Facebook Has Released a Map of Coronavirus Symptoms Crowdsourced From Its Users (technologyreview.com) 45
An anonymous reader quotes a report from MIT Technology Review: Facebook has released a map showing the proportion of people who say they have experienced coronavirus symptoms in each state in the US. The data was gathered from more than one million Facebook users who filled in a survey created by Carnegie Mellon University about whether they were experiencing symptoms like a cough or a fever. The map, which goes down to county level, will be updated every day. Facebook says it will create similar maps for other countries in the coming days and weeks, also based on survey responses.
Knowing who is experiencing symptoms and where could help health officials and governments to prepare for surges of hospital cases and decide where to allocate resources like ventilators, face masks, and personal protective equipment. Given the shortage of tests, and long delays for results, this map could be useful in helping to predict where COVID-19 hot spots are forming across the U.S. Obviously, the map is only as good as the data that's used to create it, and as you can see for yourself, vast swaths of the map don't have enough participants to yield reliable data. The map is part of Facebook's work with CMU and the CDC to predict the coronavirus's spread.
Knowing who is experiencing symptoms and where could help health officials and governments to prepare for surges of hospital cases and decide where to allocate resources like ventilators, face masks, and personal protective equipment. Given the shortage of tests, and long delays for results, this map could be useful in helping to predict where COVID-19 hot spots are forming across the U.S. Obviously, the map is only as good as the data that's used to create it, and as you can see for yourself, vast swaths of the map don't have enough participants to yield reliable data. The map is part of Facebook's work with CMU and the CDC to predict the coronavirus's spread.
I'm sure it is very accurate (Score:5, Insightful)
Re: I'm sure it is very accurate (Score:1)
Maybe you might have a handful of douchebags that would answer a unique survey dishonestly but it's unlikely to be over a million unique douchebags. This isn't some 24 hour news pole that samples 20 people and call it "most of America," but something a little more rigorous, even if they don't release all the data for peer review--we know Facebook didn't give a shit about privacy to be a reason. I wonder if this could be used as a way to account for areas of the country that simply don't have adequate health
Re: (Score:1)
Re: (Score:2)
Re: (Score:2)
Yes, 'self selection' is a problem with all surveys and non respondents can a real problem.
The solution to creating a better sample is to understand the demographics of those who answer, then adjust your results based on it. Any good survey does this already, but FB is surely the best at understanding it's respondents. Facebook already knows or has a good guess on your sex, age, income, religion, height, weight, political leanings, etc., they even have those demographics on people who skip the survey. O
Re: (Score:2)
Of course people lie. But why would they lie more on average in one location than other location? If not, the data is still reliable for comparing one location to another.
Re: (Score:2)
But why would they lie more on average in one location than other location?
Perhaps:
One location is extremely bored city dwellers couped up in their apartments.
Another location is semi-bored suburbanites doing home maintenance and yard work.
Still another is active rural folk doing the same stuff they've always done.
But hey, I suspect city dwellers think everyone is just like them.
Re: (Score:2)
But hey, I suspect city dwellers think everyone is just like them.
Having lived in both cities and in rural areas for approximately half my life each, everyone is just like them. The rural-urban divide is a myth.
Sure, there are minor differences:
Re: (Score:2)
The details might be slightly different, but fundamentally, the behaviors are basically the same. IMO, the only people who believe that the rural-urban divide is real are people who either haven't lived in both or who are deliberately trying to create that divide for their own political gains.
Except you didnt detail any understanding of rural life. The details you gave indicate that you have slapped the rural label onto suburban living. You are the ignorant one.
Re: (Score:2)
I lived in a town of 8,000 people for more than 22 years, more than an hour's drive from anything that could even remotely be called a city, and 2.5 hours from the nearest real airport. Suburban, my a**.
The biggest difference I see is that in an urban area, I don't constantly run into people I know while shopping at Wal-Mart. In a small town, people gossip more, and social pressures to conform are much higher because the small percentage of overly judgmental people have access to more information about w
Curious as to your sampling methodology (Score:1)
While degenerates really make themselves stand out, they do not make up even a large fraction of our society.
