Are We Still Autonomous in a Fully Connected World?

When did you last make a decision that didn't touch a server somewhere? Not a big decision — a small one. Which route to take. What to watch tonight. Whether to buy the thing in the cart. Try to find one from today, and notice how long the search takes.

This isn't a trick question, and the point isn't that connectivity is bad. It's that the word "autonomy" was built for a world where most decisions were genuinely self-contained, and it's not obvious the word still means what it used to mean now that almost none are.

What autonomy used to test for

The classical philosophical test for autonomy is roughly: does this decision originate from you, reflect your own values on reflection, and remain free from manipulation or coercion by others? It's a good test. It's also a test built for a world with far fewer intermediaries between a person and their choices than the one that exists now.

Consider what "deciding what to watch tonight" actually involves in 2026. A recommendation system has already filtered several thousand possible options down to twenty, using a model of your past behaviour that you've never seen and couldn't audit if you tried. The twenty you're shown aren't a neutral sample — they're the twenty most likely to keep you watching, which is a different optimisation target than the twenty most likely to be good, or true to your stated preferences, or genuinely what you'd choose given the full set.

You still pick one. That's real — nobody is forcing a specific choice on you. But is the origin of that choice still fully you, when the option set itself was authored by a system with its own objective function, one that was never disclosed and that you never agreed to optimise for?

Is a shaped choice still a free one?

This is genuinely contested territory, and worth sitting in the discomfort of rather than resolving too quickly. One position holds that as long as you retain the formal ability to choose otherwise — to scroll past all twenty recommendations and search for something else entirely — autonomy remains intact. The architecture influences without controlling; you're still the one who says yes.

The counter-position holds that this formal freedom does a lot of quiet work to obscure something real: influence that operates below the threshold of conscious noticing is a different category of thing from influence you can see and evaluate. Advertising, at least, announces itself as advertising. A recommendation system doesn't announce itself as an argument for a particular outcome — it presents itself as simply "what's available," which is a much harder thing to argue with, because there's nothing visible to argue against.

Perhaps the honest answer is that autonomy was never binary, even before any of this existed. Every choice a person makes is shaped by upbringing, culture, mood, the specific options that happened to be visible at the time. What's changed isn't the existence of shaping forces — it's that one particular shaping force is now optimised, continuously, by a system built specifically to be effective at shaping, at a scale and precision no advertiser, parent, or culture ever achieved.

The question that's easy to skip past

Here's a question worth sitting with rather than rushing to answer: if a system could predict, with high confidence, what you would choose before you chose it — not by controlling you, just by knowing you extremely well — would that prediction itself change anything about your autonomy? Nothing about the mechanism of your choosing would be different. You'd still deliberate, still feel like you were deciding. But something about the idea of a fully-connected world already knowing the answer sits uneasily, and it's worth asking why, rather than dismissing the unease as paranoia.

One honest answer: prediction that stays passive — a system that simply knows — is different in kind from prediction that gets used to shape the very options you're shown, in order to make the predicted outcome more likely. The first is observation. The second is a feedback loop, and feedback loops, once running long enough, blur the line between "this is what you wanted" and "this is what you were shown until it became what you wanted."

Somewhere between "the system knows me well" and "the system decides for me" is a line most people cross without ever being told they'd crossed it.

What staying autonomous might actually require

If none of this admits an easy resolution — and it doesn't — there's still a practical takeaway that doesn't require solving the philosophy first: autonomy in a connected world seems to require more active maintenance than it used to, precisely because the default architecture is optimised for something other than your considered preferences. Not paranoid vigilance over every choice, which isn't sustainable and mistakes the scale of the problem. But an occasional, deliberate check: for the handful of decisions that actually matter — where to live, who to trust, what to believe about something contested — is the option set in front of you the real one, or the curated one? For [everything explored around control and permission](/blog/why-most-people-have-lost-digital-control), that's usually the more useful question than "am I still free," because it's actually answerable.

An open question, left open

Autonomy in a fully connected world may not be something you either have or don't. It may be something closer to a skill — the capacity to notice, occasionally, when the twenty options in front of you were chosen for a reason other than your own good, and to go looking for the twenty-first anyway. Whether that's still autonomy in the old philosophical sense, or something new that deserves its own name, is a question this piece doesn't answer. It's one worth returning to, though, the next time a decision feels unusually easy.

Frequently asked questions

Does using a recommendation system automatically mean I've lost autonomy?
Not automatically — the formal ability to choose otherwise still exists. The more useful question is whether you'd still choose the same thing if you could see the full, uncurated option set, and how often you check.

Is this really different from how advertising or culture has always shaped choices?
The mechanism of influence isn't new, but the scale, precision, and feedback speed are. A recommendation system updates its model of you continuously and personally, which no advertisement in any previous era could do.

Can you opt out of shaped choices entirely?
Not fully, and trying to do so for every decision isn't practical. What's more realistic is applying scrutiny selectively — to the decisions that actually matter — rather than to every recommendation you encounter.

Is there a name for this specific philosophical problem?
It overlaps with long-standing debates about manipulation and "nudging" in behavioural economics, though the always-on, individually-personalised version of it is new enough that the philosophical vocabulary is still catching up.

What's a practical first step toward reclaiming some autonomy?
For any decision you'd consider genuinely important, deliberately search outside whatever was recommended first, at least once. It's a small habit, but it keeps the muscle of independent choosing from going entirely quiet.