Computation Under Uncertainty

Nicole Starosielski’s text “Thermocultures of Geological Media” [1] talks about a “culture of purity”, where our cultural imperatives have resulted in us choosing to only use pure metals and other materials in our electronics. Her main critique of this is that the purification process of metals such as copper and quartz is very energy intensive, and that developing technologies which would utilize metals of a lesser purity would result in media with a lower environmental impact. She also says that this kind of technologies, which probably would have to compromise speed and accuracy, would “…significantly alter the form of existing media texts and technologies”. I find the idea interesting but at the same time I finding it very difficult imagining how computation would work in such an inaccurate system seeped with uncertainty.

Our current models of computation rely heavily on reproducibility and stability: bits will not flip randomly (except in extreme cases) and code will always be executed in the same way. Given the same inputs, a set of commands will always result in the same outputs. Introducing uncertainty into this system would not only cause “subtle variations across media objects”, but result in bugs, crashes, corruption and loss of data. Maybe some new computational models could be developed which could better deal with randomness (quantum computation comes to mind), but currently one of our only methods of dealing with uncertainty in computation is by verifying the validity of data and performing recalculations as needed. Already a small amount of uncertainty could cause huge numbers of unnecessary CPU cycles, which across the millions of computers in use today might very well negate any environmental benefits gained from the use of impure metals. And with a high enough level of randomness, even these methods would no longer work and the system would come crashing down under the pressures of uncertainty.

The word “uncertainty” has a negative connotations, even though it is non-partial in the quality of the future it describes. Uncertain events might as well lead to unexpected successes as to devastating failure, but our negativity bias makes us focus and lay greater importance to the latter and makes us uncomfortable in situations where we have too little control of the future. Seen through this lens, the strive to control our future is a very natural trait. In fact, I believe one way to look at the evolution of organisms is as a struggle for control over uncertainty. Existence is an extremely complex system which humans and animals alike have evolved to navigate as best they can in the fight for survival. Excessive uncontrolled futures results in accidents, broken bones, death and the extinction of species.

Ultimately, I enjoyed this thermal perspective on media that Starosielski’s text gave, but question the validity of her thoughts on purity of metals and the possibility of moving away from them in our electronics.

1. Starosielski, Nicole. Thermocultures of Geological Media. Cultural Politics (2016) 12 (3): 293–309.