How Not to Know Ourselves « Machine Learning Times

Originally revealed in Medium, July 29, 2020.

Platform information don’t present a direct window into human habits. Rather, they’re direct data of how we behave below platforms’ affect.

Surfing a wave of societal awe and pleasure about “Big Data,” platforms shaped a behavior of releasing “data science” insights on what we search, like, categorical, buy, obsess over, try to cover, and like to neglect. These colourful graphics and juicy taglines — most notably from OKCupid and PornHub, whose information lay claims to the quirks and needs of our intimate lives — are at all times common novelties to behold, ponder, and reference. If realizing ourselves by way of platform information is a observe of our age, it’s definitely not confined to platforms themselves. Aspiring information scientists, curious programmers, vigilant information journalists, analysts of civic organizations and political campaigns, and (final however not the least) tutorial social scientists equivalent to myself make up the rising subject that is determining who we’re, what we do, and the way we sway within the swathes of platform information.

Such information could be spectacular resulting from their unprecedented granularity and quantity, in addition to the truth that they’re seemingly “unobtrusive” recordings of our actions when nobody is watching. These obvious strengths of information for social analysis are outweighed by an issue in what we name the “measurement conditions”: platform information are platforms’ data of their very own behavioral experimentation. Trying to know ourselves by way of platform information tends to yield partial and contorted accounts of human habits that conceal platform interventions. Moreover, although more and more produced by non-corporate actors, such information accounts and narratives are usually amenable to platform money-making and image-building.

Trying to know ourselves by way of platform information tends to yield partial and contorted accounts of human habits that conceal platform interventions.

To be clear, for years many have contested the ascendance of platform information as a staple in quantitative social sciences alongside typical information assortment strategies, equivalent to surveys and experiments. These contestations concentrate on points concerning the information’s representativeness, privateness issues, and precarious entry on the mercy of platform corporations. The “measurement conditions” downside, nonetheless, is completely totally different. In our newly published paper, Harsh Taneja and I name for consideration to the circumstances below which these information come about: what function does the measurement initially serve? As historians have advised us, measurement — or changing elements of the social world into portions in accordance with some enduring instrument — isn’t an finish in itself, however a way for managing occasions and coordinating actions.  Measurement is thus a product of the social and institutional context (i.e., “measurement conditions”) wherein it’s referred to as upon and carried out. 

To proceed studying this text, click here.


Please enter your comment!
Please enter your name here