Re: (Score:2)
From Contact Tracing in the Real World [lightbluetouchpaper.org] by Ross Anderson [cam.ac.uk]
Anyone who’s worked on abuse will instantly realise that a voluntary app operated by anonymous actors is wide open to trolling. The performance art people will tie a phone to a dog and let it run around the park; the Russians will use the app to run service-denial attacks and spread panic; and little Johnny will self-report symptoms to get the whole school sent home.
Re: (Score:2)
People who troll social media for shits and giggles are well and truly outnumbered by bored mindless people browsing social media in the first place. They won't be reflected in the stats in the slightest.
Paid for and targeted shilling campaigns are a different matter entirely.
But then self selection bias will be far worse than the above problems combined.
"dataforgood" (Score:1)
As opposed to the default usage...?
Facebook users (Score:4, Interesting)
Re: (Score:2)
COVID-19 is actually a medical marvel that needs to be studied. It has virtually eliminated all cases of plain old flu and pneumonia.
Re: (Score:2)
Perhaps this is more accurately described as a map of attention whore Facebook user density?
I'm sure these guys: https://delphi.cmu.edu/ [cmu.edu] who ran the actual survey didn't at all account for that. Thanks for your help armchair engineer.
oh Great (Score:1)
Re: (Score:2)
Er (Score:2)
Garbage in, Garbage out (Score:1)
This has got the be the most irresponsible and reckless thing I've seen Facebook do in a long.... well.. since last week I guess.
So, based on a survey, and probably a small dataset, querying people who are going to have media-induced bias into believing they have symptoms of this virus... goes into something Facebook wants people to take seriously.
Re: (Score:2)
They aren't saying those who have symptoms of COVID-19 necessarily actually have it, but my bet is it correlates pretty well.
A million people is far more people than just about anybody else could do. I wouldn't call that a small data set. And on the map, they do show regions where there isn't enough data. Like it or not, a high percent of the population uses Facebook (not sure about biases in terms of who responds to surveys). They are probably in a better position than anybody to get a good random sample (
dataforgood my ass (Score:2)
Thank You, Hypochondriacs R' Us (Score:4)
I'm A Sick Kid Now!
Now to see if it correlates with other data (Score:3)
US Health Weather Map [healthweather.us] has a map of excess fevers based on smart thermometer data. Do the hotspots on its map correlate to those on Facebook's map? Do they correlate to known COVID concentrations?
Re: (Score:2)
Re: (Score:2)
You seem unaware that both maps measure percentage of population, not absolute number of people.
Re: (Score:2)
Re: (Score:2)
To a large extent it does. Major exception is that Facebook does not have as many sick people in Florida. I guess either old people still do not use Facebook (not my experience), or they are more likely to use a smart thermometer.
Re: (Score:2)
Do the hotspots on its map correlate to those on Facebook's map?
Do you know how to open up two webpages at once? Or do you think multi-tasking is just some fad?
Re: (Score:2)
Unfortunately, I have a full time job plus I need to care for kids without preschool, so I don't have the free time to compare thousands of rural counties one by one. Maybe you do?
Has the country gone so hypocondriac? (Score:1)
Self reporting medical tests... (Score:2)
...have at least a 50% overlap to "who has hypochondria?" maps.
List of Covid19 symptoms
Common symptoms:
fever.
tiredness.
dry cough.
Some people may experience:
aches and pains.
nasal congestion.
runny nose.
sore throat.
diarrhoea.
At least one item on the list above in my experience is present in every MN resident from like, October to May. How is this useful?
Re: (Score:2)
> At least one item on the list above in my experience is present in every MN resident from like, October to May.
Diarrhea? It's always diarrhea.
This is a map of moron density. (Score:2)
Re: (Score:2)
Not a "map of symptoms" (Score:3)
English, does anybody speak it?
It's a map of people who report the symptoms of COVID-19
Re: (Score:1)
And even then, only those in the USA.
No symptoms in the rest of the world :rolleyes:
Re: (Score:2)
The /. summary and the article attached to the data both specify US - they just forgot it on the title to the graph. I think you're probably over-triggering yourself here.
I'm sure that data won't be used irresponsibly (Score:2)
So Facebook collects a survey, secretly recording your GPS location to release a map on things we already know through traditional and accurate methods such as medical assays.
This is on par with those government tracking apps they're trying to build. No use to anyone except to track the citizenry in great accuracy.
Some Maps Shouln't Be Made (Score:2)
And worst of all, it misse