Digital Enmeshment

Virtually every aspect of our lives is digitalised. Professional and private conversations, photos, schedules, notes, searches, the knowledge we consume: all of this is now encoded, stored, retrieved and mediated by electronic devices. Physical states of matter are translated into digital traces. What is the cyberspace?

Pens, notebooks and keyboards

I used to carry a paper notebook with me at all times. I would write stray ideas, doodle and lay down structured thoughts. But I have used a keyboard for most of my life. Over time, I decided I had to choose one primary cognitive outlet, and I went digital. Sometimes I still crave pen and paper. When that happens, and I cave, I feel the remnant of the world I grew up in, of an archaic education.

I do not believe my thoughts are bound to one information format. Typing is a different interaction with thought. As I type these words, both the feeling and the format differ from writing by hand. I can erase easily. I can emphasise easily. The symbols appear on my screen in this standardised fashion. The emotion, excitement or doubt that would have shaped the letters differently under the movement of my hand is largely melted away. The pause in my speech is, at best, carried through punctuation marks.

What remains is a distilled window on the canvas of my thoughts.

Some information is lost, but not all. It would be wrong to say that these words do not reflect an underlying reality of my mental states. Digital dust in a digital cloud, perhaps; but these enciphered symbols still carry, imperfectly, my intentional meaning.

This is where the question starts. Not: did the digital object preserve everything? Not: is typing the same as handwriting?

Rather: what kind of weight does a digital trace have?

Not fake or simulacral, but a new modality

This matters because digital traces now participate in private and shared systems of reference. A message, picture, calendar entry, post, recommendation profile or archived document can become a shared anchor. It may be compressed, distorted, formatted, platformed or monetised, but it still enters the world of references through which we act and recognise one another.

Before hyperhistory, humans had fewer externalised self-references. We had bodies, speech, writing, images, objects, institutions and memories. Now we also have pictures, videos, emojis, search histories, metadata, shared files, private chats, public traces and invisible patterns imprinted every millisecond as we interact with this digitalised world.

If this technology went dark tomorrow, we would survive. But we would be left with a deep open wound, like a missing limb. This echoes the extended mind and embodied cognition threads: our cognitive life does not stop neatly at the skin. It is enacted through tools, gestures, symbols, environments and other people. The digital is one of those environments.

The personal bits that we imprint onto cyberspace are part of us. We are not reducible to them, just as they are not reducible to us. They are not better than embodied life, and they do not replace it. They are another modality through which we remember, express, coordinate and become legible.

The mesh

Together, we weave cyberspace as an extension of ourselves. We share information through this vast network. We connect through these digital bits. The threads are so tightly knotted that they form a mesh: an intricate, fully interwoven pattern, connecting our deepest thoughts and our most abstract representations almost instantly, globally. We are enmeshed, not only with our digital selves, but also with others’. AI strengthens and extends this fabric. It accelerates digital enmeshment, thickens it and tightens the knots.

With advances in computing technologies, philosophical debates have often split between those who believe we already live in a simulation and those who think digital information is simulacral. Considerable theoretical and empirical progress has been made over the past century, yet the crux of the debate remains open. This is not exactly the same question as either of the two I approached elsewhere in Are You an AI?: whether we live in a simulation, and whether machines could be conscious. But it sits nearby.

Meanwhile, we have built machines that use natural language in semantically relevant ways. Their cognitive abilities remain, in large part, underexplored. Over a lifetime, a human may be exposed to language on the order of hundreds of millions of attended words, plausibly approaching a billion for heavy readers or media-saturated lives.¹ LLMs, by contrast, are trained on hundreds of billions to trillions of tokens.² The comparison is crude, but it points to a real difference in scale.

Here, Yann LeCun’s Advanced Machine Intelligence (AMI) research direction, with its emphasis on world models, hints at something important: much of what we learn comes from multimodality. If we take embodied perception, action, memory and social interaction into account, human learning cannot be reduced to language tokens. Nevertheless, language may still hold a compressed key to decipher natural reality. Not because words float above the world, but because they help weave the complex fabric in which experience becomes shareable.

Science also depends on imprints. It predicts where the overall canvas is going by pulling on threads, often from the past. We have the most detailed record of the past that humans have ever had. It is still partial, fragile and unequal, but it is growing. The digital present is constantly becoming evidence. Perhaps this is one thing we have always tended toward: closer verification of what happened, a more persistent shared memory.

So what is the present?

The present is a knot in the mesh

For anyone born in the eighties and after, the digital environment has rarely been merely secondary: video games provided fictional worlds that could be explored and acted within; social media pushed this further by making fragments of embodied experience shareable, searchable and open to others. It is tempting to dismiss digital knots because they are mediated by platforms whose interests diverge from ours. But platform ownership does not erase our agency; it shapes our vulnerabilities. These imprints form our digital ghost: partial, distorted, sometimes captured, but still potent in the world we inhabit. We are past the point where cyberspace can be treated as fantasy.

It is not a macroscopic physical space organised like ordinary matter. It is not a biological space playing by the rules of DNA. It is a virtual space built over robust encodings of our shared world. At the physical level, it records words and meanings in patterns whose regularities track our own. At the biological or psychological level, the comparisons seem to fade. Nevertheless, we co-evolve in this digital space by giving parts of our existence weight within it. Each trace is like a small pressure point in the fabric: local, partial, but capable of pulling on neighbouring threads.

Floridi’s framing of the infosphere and hyperhistory helps name this dependency. Cyberspace is not simulacral. It is now part of how we remember, act, relate and become publicly legible. Societies depend on it. Our existence cannot be cleanly disentangled from it.

We are digitally enmeshed.

Footnotes

Footnotes

1. I am treating this as an order-of-magnitude estimate, not as a settled statistic. For spoken output, Mehl et al. estimated that participants spoke about 16,000 words per day, with large individual variation; a later replication summarised by the University of Arizona reports an average closer to 12,700 words per day by 2019. For childhood input, Hart and Risley’s much-discussed extrapolation estimated roughly 13 to 45 million words heard by age four across socioeconomic groups. For reading, the U.S. Bureau of Labor Statistics’ 2025 American Time Use Survey reports 0.27 hours per day spent reading for personal interest among people aged 15 and over, and Brysbaert’s meta-analysis estimates adult silent reading at 238 words per minute for non-fiction and 260 for fiction. A rough model—25 to 45 million early words, plus 65 adult years times 365 days times 15,000 to 30,000 heard/media words and 4,000 to 15,000 read words per day—lands around 0.5 to 1.1 billion words. See Mehl et al., “Are Women Really More Talkative Than Men?”, University of Arizona, “Are we talking less?”, Hart & Risley, “The Early Catastrophe”, BLS, American Time Use Survey Table 11A, and Brysbaert, “How many words do we read per minute?”. ↩

2. GPT-3 was trained on 300 billion tokens, while Meta reported LLaMA 65B and 33B were trained on 1.4 trillion tokens, and the Llama 2 model card reports 2 trillion tokens. See Brown et al., “Language Models are Few-Shot Learners”, Meta AI, “Introducing LLaMA”, and Meta’s Llama 2 model card. ↩