FaceFlow – Random Video Chat

The first stranger FaceFlow handed me was a man in a grey hoodie, lit from underneath by his own laptop, and for about two seconds he looked like every reason to press Next. Then he asked whether the thing behind me was a real fireplace or one of the electric ones, and we spent eleven minutes on it. I’ve thought about that call more than it probably deserves, mostly because of how close I came to skipping it on nothing at all.

Which is the honest place to start. Plenty of calls here go nowhere — a fair share end inside three seconds, and you’ll be on the giving end of that as often as the receiving end. There are flat half-hours where the same few faces come round again.

What I keep turning over is why those three-second endings happen so easily. Meeting somebody anywhere else hands you a quiet pile of information you never asked for and barely notice using: roughly how tall they are, how they came through the door, what they’ve got on, who they turned up with. A random video chat throws every one of those away. Everyone arrives at the same size, in the same rectangle, lit by whichever lamp they happen to own.

So you get a face and a voice. That’s the lot.

Put like that it sounds thin, and for the first few minutes it is. But the two cues you’re left with were carrying most of the weight anyway, and the ones taken off you are mostly the ones you’d have used to sort people before ever speaking to them. Nobody turns up pre-sorted here, and nothing in the picture marks out which of them is worth staying for.

What the rectangle leaves out

Mechanically there’s almost nothing to it. Open the page, let the browser hand over your camera, and it starts pairing you with whoever else has just done the same. No account, nothing to download — it runs in the tab you already had open. Skip when you want to. The other person can too, and will.

Which means the small amount you can control is worth a minute of thought, because it’s the whole of what a stranger has to go on. A few things that genuinely change how the first ten seconds land:

• Get the light in front of you. A window or a lamp facing your face rather than behind your head — a backlit silhouette reads as somebody hiding, and gets skipped for it.
• Camera roughly at eye level. Propped on a couple of books beats held in your lap.
• Check what’s in shot behind you before you start — post on a table, a delivery label, the view straight out of your own window. None of that needs to be part of the introduction.
• Take the skips lightly. Being dropped in two seconds usually means their evening was elsewhere, not that yours is going badly.

You’re deciding about a person on a face and a voice, and they’re doing the same to you at the same moment. It’s less to work with than you’ve ever had, and it works far more often than it has any business working.

Who ends up in it, and what that’s worth

Because there’s no sorting, who you get comes down to overlap — the people also awake, also with an hour going spare, at that exact moment. An hour or two either side of midnight that means plenty of people whose afternoon is only just getting going. That’s the part of FaceFlow I’d have talked myself out of if the site had let me filter first. The same flattening that costs you every useful clue also means a stranger four time zones away turns up the same size as one three streets over.

Fair warning: two people with nothing arranged between them sometimes have nothing to say either, and no amount of good lighting fixes that. Evenings run busier than afternoons, weekends busier than midweek — but nobody can promise you a good one.

What I’d say, if you’re weighing it up: the mistake is judging FaceFlow on a single call, the way you’d judge a video or a song. A single call is a sample of one, drawn blind. A run of them, on an ordinary evening with the camera lit and your expectations near the floor, is a different measurement altogether. Try an evening of it rather than a call or two, and judge the place on the ones you nearly skipped.