[kube-system]

That works works out for TVs because of a few specific things:

1. The consumer wants to hook their TV up to the internet to get content

2. TVs ads have a well established market of buyers who will pay for that data

3. TVs are expensive and consumers often buy primarily on price

By comparison, there is no market for the data from my coffee pot, and little to no incentive for a consumer to choose a model that collects data over one that doesn't.

[dmm]

> 1. The consumer wants to hook their TV up to the internet to get content

As the cost of 5G iot chips fall it won't matter what the consumer wants.

> 2. TVs ads have a well established market of buyers who will pay for that data

The data collected from Smart TVs is much more than ads. They track and report everything you watch, including dvds and blurays, with media fingerprinting techniques.

> By comparison, there is no market for the data from my coffee pot

Oh but there will be. Soon some data scientist at your health insurer will notice that people who drink more than 3.5 cups of Folgers a day are 5% more likely to suffer heatstroke or whatever. Then they can adjust premiums and deny claims more effectively. Not allowed to use this information due to regulations? Apply parallel construction and use it to focus limited investigation resources.

I really hope you're right and this doesn't happen.

[kube-system]

> As the cost of 5G iot chips fall it won't matter what the consumer wants.

If the consumer doesn't want that connectivity as a feature, you'll need to find a buyer for the data who is willing to subsidize >= 100% of the price of the additional hardware and service to have a viable product. Alternative data from other sources is probably still going to be a lot cheaper for most things... because a lot of competing data sources have way less overhead, approaching zero in some cases.

> The data collected from Smart TVs is much more than ads. They track and report everything you watch, including dvds and blurays, with media fingerprinting techniques.

I was referring to the purchasers of the data, not the subject of the data.

> Soon some data scientist at your health insurer will notice that people who drink more than 3.5 cups of Folgers a day are 5% more likely to suffer heatstroke or whatever. Then they can adjust premiums and deny claims more effectively.

If they wanted to do this, they could have done it any time over the past 20+ years ago by purchasing transaction data. But insurers have already found better ways to collect even better quality data: Just ask for it directly and offer a discount.