[prepend]

This is interesting. What’s the data use agreement like on this sale? Does google get to retain all of Zagat’s historical data? Or is it sold to the Infatuation?

They just bought Zagat in 2011, so selling it so quickly seems odd. It’s an undisclosed amount, so I assume it’s at a loss from its $151M purchase. I hope it eventually becomes public knowledge through Google’s SEC filings.

If this is just a way to buy deep data sets, then this makes sense. $150M for likely the best training set for food review in existence makes sense and therefore isn’t Google stupidly investing in review sites. It would also explain the buy and sell technique used on Boston Dynamics, Moto, and others.

[trevyn]

>It would also explain the buy and sell technique used on Boston Dynamics, Moto, and others.

Boston Dynamics was all Andy Rubin, and it didn't really fit when he left and Robotics collapsed as a result, and Moto was for defensive patents for Android (which they kept).

I haven't followed Zagat, but a sibling comment here mentioned it was Marissa Mayer's doing, so that would fit with the Andy / Boston Dynamics pattern.

There is also a PR disadvantage to feeding the idea that "Google owns everything".

[lacker]

They just bought Zagat in 2011, so selling it so quickly seems odd. It’s an undisclosed amount, so I assume it’s at a loss from its $151M purchase.

Buying Zagat seemed like it was Marissa Mayer's project, so when she left it is possible there was no longer a real champion of the Zagat acquisition left at Google.

[chkuendig]

Relevant: http://www.businessinsider.com/google-zagat-story-2013-6?IR=...

[jonknee]

> They just bought Zagat in 2011, so selling it so quickly seems odd.

That was 7 years ago...

[nailer]

Exactly - a lot can change in that time. Years ago Zagat was an unplayable boss, now he's a playable character.

[chatmasta]

If they are using the data internally (e.g. for ML training), they can get away with not buying it. They already scrape every website in existence and feed their data to ML training algorithms, it's not like Google needs to pay for a data set just to use it internally.

[sf_rob]

Publishing something does not make it public domain. I am not a lawyer, but saying they can do anything with what they scrape seems questionable.

[chatmasta]

I mean in practice they can get away with it. As of now, there are no audit trails required for training of machine learning algorithms.

[ucaetano]

I don't think that there's an issue with someone reading 100 reviews of a certain restaurant on the internet, and summarizing them by writing a new review:

"Most people who talked about it mentioned the great salad bar, but said that the fish was sub par. They commented on it being lively and child-friendly"

That's what I'd guess they are doing.

[prepend]

Zagat likely had much more data than what was publicly available. I have no visibility, but can imagine non-public data useful for training like individual reviewer biases, reviews through their edit lifecycle, stuff like that.

[malchow]

Is data stripping the new asset stripping?

[ucaetano]

Not really, data is replicable, assets aren't.

So you can buy a company, strip all of its data, and resell for a similar value than you bought it for. You can't do that with asset stripping.

But, on the other hand, if the data striping happens again and again, the company becomes worthless, even if it still has all of the assets and data.

Ah, the wonders of the information economy...

[malchow]

Data is replicable, but any data set large enough to be globally useful has a time bounding function. A snapshot of the whole world today is worth less than perfect information just about how humans drive over 200 years / 1.752e+6 hours of observation. Don't you think?

[ucaetano]

Sure. But that doesn't have anything to do with my comment.

As information goes, [A, B] is more valuable than [A], and the value of [A] declines with how many people/companies have A.

[prepend]

It would explain a lot.

[koolba]

> If this is just a way to buy deep data sets, then this makes sense. $150M for likely the best training set for food review in existence makes sense and therefore isn’t Google stupidly investing in review sites. It would also explain the buy and sell technique used on Boston Dynamics, Moto, and others.

It'd only make sense if they couldn't acquire the data set directly for less than the price difference and transaction. An M&A deal for Google isn't cheap and comes with a lot of oversight, risk, and scrutiny.

[prepend]

In some situations data can not be sold if the privacy policy only allows internal use. For example, gmail doesn’t allow my mail to be sold or loaned to anyone, but google can access the data to show ads and for quality control. (See https://www.google.com/policies/privacy/#infouse)

Google is unlikely to be purchased. But if the NSA bought google, then it could use my gmail data -per the privacy agreement that I agreed to- “services to provide, maintain, protect and improve them, to develop new ones, and to protect Google and our users. We also use this information to offer you tailored content.”

This is only possible if the organization is purchased.

[ballenf]

Could this be related to antitrust scrutiny in EU?