[i_am_jl]

I've seen these advertisements too, also only when my phone had been playing unattended for some time.

I have a (unsupported, unsubstantiated) theory that YT detects phones of "sleepers" and pushes more profitable content with the understanding it won't be skipped.

I've got a few spare phones, maybe I'll run an experiment.

[ksaj]

With YT, it might be an account-specific metric. Ie: flagged as a frequent sleeper. This would not surprise me, since they track just about every other metric possible against your account.

You can have multiple YT accounts on a single gmail acct, but I don't think that'll fool them. They know where you initially logged in from. So you will likely need multiple gmail accounts to do this kind of experiment.

[i_am_jl]

Good shout.

They don't have SIMs, they'll be connected to a VPN router, and I'll create new Gmail accounts for each device, from each device.

[kube-system]

I'm not sure why it would specifically be targeting "sleepers"... there are a lot of reasons why someone might not skip ads... people who are sleeping are probably the least valuable of them.

It could just as well be something super valuable -- like an unattended kiosk device playing youtube to a crowd of people.

[i_am_jl]

Regarding the kiosk, I wholly expect that an unattended device with YT on auto play will ratchet up the length/frequency of ads as long as they're never skipped.

Someone who falls asleep watching YouTube will skip ads, unless they're asleep.

The idea is that if YT can infer that someone is asleep (location, no movement, no sound, low light, night) that they can show the longest, most skip-inducing ads that they've got since they know they won't be skipped.

The difference between the kiosk and the sleeper is that if the sleeper gets a 20 minute ad at 2pm while they're eating lunch, they'll skip it. YT is incentivized to show the most profitable ad that someone won't skip.

The value in identifying sleepers isnt showing a long ad, it's showing a long ad with the certainty that it won't be skipped.

[Gabrys1]

I don't think they specifically target people who tend to go to sleep. But, having worked in the ad engineering, I can imagine they do know how often specific users skip ads and target ads based on that property.