Can Algorithms Discern My Meanings?
I suppose that, theoretically, they could, but only theoretically, and only given a lifetime’s worth of customized definitions.
In current reality? …not even close.
Let’s say that I spend hours exploring a site that sells sex toys. How can it be known that I’m doing that because a friend told me about how easy the site is to use and I’m interested in examples of user-friendly design? Or perhaps that I stumble across that site on a home computer that my son last used and wander around in a stunned state before learning that it was the cute girl that my son’s friend brought over who had been hanging out there. Separately, my husband repeatedly snipes auctions of decidedly femme collectibles for me, but does not shop for himself online. Take a number of such examples without context, and you start constructing the image of a decidedly eccentric (to put it mildly) family, when the reality could be the opposite.
Most of the gatherers of information on what we do online aren’t specifically interested in me and my family or in what might appear to be aberrant behavior for people in our demographic, although they would be if they could determine context. Tracking the habits and behavior of millions results in such anomalies falling by the wayside. If the systems are sophisticated, and aided by additional market information tools, the resulting information can then be developed in more detail, revealing individual wavelets within the waves across the tides of traffic movement. Little of it, though, is presently about individuals in particular.
Its primary usefulness is to marketers. For example, putting the wavelets together, they might discover that people who like to look at or buy red cars are twice as likely to buy a Camaro, and this information will lead them to put a red Camaro in their online ad. Even this group behavior information could be an anomaly, though, and needs to be consistently repeated over time to become reliable data. By also plugging this sort of information into a recommendation engine, they might increase online sales (or at least clicks) substantially. This is generally an inoffensive process to us, and many online shoppers actually like the results, finding books, music, and other merchandise to their taste that they might never otherwise have found.
Marketers also include those packaging and presenting a political candidate to potential voters, and this segment’s agendas can be tangentially tied to the secondary group to whom this information is useful, namely researchers (including academics). A tie can occur, for example, where the researchers’ data aids political agendas and results in elected representatives supporting funding for more data collection.
Regardless of the goal and end uses, none of this data gathering and weighing involves relevance and meaning to the individual except when it is 1. part of an accurate enough wavelet to filter back product recommendations or 2. the individual seeks widely popular or repeatedly accessed/recommended data, thereby improving shopping and researching experience and results.
Developing better search engines and relevancy structures matters a lot to anyone with something to sell online. They know that they can barely see the tip of the iceberg of things that I, or anyone else, might be interested in buying, joining, etc. Hundreds of millions of dollars are poured into tracking our paths and destinations and online behavior towards hopefully uncovering this ‘iceberg’. Such building could also matter a lot to societies as a whole and organizations and communities in particular. It might, in an integrated view, matter more to the second constituency if it can be customized. Recognizing that could also turn out to be a boon to the first group.
Such discovery could happen. It won’t happen, however, by following us all around with ever more complex and sensitive algorithms, and not just because the richer a user’s activities, the more likely that user is to be aware of being tracked and of learning how to evade it. It won’t happen until I, the user, have a stake in revealing my context and meanings, incentives to do so that benefit me and my fellowes on levels far beyond those deemed sufficient so far, and the structural tools that empower me to do it.
Our personal meanings are important to us. They have value. They represent enrichment far beyond the scope of today’s online transactions, social or commercial. I believe that sharing and connecting them responsibly and respectfully would result in societal changes that we haven’t fully conceived yet.
