Chat for video: the argument for changing the hour, not the site

Every guide in this category ranks services. I've read a lot of them and I can't find a ranking that survives contact with a second evening, because the thing being ranked isn't stable: what you meet depends more on when you opened it than on which address you typed.

So the recommendation is a schedule rather than a site. If the version you remember was 2011 at eleven at night, the nearest thing available now is not a site called something else — it's the same site you already tried, opened at two. That sounds like a dodge. The rest of this is why it isn't.

Not a mood, and not a type of person. It selects a longitude. At two in the morning in Europe almost nobody you meet is in their own night — they're in an evening somewhere else, doing unremarkable evening things, and the quiet on your side is not shared.

That table is arithmetic rather than research, which is why it's the part of the argument I trust most. Whatever a service claims about matching, it cannot pair you with people who are asleep, and the sleeping half of the planet is the strongest filter in the system.

Which also explains the most common bad first experience. Somebody tries a chat without account on a weekday afternoon, gets four poor pairings and closes the tab.

They have sampled the thinnest chat for video hour there is and concluded something about the service instead.

A man on a pier in Helsinki, middle of a weekday, in the quiet stretch when nothing bites. He had the phone propped so that the frame held a sleeping dog and about a metre of water. He talked for a long time and never once moved the camera to himself. Nothing else in a chat with strangers tells you as much before the first word.

• Something static and alive in frame — a sleeping animal, a fire, a view the person is also watching — means they have settled in. Expect a long session.
• A dark room and a face at close range means the phone is in a hand and the hand will get tired. Ninety seconds, usually.
• A frame that changes every few seconds means somebody is walking, and walkers leave when they arrive.
• No object at all, and the preview turned off, usually means a session that is being conducted rather than had.

Three of them, and only two get a proper answer.

They are in the order people raise them, with the weakest answer last so that it isn't buried.

If the hour matters that much, why does no service say so?

Because an hour is not sellable. A filter, a subscription, a queue position — those are things a service controls and can charge for. A no signup video chat has even less to offer in that line, and the hour belongs to the planet either way, so nothing about it reaches the interface.

Doesn't a thin hour just mean the same people repeatedly?

It can, and on smaller services it does. The test takes two minutes: if you meet somebody twice in one sitting, the pool is small enough that the hour argument stops working and you do need a different address.

Isn't this just nostalgia with a timetable attached?

Possibly. I can't rule it out and I've stopped pretending otherwise. The mechanism, though, is checkable and the nostalgia isn't; where the two disagree, believe whichever one you can measure.

Number two worries me most. A schedule is useless advice on a service with a few hundred people on it at once, and from outside there's no reliable way to tell how many that is.

Around one in the morning the pictures get worse, and the usual explanation is congestion. That explanation looks wrong to me: late connections are the quietest of the day on most residential lines. What changes is where people are sitting.

• Ask where the router is. Late sessions are usually held from a bedroom, and a bedroom is normally the furthest room from it.
• Ask whether they moved recently — people relocate to talk without being heard, and the move costs them signal.
• Check your own side the same way before blaming theirs: if your preview is soft too, you are both at the edge of something.

None of this is fixable by a service and all of it is fixable by a person — a useful line to draw before deciding to talk to strangers somewhere else instead.

Open the same address three times: once at ten, once at one, once at four. Ten pairings each, no filters, no attempt to improve your odds. Note the language of each and roughly how long it lasted. That's the experiment, it costs an evening, and it settles the question for your part of the world better than any list of sites.

If the three sittings look the same, the argument is wrong and you should ignore it. If they look as different as they do for me, then the next question isn't which service to use — it's what you want from an evening, and that's a better question to be stuck on.