You have been comparing yourself to a casting call

I want to make a statistical argument and leave the moralising out of it, since the moralising is what has made this conversation useless for twenty years.

Set aside whether pornography is good or bad, healthy or unhealthy, art or exploitation. None of that is the subject. The subject is a sampling problem, and it happens to be one of the cleanest examples of a sampling problem you'll meet in ordinary life.

It goes like this. An adult performer isn't a randomly selected man. He's a man who was selected.

That's the article. The rest is elaboration.

What selection does to a distribution

Picture an industry that hires professional basketball players and broadcasts them constantly. If that were the only place you ever saw human beings standing next to one another, your sense of average height would end up catastrophically wrong. Not slightly wrong. Wrong by something like a foot.

You wouldn't experience that as distortion, either. You'd experience it as observation. You would feel like a person who had seen a great deal of evidence, and you would be right about that. The evidence would simply all have come through one filter.

Notice the difference between the basketball case and this one, though. Everyone knows professional athletes are selected for extremity. Nobody watches the NBA and concludes they're short. The selection is legible: discussed, joked about, printed on screen beside the player's name.

In the case we're discussing, the selection is invisible and never named. No caption. Nobody announces before the scene that the following individual was chosen from a large applicant pool partly on this specific characteristic. The filter runs silently, and the viewer's brain does what brains do with silent filters, which is treat the sample as the population.

Three more distortions stacked on top of the first

Selection is the big one, but it doesn't work alone. Three technical effects compound it, and all three run the same way.

1. 1.Lens and framing. Focal length alters apparent proportion, which is why portrait photographers favour particular lenses, and why a wide-angle shot taken close up makes whatever sits nearest the camera loom over everything behind it. Commercial production is shot deliberately, by professionals, to flatter. There's no version of this in which the framing is neutral.
2. 2.Comparative context. Whatever shares the frame sets the apparent scale. Oldest trick in visual composition, and knowing about it doesn't stop it working.
3. 3.Your own viewing angle. You have observed your own body, all your life, from directly above: foreshortened, at the precise angle that minimises apparent length. Every other one you have seen side-on. Worth saying loudly, because this isn't a media effect at all. It applies in a changing room, in a photograph, anywhere. You have never once seen your own from the angle you've seen everybody else's.

The arithmetic of exposure

Now put a clock on it, because the sheer scale is what separates this from every previous generation's version of the problem.

A man in 1975 might see a handful of these images in a year, on paper, badly reproduced. A man today can see more in an evening than his grandfather saw in a lifetime: high definition, an effectively unlimited catalogue, starting from roughly the age he first gets an unsupervised device. Which is to say, before he has finished developing.

That isn't the same phenomenon at higher volume. It's a different phenomenon. A reference distribution built from ten thousand selected observations and absorbed during adolescence isn't a bias one article can talk you out of. It's the substrate your intuitions are made from.

Which is why telling a man the average is about 13 cm often does nothing at all. He hears the number, checks it against ten thousand stored observations, and the observations win. They're more vivid, more numerous and more emotionally weighted than a figure in a paragraph. Vividness beats statistics, and always has.

So what actually shifts it

My honest view is more limited than the confident advice you'll find elsewhere.

A corrupted reference distribution doesn't get fixed by a single fact. You fix it the way you fix any miscalibrated instrument, with repeated exposure to correctly labelled measurements, over time, until the new distribution carries enough weight to compete with the old one.

That's why this site works the way it does. Not one number delivered once in a reassuring tone, but a number a day, from a named study, with sample size and method attached, guessed at before it's revealed. A guess you got wrong sticks far better than a fact you were handed. That isn't a design flourish, it's the oldest result in learning science. Predict, then find out, every day.

It won't undo ten thousand observations in a week. In my experience, and going by the messages this site gets, it does something noticeable in about a month. The internal sense of "typical" starts drifting back toward where clinicians have always said it was.

The part I want to say gently

If you're reading this and recognising yourself, I want to tread carefully, because this is the point in an article like this where someone usually starts issuing instructions about your habits and your character.

I won't. You did nothing wrong by having eyes. The filter was invisible by design, nobody labelled it, and you were young when it was installed. Forming an inaccurate estimate from a biased sample isn't a moral failure; it's the correct behaviour of a healthy brain given bad inputs. Any statistician would have predicted it, and any of us would have done the same.

What you can do is stop treating the estimate as evidence. It never was evidence. It was a casting call, well filmed, shown to you ten thousand times.