On ignorance as a product of digital infrastructures
There is a convenient notion regarding information that warrants dismantling: that we live surrounded by it, that the problem of our era is excess, that never before has so much been available to anyone. This is a half-truth, and the missing half is the significant one. For alongside that abundance, there occurs, in a less visible and more decisive manner, its reverse: an unequal distribution of what each individual can actually see, know, or find. We do not live solely in an attention economy. We live, increasingly, in an economy of ignorance: not that which it displays, but that which it renders improbable.
This is the sixth text in the series on infrastructures that I have been publishing in this public notebook, and in a sense, the one that turns all the previous ones on their head. In the five preceding entries, I argued that digital and cultural systems capture, displace, homogenise, and financialise, and that responsibility for their effects is structural, not individual. Here, I address what those systems do by not showing: the ignorance they produce, and why that production matters for considering what art can even emerge.
1. Ignorance is not a void, it is a product
The technical term exists and should be utilised. Robert Proctor termed agnotology the study of ignorance understood not as a simple absence of knowledge, but as something that is produced: culturally and politically, through secrecy, selection, and suppression. His classic case is the tobacco industry, which for decades did not merely conceal the truth: it actively manufactured doubt, funded studies, and sowed confusion. Ignorance, in that instance, was not a void that no one had yet filled. It was a result, fabricated with intent and resources.
What I propose is to shift that perspective to algorithmic infrastructures, where ignorance is produced in a more subtle manner than the deliberate falsehoods of the tobacco industry. No one needs to lie. It suffices to consider how the system is configured. Algorithmic ignorance is not manufactured by concealing: it is manufactured by deciding what appears, in what order, with what probability, and to whom. And that, unlike classic secrecy, is invisible to those who suffer it, because one cannot miss what one never knew existed.
2. Four ways of not seeing
Infrastructural ignorance is not a singular phenomenon. It adopts at least four forms, and it is useful to distinguish them because they operate in distinct ways.
The first is invisible subtraction. Personalisation systems display to each user what aligns with their previous preferences, and in doing so, isolate them from what does not fit. Eli Pariser termed this the filter bubble: a reduction in informational diversity that the user does not perceive, because the system spares them even the experience of absence. You do not see what is missing, nor do you see that it is missing. The bubble is perfect when one does not even notice it is a bubble.
The second is the exploitable void. There are queries for which content barely exists—the data voids studied by Michael Golebiewski and danah boyd—zones with hardly any results where whoever arrives first with dominant material occupies the space without competition. The void is not a system failure: it is an operable territory within it. Invisibility, here, is not passive. It is a structural opportunity for those who know how to exploit it.
The third is the hierarchy of exposure. Recommendation and ranking algorithms do not merely distribute content: they determine its visibility on a continuous gradient. Something may be neither prohibited nor removed and yet remain so low on the scale of exposure that its existence is, for practical purposes, null. Recent research documents reductions in visibility of several orders of magnitude without the user perceiving it. To ignore, here, is no longer not to know: it is to fail to reach the threshold of visibility.
The fourth is the structural absence of origin. What a dataset does not contain is not merely difficult to find: it does not exist for the system trained upon it. Each dataset encodes prior decisions of inclusion and exclusion, and those decisions determine what can be retrieved, recombined, or generated, and what remains outside the field of the possible. In generative systems, the structure of the corpus does not merely affect the results: it constitutes the space of the producible. What the dataset does not collect is not invisible. It is non-existent for the machine.
3. Why this is an issue of art, and not merely of information
Thus far, this might appear to be a matter of disinformation, politics, or internet search. It is, but it does not end there, and the point that interests me is different: these four forms of ignorance condition what art can even emerge.
In my work on structural surplus, I maintain that the artistic event resides neither in the object nor in the author's intention, but in a reorganisation of the field of the possible that a receiving system registers. Furthermore, for this reorganisation to be fully realised as art, it requires a field of reception where it can be perceived, interpreted, and recognised. Today, this field of reception is increasingly an algorithmic infrastructure. What becomes visible, what reaches the threshold of exposure, and what a system can retrieve or generate is mediated by architectures that determine, without explicit declaration, what enters the field of the possible and what does not.
