[shadowgovt]

The penalties will also accrue even to people who aren't customers of the company. It may be harder to see when that occurs.

Google's speech model, for example, was bootstrapped on the audio collected from GOOG-411. Is the GDPR intended to prevent things like that? Then it's "intended" to hinder development of practical speech models the likes of which hadn't been seen from decades of research that lacked access to hundreds of hours of input from users asking real questions in real environments.

[jariel]

"Then it's "intended" to hinder development of practical speech models the likes of which hadn't been seen from decades of research that lacked access to hundreds of hours of input from users asking real questions in real environments."

This is simply not true.

Google can absolutely obtain data from users in a manner that's transparent and clear for specific purposes.

They have 40K highly paid Engineers and $100B sitting in the bank. They have ample resources with which they can draw meaningful data necessary - but they'd rather not, if they can just use your data for whatever they want.

[shadowgovt]

Not when they don't even know what future purposes look like.

[robin_reala]

Your comment implies the cost was worth it, but there are definitely people that would disagree.

[shadowgovt]

And they are welcome to not use GOOG-411, Google's voice recognition, or any of the applications derived thereof.

Me personally, I'd been waiting for the breakthrough that Google hit upon to get reliable voice recognition for a couple of decades.

[robin_reala]

Uh, if they don’t use it then the cost/benefit ratio is even more balanced over to the cost. The point is that the cost was never consulted in the first place when the benefit was sought.

[joshuamorton]

What is the cost to someone who never used GOOG-411?

[wizzwizz4]

But whose data was gathered and mined for the purposes of its development.

[joshuamorton]

I'm responding to the statement that GP made which was that to someone who never used GOOG-411, the cost-benefit ratio is further into the cost category [compared to someone who did use GOOG-411]. This would imply that there is some cost to someone who never used the service, hence my question.

[shadowgovt]

Basically zero.

[JadeNB]

To the extent that users benefit from desired violations of their own privacy, they can consent to it. To the extent that other users benefit from violation of my privacy, they have no right to it.

[shadowgovt]

How do you apply such reasoning to, say, traffic counters on highways?

Nobody has a right to know whether the highway is congested because your car may have been one of the counted datapoints?

[JadeNB]

The government has a right to that information. For-profit companies do not. (Also, traffic counters can easily be anonymised, if they're not already, whereas Google's data is essentially worthless to them if anonymised.)

[shadowgovt]

> The government has a right to that information. For-profit companies do not.

That seems like a very arbitrary distinction, and I'm not sure how one arrives there. There are a lot of for-profit companies that build business around data analysis of things that most people could count (given time and inclination to do so) but do not, be it cars on the road or the average price of gasoline state-by-state. Is that a moral red-flag?

[JadeNB]

> That seems like a very arbitrary distinction, and I'm not sure how one arrives there.

I think a blurring of the difference between "things that the government can do" and "things that a private company can do" is a very unfortunate (and also very deliberate) development in modern society.

To me, the distinction is far from arbitrary; it is fundamental. Democratic governments are elected, and at least nominally accountable to the people over whom they exercise their reserved rights; for-profit companies are generally accountable only to their shareholders, not to those over whom they exercise these privileges—and technically to governments, although governments seem unwilling to exercise much of that right of restraint.

(This is a response to your point about the distinction being arbitrary, not to your question about private companies analyzing data. If a dataset of which I am a part is used without identifying me as part of it, or after obtaining my consent to be included in it, then my qualms about it go away. If, however, the company gathering this information, while counting the traffic flow of which I am a part, decides to sell information about where they've noticed I like to shop to advertisers, then, yes, that's a moral red flag for me.)

[shadowgovt]

For-profit companies are also accountable to the market; a company lying about the traffic numbers, for whatever reason, loses when a competitor offers more accurate numbers.

What happens when the government chooses to lie about the numbers? What if it's doing so because the public has signaled it wants to be lied to (which happens from time to time)?