Google and it's flawed Data experiment to further bolster Global Warming Ideas and non-sense laws!

Google just announced that they will continue their air quality "experiment" in the Bay area, Central Valley and Los Angeles. While I admire their dedication of using statistical data to proof certain situations (instead of just claiming it), I highly doubt that their process will generate anything else than flawed data.

The biggest issue I see is that this data will be consumed by the EPA and other Government agencies and presented as "facts" to the public which lacks the understanding how this data is collected. I already see the headline: Your morning commute is the reason why you are sick! Well traffic is probably more responsible for that than air quality.

The basic idea is that Google is using it's street view car fleet to collect environment data while driving around cities. This data is later processed and scaled to a certain time range in which it was collected. The primary concern here is that the sample rate is just too little and will never reach statistical significancy. Basically worthless data. There is also another issue that the sensor might read proportional higher the street view car instead of the environment (I saw a couple VW's in their fleet, too). Also there is no mentioning of any independent - unbiased organization that actually verifies the sensors in contrast to the data that needs to be collected. You have to "trust" that Google and Aclima doesn't screw the data over.

Now the biggest flaw is that they drive once or twice through a street and take that as their base for that region. Well the actual air pollution shifts with the surrounding conditions, e.g. Rain, Humidity, Season, etc. It's not a constant that is proportional. So that can go two ways: You could pick the worst day with a high pressure system and the readings will be off the charts resulting in intense doom predictions OR you could have the best day ever and you come to conclusion that 100k have no impact. Obviously those are extremes but if carefully selected, the data is screwed again.

Overall it seems a nice project but this data gives me the chills because it's way too easy to manipulate and it probably will be ending up with Governments like in California who are starting enacting ridiculous environment laws based on this super flawed data.

Nevertheless it's interesting to watch how this "experiment" of the "experiment" unfolds ..

aclima.io...
insights.aclima.io...

Wow so much hate towards a company doing research....
Let me guess, if the results came back clean and rosy you would be the first to say how legitimate their test are.

I think it's great news! It will give us a true understanding of pollution and how/when to avoid it when commuting, look at the Japanese and their face mask....

originally posted by: muSSang
Wow so much hate towards a company doing research....
Let me guess, if the results came back clean and rosy you would be the first to say how legitimate their test are.

I think it's great news! It will give us a true understanding of pollution and how/when to avoid it when commuting, look at the Japanese and their face mask....

Not at all - actually I am all for it but at least do it in a way that is not flawed from an engineering standpoint from the get go.

By your logic, you would never question anything. Even you see an issue with it. Nice!

a reply to: flyandi

I'm pretty sure they know about the fluctations in air quality depending on a plethora of variables. Perhaps they are taking those into consideration and also making a note about the conditions, time of day, time since last rain fall. You know, since those things are important too. Or maybe you should call google and let them know about their potential oversight. Or maybe leave the science to the scientists. Just a thought.

a reply to: flyandi

So who should be doing the research?

a reply to: flyandi

I question everything, but I also give an unbiased opinion until the data presents it's self, come on man un brain wash yourself.

Just look at the Denver data alone. They collected for one day and they base all their assumption on ONE day in the middle of Summer! Sorry but that's all hype with no value.

I don't say that collecting this data is bad but collecting it with "cars" that hit certain street once per year and base all their findings on that very single event is just flawed!

If you look at the Denver test data, it was one tiny small area and they ran 3 cars. Even with these 3 cars, they couldn't cover that area for every hour and you can see how they drove based on the rise of the NO2 because of time overlaps. What are they going to do in California? Sending 10k of cars on the streets to cover every street at every hour point? Sure that will help to decrease air pollution.

Anyway, I am standing by my opinion that this data is highly flawed just because they way it is collected. They will capture tremendous amount of data with little to no significants.

It would be better to position these environment sensors in strategic location and run them 24h/7d to collect data that actually can reach statistical significants.

a reply to: flyandi

You don't think that, over time, this data could add up? Or even be added to other programs that are also collecting data?

Collecting data is exactly what we should be doing. If Google can multitask several things at once, isn't that a good thing?

originally posted by: Woodcarver
a reply to: flyandi

You don't think that, over time, this data could add up? Or even be added to other programs that are also collecting data?

Collecting data is exactly what we should be doing. If Google can multitask several things at once, isn't that a good thing?

That's not how statistical data collection works. You can not just collect data for the sake of collecting data. The data needs to be in model that can be sustained, especially over time and clearly that is impossible to do with Street View Cars which just drive a street maybe three times a year if at all.

To make any sense, you would need at least drive once per week on the same day. That will give you 52 data points per collected position. Think about the logistic that is needed to run this endeavor properly. Might work for a very small part of a large city but that is not what Google will do. They will use the small sample data and enlarge it in statistical model that can be applied "abstractly" to every location.

Again, it's interesting how data collection will result in non-sense results and how the general public will buy it without questioning it considering that the EPA already has her fingers in this project.

a reply to: flyandi

How do you know they will do that? Is it impossible to imagine that maybe they know something about their processes that you don't know? Or maybe that you don't have all the info needed to make the assesment and claims that google is not qualified to collect environmental data correctly?

Do you have an inside edge on this story? What makes you qualified to say that google is going to mess this up?

originally posted by: Sremmos80
a reply to: flyandi

So who should be doing the research?

Everybody, or at least lots of people.

The issue isn't disinterest in collecting empirical data but in the presumption of that data providing actionable intelligence, it cannot.

originally posted by: flyandiGoogle and it's flawed Data experiment to further bolster Global Warming Ideas and non-sense laws!

It is always easier to point fingers than to look in the mirror and honestly answer the question of how the world is a better place with "ME" in it?
What kind of car do I drive?
What is my contribution on my share of the poisoning of the planet?
And then doing something, being the 'change' that I want to see!"?
What Lord Google does, or not, doesn't absolve me of my responsibility to the planet (and YOU as a feature of it)!