The consequence is significant. A work may produce surplus—effectively reorganising the field of the possible—yet never be realised as recognised art, not due to a lack of merit, but because the infrastructure governing its visibility kept it below the threshold, outside the dataset, and within the incorrect bubble. Infrastructural ignorance does not destroy the surplus. It performs a more insidious action: it renders it improbable to encounter. It leaves it existing in a zone of practical invisibility where, for the purposes of the field of reception, it is as if it did not occur.
This connects to the arguments I presented in the previous texts of this series. If the apparatus of critical validation produces experience rather than merely reflecting it, and if financial logic reorganises which works are seen, one must now add a further layer: algorithmic infrastructure determines the very probability of appearance. It does not censor; it configures the threshold. A surplus that does not reach the threshold of visibility remains, for the system, in the same position as a surplus that does not exist.
4. What this is not
Three interpretations should be deactivated before concluding.
This is not a conspiracy theory. I do not maintain that anyone deliberately decides which art to conceal. Infrastructural ignorance operates for the most part without intention: it is the effect of a design oriented towards other objectives—engagement, prediction, commercial capacity—rather than a plan for suppression. This makes it more difficult to combat, not less: there is no culprit to identify, only an architecture that produces the effect as a byproduct.
This is not nostalgia for a world without mediation. All visibility has always been mediated: by the printing press, by galleries, by critics, and by museums. I do not advocate for pure access to art without filters, which never existed. The question is not how to eliminate mediation, but which type of mediation favours the emergence of the new and which type renders it improbable.
This is not determinism. That the infrastructure configures the threshold does not mean it determines everything. There remains surplus that breaks through, work that appears against the logic of the system, and reorganisations that the algorithm did not foresee. However, one should not confuse the exception with the rule: that something breaks through despite the system does not prove that the system is neutral. It proves the contrary.
5. Where responsibility lies
The conclusion is the same one that runs through the entire series, and for this reason, this text brings it to a close. If ignorance is produced infrastructurally, then ethical responsibility does not lie primarily with the user who fails to search effectively, nor with the artist who fails to promote themselves, nor with the spectator who fails to find. It lies with those who design, finance, and operate the architectures that decide what becomes probable and what becomes improbable to appear.
Asking the individual to escape their bubble, to find what the system buries, and to see what the architecture renders invisible is to shift onto them a responsibility that is beyond their reach, much like demanding individual sustainability in the face of an industrial design problem. The decisive lever lies elsewhere: in the composition of datasets, in ranking criteria, in architectures of visibility, and in the frameworks that regulate all of the above. To intervene there is to intervene in the very conditions under which a surplus can appear or remain buried.
The question that should be maintained in the face of any visibility system is not what it shows me. It is what it renders improbable for me to see, and who has decided that threshold. Because what fails to appear does not cease to exist: it simply occurs elsewhere, outside the field that the infrastructure illuminates, and at times invisible to it.
On open conversation
With this text, I conclude the series on digital and cultural infrastructures that I have been publishing in this public notebook: sustainability and ethical consumption, institutional validation, the displacement of the value of cultural labour, the geographical distribution of cultural power, the financialisation of the market, and now the infrastructural production of ignorance. All six share a fundamental thesis: in contemporary systems, the decisive factor is not individual conduct but infrastructural design, and ethical responsibility is distributed proportionally to the power of those who design and operate these systems.
If anyone wishes to intervene from the perspective of the sociology of algorithms, platform studies, the political economy of information, or professional experience in the sector, this notebook remains open.
Sources
Proctor, Robert N., and Londa Schiebinger (eds.). Agnotology: The Making and Unmaking of Ignorance. Stanford University Press, 2008.
Pariser, Eli. The Filter Bubble: What the Internet Is Hiding from You. Penguin Press, 2011.
Golebiewski, Michael, and danah boyd. Data Voids: Where Missing Data Can Easily Be Exploited. Data & Society, 2019. https://datasociety.net/library/data-voids/
Esteban Ruiz, J. A. Art as Structural Surplus: Toward a Relational Ontology Beyond Human Authorship. PhilArchive and Zenodo, 2026.
Esteban Ruiz, J. A. The Price of the Model and Who Trains the World. Public notebook at juanesteban.art, 2026.
Series · Infrastructures
This piece forms part of a six-part series examining the influence of digital and cultural systems upon art. The overarching thesis posits that within contemporary systems, the decisive factor is not individual conduct but rather infrastructural design. The series commences with the capture mechanisms of these systems and concludes with their exclusionary practices.