August 10, 2018
There are three kinds of lies: lies, damned lies, and statistics.
— Benjamin Disraeli
There is an unprecedented crisis sweeping the world. It is a ballot stuffing at the widest scale, a sybil attack against the fabric of the modern world itself.
The modern web, the corporate-controlled cyberspace, is essentially composed of infinitely scalable and increasingly complex popularity contests. They collect millions of records in order to simulate millions of people. There is a critical weakness: they confuse the wisdom of the crowd with the tyranny of the majority.
The cybernetic simulacra of the post-truth era: publishers, state actors, and manufacturers, optimizing for views, votes, and profit, are all eschewing the truth.
Facebook is the poster child of fake news — with disastrous geopolitical ramifications — but it's not the only culprit. Google continually grapples with attacks on its ranking algorithm. Amazon is suffering from a systemic crisis of paid reviewers. These are just the most prominent examples. The alarming thing here is not only the deceptions themselves, or the advanced means by which they're carried out, but the scale at which they can perpetrated, permeating all of the new medium.
In the post-truth era of big data, crowdsourcing becomes botsourcing. Petabyte-scale ranking algorithms are continually gamed by adversarial actors, continually caught by software engineering teams, and then re-gamed by the adversarial actors — a perpetual game of cat and mouse clicks.
It's not a distributed denial of service, but a distributed deception.
This process hasn't run its course yet; in many ways it applies unevenly and hesitantly. But I would argue that it's inexorable, something that will soon apply to everything. The point of this essay, then, is not only to codify this hyperreality, but also an attempt to think about how we might move past this, both as the consumers and producers of these data representations.
The objects are no longer commodities: they are no longer even signs whose meaning and message one could decipher and appropriate for oneself, they are tests, they are the ones that interrogate us, and we are summoned to answer them, and the answer is included in the question.
— Jean Baudrillard, Simulacra and Simulation
Once its fundamental integrity is compromised, what more value does a ranking algorithm have?
As Baudrillard might say, this data becomes hyperreal.
In any structure resembling a marketplace (a market of products, or commodities, or ideas), once the ranking ceases to be a predictor of true quality, it only serves to build confidence within its audience. It's a market for lemons that the consumers still trust because they have no other choice (or no other choice that's as easy). In this sense, the value accorded to the data is almost purely affective.
A sufficient mass, once convinced of the merit of the thing, keep the simulacrum alive. A positive feedback loop of halo effects, a memetic delusion. The likes and stars and upvotes continually ensure that the thing stays at the top of the homepage.
Thus data as the ouroboros, the snake that eats its own tail. **In this capacity, data no longer represents or even distorts the underlying reality — there is no underlying reality; the data represents and reinforces itself.