Free to chat online: ten mistakes, logged from both sides

What the report button actually does

The most common mistake is also the cheapest, so it goes first. You press report, the screen changes inside a second, and it's very hard not to read that speed as the system acting on what you said. It isn't. Two things happen at once, and your report isn't the one you're watching.

On the line the press does three things in order. Your side tears down the link you had. Then the client goes back to the server for somebody new. And the server, having just been told a slot came free, hands you one quickly — because you're the one who freed it. That's all the speed is. You're not being rewarded; you're being recycled. (Forty-one presses over three weeks: median wait after a report, four seconds; median wait after an ordinary skip, nine.)

Does the other person get told?

Not by your press, and not straight away. What they see is the call ending — the same thing they would see if you had simply moved on, which is the point.

Report first, or leave first?

Report first. Leaving drops the session, and on most rooms the form then loses the one thing it was going to attach: the partner id it held a second ago. The button still submits. It submits about nothing.

Reports do get read. Not at the speed the screen suggests — in batches, hours or days later, by a person or by a queue, and the outcome is never shown to you. Unsatisfying, and also the only arrangement that survives contact with the scale involved. Anything visible would be gamed inside a week.

So the useful habit is a dull one: write down the time. Not a screenshot — a time, to the minute, and which room. Every room that's free to chat online logs by the clock, and if it ever matters, the clock is the only thing you'll have that somebody else can look up.

The night the big one closed

November 2023, and the service everybody means when they say this category shut its doors. What followed wasn't an even spill. Some rooms took a crowd overnight and some took almost nothing, and the difference wasn't the home page.

It was the phone. The rooms that grew were the ones already working one-handed, portrait, with the next button under a thumb — mobile video chat, not a desktop page squeezed smaller. I've no traffic figures for anybody else's site and I'm not inventing any; what I have is thirty evenings of my own, and the rooms where my phone worked without rotating were the ones where somebody answered.

The line side of that is dull and decisive. A phone does not give you a worse connection than a laptop. It gives you a less stable one, because it changes masts, and each change is a renegotiation while you are still talking.

The free to chat online mistake people made — I made it too — was blaming the room. A room is a matching server and a page. It is almost never the thing that broke your call.

Everything this page has not got

Through the negative first, because it is shorter. Nothing on this page sorts people for you. There is no ranking, no recommended partner, no algorithm that has learned what you like, and no list of who is on right now.

There is no account either, so there is no history, no inbox, no folder of people you met on Tuesday. Nothing is saved between visits except which theme you picked.

There is no filter that does what people expect a filter to do. Some rooms sell one by country; what it buys is a narrower crowd, and a narrower crowd is a longer wait, because the room now needs two matching people pressing inside one second, where before any pair at all would serve. Nobody prints that arithmetic on the sales page.

And who you meet is not something a page can promise. This one certainly can't. Anybody who does is describing a room they don't run either.

What there is: a door to a room where you can talk to strangers, running on somebody else's machine, opening in this same tab, free at the point of use. That is the whole inventory.

Go through to the room

An hour straight, and the tiredness after

Somewhere around the fiftieth minute of continuous pressing, I stop being any good at this. The log is fairly clear about it: in the opening twenty minutes my median time to a first sentence runs about six seconds; in the sixth twenty it's eleven, and the number of calls I end before the picture has even resolved roughly triples.

Nothing has changed on the line. Nothing at all — same network, same hour, same room. What has changed is that every connection is a small decision, and a few hundred small decisions in a row flatten anybody. The tell is exactly that last number: when you notice you are pressing next before the picture arrives, you are no longer choosing. Stop there. It comes back after twenty minutes away.

Why there are no adverts between the calls

People assume there must be, and that the absence means something is being taken instead. Worth taking apart, since it is the most common suspicion I get asked about.

An advert between two calls has about two seconds to load, render and be seen. Two seconds is not a slot anybody pays for. Worse, on the line it is actively expensive: the frame would be fetching while the new connection is being negotiated, over the same upstream that is about to carry your face, and the thing it would slow down is the very moment you are being judged on.

What pays instead is the crossing. Somebody goes over from a page such as this, and if they later hand the room money, part of it returns here. Most crossings pay nothing. That is the entire economy, and it is written out properly on the disclosure page.

The mistake is assuming free means your face is the product. In this corner of it, the product is the click — which is a lesser thing to be, and still worth knowing, since it explains why every page in this category wants you to press.

How a person who is about to go stands

Here I have to be careful. Reading a person off two seconds of a moving picture is where most of the bad advice in this category lives, and I do not want to add to it. What follows are tendencies I wrote down, not verdicts about anybody.

