Digital-trends-innovation Cover | Pulse Moments

The Algorithm Is Not Deciding What You See — You Are, Repeatedly

Recommendation systems are usually described as making choices on our behalf. That description obscures the mechanism. These systems are largely reflective — they optimise against measured behaviour, and measured behaviour frequently differs from what people would choose deliberately.

The system measures what you did, not what you value

Watch time, hover duration, scroll speed and re-visits are the signals available. Intention is not among them. If you stop on something you find irritating, that is indistinguishable from interest. The system is not misreading you; it is reading a version of you that is defined entirely by involuntary responses.

Engagement optimisation has predictable side effects

Content that provokes strong reactions produces more measurable engagement than content that is merely good. A system optimising for engagement will therefore surface the provocative over the substantial — not through any editorial intent, but as an arithmetic consequence of the objective function.

The loop tightens over time

Each interaction narrows subsequent recommendations, which narrows subsequent interactions. The result is a feed that becomes progressively more specific and progressively less surprising. This is frequently described as a bubble, though it is more accurately a groove worn by repetition.

Interrupting it requires deliberate input

The practical levers are limited but real: explicit signals like “not interested”, following sources directly rather than accepting suggestions, using chronological feeds where available, and periodically clearing watch history. Each of these substitutes stated preference for measured behaviour, which is the entire fix.

The genuine concern is not personalisation

Personalisation is often useful. The problem is that optimisation targets are set by the platform’s commercial interest and are neither disclosed nor adjustable. A system optimising for your satisfaction and one optimising for time spent produce very different feeds, and users cannot tell which they are receiving.

The system is a mirror pointed at behaviour you did not consciously choose. Changing the reflection requires changing the behaviour, or overriding it explicitly.

Similar Posts