Moment of Inertia, Issue #11 | Man vs. Machine

We've never relied more on the data sent to our devices, but it's important to note that while the data and its distribution itself is digitized and automated, the collection of it rarely is.


Man vs. Machine

When was the last time you drove somewhere without GPS navigation? Even when we make trips between places we go to all the time (e.g., work or a friend's house), we'll take out our phone and ask for directions so as to get the most optimal route possible, reducing time, or tolls, or mileage .As the article about "Haberman", Queens in this weekend's M&D shows, though, there can be strange disconnects between the "ground truth"of a digital map and the "truth on the ground" about a place described by that map. For all the faith we place in the reliability of our digital systems, though, they're certainly not infallible and even the smallest of human errors can lead us dramatically astray(when we otherwise might have been fine, sans computer assistance).

Human error interfacing with these systems is only part of the issue, though. The computer can only execute programs to the best of our ability to enter the necessary prerequisites. The Haberman situation brings to light a much more important issue when considering the interface of Man and Machine. With the exception of the last quarter-century or so, all of human knowledge that we can now access on a computer or smartphone didn't start out on computers and smartphones, it was physically recorded somewhere - be it in a book, on parchment, or etched into a stone tablet. Furthermore, this information then had to be digitized, a process that (more often than we might think) requires A LOT of human input. If we can't type the right address into our GPS, why would we be any better at interpreting subtle differences between hand-set typography?

It's also worth noting there's a lot of human knowledge out there that we haven't gotten around to digitizing and/or making accessible. A few years ago I did some Fermi estimation on Wolfram | Alpha regarding the state of human knowledge and found the following relations:

  • About 10% of all the information on the internet (let's take this to represent all digitized data, since we're doing Fermi estimation anyway) is indexed and searchable (i.e., you can find it through a Google search).

    • Where's the other 90% of internet information? That's what is otherwise known as the Deep Web (of which the infamous Dark Web is a part, which is not dark in color scheme). It's just a bunch of websites that you can't get to through a google search or conventional URLs. (It's also worth noting that the source I cited in that hyperlink actually says my Wolfram | Alpha Fermi estimate was an order of magnitude or two too high.)

  • The more astounding thing, though, was the estimate that only 10% of all human knowledge is even digitized and available online in any form. That means a Google search is, at best, uncovering 10% of 10% of all human knowledge (aka, 1%).

All of this assumes that data is being gathered and cataloged in a more or less random and consistent fashion, which is very often not the case. As the saying goes, "history is written by the victors" and the majority of peoples and civilizations across time have not been the victors. The result of this systematic imbalance of information is a data-logging system that only makes it harder as time goes by to rectify these knowledge gaps that have formed.

This isn't to say we shouldn't trust the driving directions our phone gives us, but rather that, when we look at information - especially as we look further and further back in time - we have to recognize that there were no digital sensors to record peoples' movements, thoughts, and actions. Our contemporary base of information is incredibly thorough, but it is built upon a foundation that becomes ever more tenuous as we go back in time - and sometimes we're forced to look back and glean what knowledge we can to inform our decisions in the present and future. Just understand that that information (and indeed, all information to an extent) interacted with a human or human-supervised process at some point, and we're not always the best at recording every detail, be it out of intentional omission or just a failure to read someone's handwriting.