The eyes go down and slightly left or right — not to their own preview, which sits in a corner, but below the camera, to where the next button is. It is the same glance you make at a door handle.

The shoulders square up a fraction. People push back from a desk before they press.

A hand leaves the frame and does not come back. That one is close to reliable, and it is also the least useful, because by the time you have seen it the decision is made.

Which is the line side of this, and the part that is not psychology at all. You are watching something between a fifth and half a second old, depending on the network. Their hand left before you saw it leave. You are always reading the past.

The mistake is the natural one: trying to hold somebody who has already decided. Talking faster, asking a second question on top of the first. It makes the remaining seconds worse for both of you, and it does not change the outcome. In video chat with strangers, leaving is cheap by design, and the design is not going to be argued out of it.

Say the sentence anyway. About one time in twelve, by my count, the hand comes back.

Language practice, which is a side effect

Best hour I had in the whole month was a farmer outside Lisbon, before dawn, just finished milking, and he spent it taking a ballpoint pen apart on the table. Spring, tube, cap, the little brass nose. My Portuguese is bad and got slightly less bad. That was not why the call lasted.

It writes worse after you put it back. Every time. I have done this with maybe twenty pens and it is always true, and I still do it.

the farmer outside Lisbon, translated roughly, with his permission

His hands were busy, so the silences stopped being pauses. That is the mechanism, and it carries over: the best person to practise a language with is somebody doing something with their hands, because neither of you has to produce a sentence to keep the thing alive. There is a line cost — a still frame compresses beautifully and two moving hands do not — and it is worth every bit of the bandwidth it eats.

A slow line, and what to do in the middle of a call

The entry that cost me the most calls before I understood it. A phone that starts on Wi-Fi and walks out of range does not drop the call. It renegotiates, quietly, and the outgoing picture steps down — and it does not step back up when the signal recovers. It stays down until the call ends. (Eleven times in the log; every one recovered on the next call and none recovered during the call it started in.)

The reason is that the step down is agreed once, at the moment the network flinches, and there is nothing in the handshake that goes back to ask for more later. Your side sees their picture, which is unaffected, so nothing looks wrong to you. To them you have been slowly turning into a smear for five minutes.

What do I do when I notice?

End the call and start a new one. That is the only thing that resets it. It feels rude and it is the opposite: the next two minutes with them are otherwise spent looking at a version of you that is worse than the real one.

Can I stop it happening?

Pick one network before you start and stay inside it. If you are going to walk around, start on the mobile network and never join the Wi-Fi. A stable poor line beats a good one that changes.

Which is why mobile video chat rewards standing still more than it rewards a fast connection. I have had better calls on a train seat that never moved between masts than in a kitchen with fast broadband and a phone that kept switching to it.

Conclusion, kept separate from the observation, as with the rest of this log: the number to watch is not speed. It is how many times the network changes its mind during a call.

The details that give themselves away

Not the accent. People guard the accent and then type two words into the side field, and the keyboard does the rest — the autocorrect puts its own diacritics in, the quotation marks come out in the shape that country uses, the comma falls in a decimal. I have placed people within a country or two from six typed words, and I was not trying to.

The other one is the window behind you and the clock. Light says the hour, the hour says roughly the longitude, and the two together narrow things more than any answer you carefully withheld. Not a reason to be frightened, though. A reason to settle beforehand what you do not mind giving away, because in the moment you will not be settling anything.

And the shape of the worry is usually backwards. Nobody in video chat with strangers is sitting there assembling a file on you; the guesses land sideways, out of the corner of somebody's attention, and they are gone by the next call. The thing worth watching is the other direction — you assembling the file yourself, one careless detail a session, in front of a few hundred separate people, until one of them turns out to care.

If the person is a child

Last, because it is the costliest, and the only entry in this log where I am not going to be measured about it. End the call. Immediately, not after working out whether you have understood correctly.

Then report, and write down the time and the room, which is the one place in this log where the dull habit from the first entry earns its keep. The report is read later by somebody who was not there, and the clock is what lets them find the session.

Do not try to handle it inside the call — not by asking questions, not by warning them, not by staying to see. You can verify nothing, you can follow up on nothing, and every extra second is a second on record with a child at the far end.

That is the end of the log. Most of what happens in a room that is free to chat online is duller than any of this: waiting, pressing, a few seconds of somebody's kitchen, and now and then a farmer with a pen in pieces.

The dull part is the good part. People who talk to strangers for a month end up with a list much like this one; mine only differs in being written down. Go in knowing which of the ten you are about to make, and you will mostly make a different one.