31 comments

[ 2.7 ms ] story [ 68.3 ms ] thread
I think a more fair assessment would be: "Google is tracking people's movements"
Yeah, but it's not clickbait-worthy unless it has "corona virus" in the title these days.
Would have been an even better title if they somehow worked Zoom into the title.

“Google Is Tracking People's Movements in Their Communities During Coronavirus although they are mostly working from home using Zoom.”

Why is Google not warning people that they have been in contact with potentially infected individuals? Implementing this would probably be trivial for them.

It almost feels like breaking some higher contract with society by reserving that power for advertising and not using it to tackle the pandemic.

How would Google reliably know which people had it, so it could warn people they'd been near?

(Disclosure: I work for Google)

They don't go outside anymore and search for "when do I have to go to the hospital if i have coronavirus". Or you can just report it.

(Disclaimer: I work on an app that tries to do that. Ideally privacy preserving. And we implemented it using Google Nearby API to detect whether people are close. But that sends all messages over Google servers and that would be a deal breaker for us, but the tech itself would be really cool to have. Let me know if you can help.)

Also Related Work:

Singapore doing this: https://www.tracetogether.gov.sg/

This as hardware: https://estimote.com/wearable/

(Disclaimer: I work on an app that tries to do that. Ideally privacy preserving. And we implemented it using Google Nearby API to detect whether people are close. But that sends all messages over Google servers and that would be a deal breaker for us, but the tech itself would be really cool to have. Let me know if you can help.)
(comment deleted)
You can just edit your first post. It's inside the two hour window. All these addendums-as-posts clutter things up.
A lot of people don't go outside anymore and search for symptoms. Even people who don't normally worry are going to search since it's allergy season, and some symptoms overlap.
1. Doesn’t work when there’s a general lockdown

2. People could just be searching out of curiosity and heightened awareness

3. Can’t self-report if you can’t get tested

How would you verify that people who search for these things actually have the virus? They could be doing preemptive research.

And how would you find a list of such questions without just broadly matching on terms like "coronavirus"? The task seems very complex to me.

You're right. Both availability of testing and how much the lock down is relaxed increase the need of this drastically. But those are both things that are likely to increase during the course of the year.
That has way too many false positives to be worth notifying potential contacts over.

You could have something where people who have it can say "I have it, please let others I was near know" but unless we get testing up dramatically that won't work either.

Exactly, these are the things to get right. Hard problems but for sure not impossible.

Also, the alternative of total lock down is hard to beat when it comes to false positives. ;)

(comment deleted)
Hey, maybe all the billions of fines that Google paid to governments can be used to tackle the coronavirus crisis now, right?

... right?

(comment deleted)
That would still be a legal minefield, and rightfully so.
This post is describing https://www.blog.google/technology/health/covid-19-community...

About privacy they say: "For these reports, we use differential privacy, which adds artificial noise to our datasets enabling high quality results without identifying any individual person. The insights are created with aggregated, anonymized sets of data from users who have turned on the Location History setting, which is off by default."

This seems pretty reasonable to me?

(Disclosure: I work for Google)

Given a pattern of movement it's trivially easy to identify an individual by cross referencing with other cheaply/freely available data sets.
Fortunately this is highly aggregated data without individual movements.
This is built on the normal Google location tracking systems. We both know that while they may be providing aggregate stats now they have all the data they need to provide individual location stats. Because the capability exists there is the temptation to use it, especially if the situation worsens more than expected.
In the same way 911 was used to normalize 4th amendment violations, which have now lasted for 20 years...Covid19 will ratchet this up a level to include checkpoints, electronic tracking, social credit scores, etc.

Every year they will trot out their next virus scare. We will look back on the Covid19 scare in awe of how innocent we were in light of what’s coming.

You know what, this time I’m OK with that.

Obviously the preference would be this powers be rolled back, and realistically I say you’re overselling an otherwise valid point...

But other countries are showing it works to slow the spread in a meaningful way.

Here in NYC people are dying left and right. Hospitals are overwhelmed, people who code are being left for dead because they don’t have the resources to save them.

My mother who’s a nurse in CT had to be snuck proper PPE by a concerned coworker because their hospital is rationing it to unsafe levels, allowing healthcare workers to be exposed to confirmed COVID patients with insufficient protection.

I don’t care if they give us all ankle bracelets at this point, if we don’t act decisively now, millions will die.

This isn’t some shadow war with DOD telling us the terrorists will get us if we don’t break encryption.

The bodies are piling up in plain sight, we need to stop this by all means necessary.

I'm glad Google has time for this yet can't build a simple website to help us.
This is simply an aggregate (on country level) on how many people (as change to baseline) have been at what sort of location at what time (with a granularity of days):

* Retail & recreation

* Grocery & pharmacy

* Parks

* Transit stations

* Workplaces

* Residential

It's quite clear that this can be easily derived from the information they have in maps.google.com plus phone movements. It's unlikely that this data presents any privacy concern.

On the other hand it's probably mainly useless with regards to fighting Covid19. As the data is not detailed enough for such a purpose.

Also I think we should point out that releasing this data in PDF form seems to be deliberate to prevent anyone else from providing meaningful insights by adding infographics of their own.

[Edit: Clarification that does not contain number of people but simply difference to baseline]

Isn't it what they already do anyway?
I trust google over facebook. (Which isnt really saying much)