The you on camera is not quite the you in the room

Short answer: on video you can see your own face, which no ordinary conversation has ever asked of you. Managing that picture takes attention away from the person you are actually talking to, and the split is what makes people feel slightly performed on camera. It is fixable, and mostly by covering up your own thumbnail.

Why do I act differently the moment the camera is on?

Because you can see yourself, which is a condition humans almost never operate under. In a normal conversation you have no idea what your face is doing; on video you have a live feed of it in the corner, and part of your attention goes to managing the picture instead of the person. That split is the whole effect. It is why people sound slightly performed on camera and slightly relieved the moment it goes off. Nothing has gone wrong — you are simply doing two jobs at once, and one of them is being your own audience. It fades as the conversation takes over, which is why the first thirty seconds of any video call feel stiffer than the ten minutes that follow.

Am I being fake if I act different on camera?

Not usually. It is closer to the version of you that shows up at a party where you know two people: a bit more animated, a bit more edited, still recognisably you. Most of us run several of these and call them all ourselves. The one to watch is different — if you only say things you think will not get you skipped, the conversation empties out fast and you will not remember any of it. So the test is not whether you are performing, because you are, and so is the person opposite. It is whether the camera version says things the room version would not. Louder is fine. Different in substance is worth noticing.

Why do I keep watching my own thumbnail?

Nearly everyone does, and it is one of the easiest things to fix. Self-monitoring pulls your eyes off the other person, which reads on their end as distraction or disinterest, and it feeds a loop where you are grading your own face in real time instead of listening. The practical move is to shrink or cover the self-view once you have checked your framing — a sticky note works. Look at the other person, not at yourself looking at them. Conversations get noticeably better within about a minute, and the effect is entirely out of proportion to how small the change is. It also stops the slow drift where you start adjusting your face for the camera instead of for the person, which is the thing that reads as distant on the other end.

If the camera is the part you dread rather than the part you manage, start here instead.

Does the camera make me more honest or less?

Both at once, in different registers. The visual channel is heavily managed — the angle, the light, the tidied corner of the room, the shirt you changed into. The verbal channel often goes the other way, because a stranger with no connection to your life is unusually safe to say true things to. So you get people carefully staged and unusually candid in the same breath. Worth knowing about yourself: whichever channel you are managing hardest is where you are most anxious. It is also why a chat can feel more honest than a conversation with someone who has known you a decade.

Why is it easier to say real things to a stranger?

Because there is no future to protect. Anyone who knows you holds a running record, so a confession has consequences that outlast the conversation. A stranger on a random chat has none of that — the room closes, there is no friend list, no history, and you will almost certainly never be paired again. That combination makes people unusually direct, and it is one of the genuinely good things about the format. The same property is why it should not become your only outlet: consequence-free honesty is useful, but it does not accumulate into anyone actually knowing you. The people who hold your history are the ones who can do anything with it, and a stranger, by definition, cannot.

Which version should I be aiming for?

The one you can hold for an hour without effort. That is the only real test, and it rules out both extremes — the fully unedited version you would not be comfortable broadcasting, and the polished character that collapses the moment someone asks a follow-up. In practice it means fixing the things that are just logistics, like light and framing and background, so that none of your attention is going there, and then spending the rest on listening. People tend to be far better at spotting effort than at spotting flaws, and effort is the thing that reads as fake. An ordinary face has never been the problem; the sense of a bit being performed at someone is.