) Less time for planning means that cost-efficient solutions will rarely be found, workers may need to be paid overtime, etc. ("Good, fast, cheap" again.)
This group of costs could be conceptualized as representing an inverse relationship between speed and efficiency.
And finally, one other cost that I cannot readily fit into either of the above categories:
- speed and stress: All other things being equal -- and assuming that you start above some fairly low threshold of speed -- doing a task faster (or, similarly, doing more tasks in a given period of time) is more stressful than doing that task more slowly (or completing less tasks in a given period of time). It is more physically and cognitively demanding, more fatiguing, and creates a greater degree of anxiety.
Other thoughts?
Recent Comments