Reading Floridi, Fifteen Years Later

When The Ethics of Information appeared in 2013 it was reviewed politely, cited frequently in its own subdiscipline, and largely overlooked outside it.[^1] The political moment was the wrong one for the book. The questions Luciano Floridi was asking — about the moral status of informational entities, about the ontological commitments implicit in our use of digital systems, about the architecture of an ethics adequate to the infosphere — were widely held to be the kind of abstract philosophical questions that one could safely defer until the practical urgency of the technology made answers necessary.

The urgency has now arrived. It seemed worth returning to the book on something close to a decade-and-a-half anniversary, and asking how well its central arguments have aged. The short answer is: better than the contemporaneous reviews suggested they would. The longer answer requires a few pages and is the rest of this notice.

The Argument, Recalled

Floridi's central claim, briefly put, is that the contemporary ethics of information cannot be adequately handled within a framework that takes the human moral agent as its central object. Information — or, more carefully, what Floridi calls informational entities, ranging from individual data items to the larger semantic structures within which they are embedded — has a moral standing of its own. The destruction of an informational entity is, on Floridi's view, a moral matter, not because the destruction inconveniences some human agent (though it often does) but because the entity has, in its existence, a kind of value that an adequate ethics must recognise.

The position is sometimes shorthanded as information ethics or as ontocentric ethics. The shorthand has not helped Floridi's reception, because both phrases sound at first like an inflation of the moral domain into territory it does not belong in. The strongest interpretive move, I think, is to read Floridi as offering not a rival to traditional human-centred ethics but a supplement to it: a way of acknowledging that the entities we now spend our lives creating, storing, sharing, and destroying have a moral character that anthropocentric ethics is poorly equipped to discuss.

The book devotes its middle chapters to working out the implications of this position in detail — for privacy, for intellectual property, for environmental ethics applied to the digital environment, for the design of artificial agents. The chapter on privacy, in particular, is one of the best things written on the subject in the last two decades, and I have used it in teaching as recently as last year.

Why the Book Was Underestimated

There are at least three reasons why The Ethics of Information did not, on first publication, land with the force its arguments warranted.

The first is generic. The book belongs to a tradition of analytic moral philosophy that, while perfectly readable, is not optimised for popular reception. Floridi writes carefully, defines his terms, and works through objections; the cumulative argument requires sustained attention from the reader. In an intellectual climate that increasingly rewards the punchy thesis statement and the publishable op-ed, books with this rhythm do not get the attention they deserve.

The second reason is that the book's most provocative argument — the moral standing of informational entities — was easy to caricature. Critics took it to imply that one was morally obligated to back up one's hard drive, or that the deletion of a spam email was a small evil. The caricature is unfair. Floridi was careful to distinguish degrees of moral standing, to defend the proportionality of obligation to entity, and to deny the more obvious reductios. But the caricature was sticky in a way the careful response was not.

The third reason — and the one that strikes me as most important on rereading — is that the book's audience had not yet arrived. In 2013, the people who would have most benefited from Floridi's arguments — designers of large data systems, executives at platform companies, regulators in nascent digital-rights agencies — were not yet reading philosophy. They were still operating under the working assumption that the ethical questions raised by their work could be safely handled by adjacent professions: by lawyers, by communications staff, by the occasional academic in a panel discussion. The thought that the ethics of these systems might require a serious theoretical foundation was a minority position even within the firms that most needed it.

That has changed, and the change is recent. Floridi's later book, The Ethics of Artificial Intelligence (2023), is, to my reading, a less satisfying work than The Ethics of Information — it is more compromised by the need to engage with the policy conversation as it now exists — but its commercial success is a sign that the audience has finally caught up to the questions.[^2]

The book was not ahead of its time. The time was behind the book, and has only recently arrived.

What Has Aged Well

Three things, in particular, have aged well.

The first is the privacy chapter, which I have already mentioned. Floridi's account of privacy as a function of informational friction — the difficulty with which information flows from one context to another — is, in my view, the most useful contribution any philosopher has made to the subject in the last twenty years. The contemporaneous literature on privacy treated it largely as a question of control: the right to decide who has access to information about oneself. Floridi's reframing in terms of friction was prescient. It captured the way in which the contemporary erosion of privacy is not primarily a story about explicit consent and its violation but about the engineering of contexts so as to remove the friction that privacy depends on. The platforms have not, by and large, tricked us into revealing things. They have simply removed the friction that, in earlier centuries, made not revealing them the path of least resistance.

The second thing that has aged well is the book's account of the ethics of design. Floridi argues, in several places, that the moral character of an information system is fixed largely at the point of design, not at the point of use, and that the designers therefore carry a kind of moral responsibility that engineering culture has not historically encouraged them to acknowledge. This was a contested claim in 2013 — the prevailing view was that systems were morally neutral and that responsibility lay with the users — and it is now closer to the orthodoxy in serious discussions of platform regulation. The shift owes a lot to Floridi, though much of the credit has gone elsewhere.

The third is the book's general posture, which I would describe as non-alarmist seriousness. Floridi is not one of the philosophers who has built a career on the existential risk of AI; he is uninterested in the more sensational versions of the long-termist argument. But he is also not one of the philosophers who has dismissed the moral questions raised by digital technology as already-answered by existing ethical traditions. The book occupies a difficult middle position that, in my reading, the moment has vindicated. The serious problems of information ethics are not new in kind; they are old problems in new forms, and they admit of careful work rather than of either alarm or dismissal.

What Has Aged Less Well

It would be misleading to leave the impression that the book is entirely above reproach.

Two things, in particular, have aged less well. The first is a stylistic matter: Floridi's commitment to coining technical vocabulary — infosphere, inforg, ontocentric, Level of Abstraction — is a barrier to readers approaching the work from outside academic philosophy. The vocabulary is, in most cases, doing real work, but the cumulative effect is to make the book feel more arcane than it needs to be. Shannon Vallor's Technology and the Virtues, published three years later, covers some of the same ground with much less terminological apparatus, and reaches a wider audience as a result.[^3]

The second criticism is more substantive. Floridi's framework is largely silent on questions of political economy. The platforms whose informational systems are at issue are not abstract designers operating on neutral data; they are firms with particular interests, operating in particular markets, under particular regulatory regimes. The ethics that The Ethics of Information offers is, by design, prior to these questions — it is meant to provide foundations on which more concrete normative work can rest. But a reader who comes to the book hoping for guidance on the regulation of large platforms will be disappointed. The work has had to be done elsewhere, and the references most often cited in that regulatory literature are not Floridi but Mittelstadt and his collaborators, who have explicitly tied the philosophical work to specific algorithmic systems.[^4]

A Closing Note

I have been reading and rereading The Ethics of Information for fifteen years, and the experience has been one of slowly growing respect. The book did not announce itself with the force of, say, Norbert Wiener's The Human Use of Human Beings — another work that took several decades to find the audience it deserved.[^5] But it has, I think, more durable arguments than its initial reception suggested, and it deserves to be read again, slowly, by anyone who proposes to think seriously about the ethics of contemporary information systems.

A reader new to the work should start with the introduction and the chapter on privacy. A reader returning to the work should pay particular attention to the discussion of design ethics, which seems to me to have grown in relevance as the design decisions in question have grown in consequence. The book is not, in either case, a quick read. It rewards the time it asks for, which is more than one can say for most of what is currently in print on these subjects.

[^1]: Floridi (2013). [^2]: Floridi (2023). [^3]: Vallor (2016). [^4]: Mittelstadt et al. (2016). [^5]: Wiener (1954).