Facebook data downloads

[ckn]

I read this article earlier today about Facebook data and decided to get a copy of my own data.  Go to Settings and at the bottom of General, there's an option to "Download a copy of your Facebook data".  Give it a few minutes to get you the link and have fun looking through it.  Extract the data somewhere and open the index.htm file.

So, apparently I'm in the "Established Adult Life" friend peer group.  I don't think I've been called an "established adult" before, something very dull about that.

Facebook have given my contact details to:

- Spotify (I don't use it and never have so they've no reason to give them my details)
- Outdoor and Country (I don't use it and never have so they've no reason to give them my details)
- Just Eat UK (I don't use it and never have so they've no reason to give them my details)
- O2 Business (I don't use it and never have so they've no reason to give them my details)
- O2 (I don't use it and haven't for years so they've no reason to give them my details)

My adverts are tailored on certain criteria, including:

- Nicola Sturgeon
- Motorcycle speedway (although I doubt I've ever shown ANY interest in this in my entire life)
- Landscape photography (I enjoy it but I'm probably in the category of people who view this as nice but that's about it)
- Laphroaig (fair enough...)
- God (see "motorway speedway" above)
- Frankie Boyle (fair enough...)

I've no idea what sort of adverts those things would serve to me!

The bit that annoyed me most though was the IP and location tracking it did.  It hides the locations it tracks you by in the "Login Protection Data" on the security tab, probably because it knows most people won't scroll down that far.  I wouldn't mind so much but it's wildly inaccurate, it currently says I'm in Clacton-on-Sea.  If it's going to record something as sensitive as my IP address details then it should be more accurate with it, or stop pretending they know what they're talking about.

It's scary really what sort of stuff Facebook keeps and who it gives it to.  It's even more scary how wildly wrong its predictive behaviour is in the way it makes money through adverts.

[Farmduck]

I mustn't use it as much as you.

My contact was shared with ebay and Fiverr (one of those companies that finds 3rd-World people to work for $5/hour)

My ads were RL and Gold Coast Titans??? (dunno why the Titans but I use adblock anyway so I've never seen any ads)

I don't have tracking because I only use the Webz from my desktop at home.

[gingerjon]

Hmmm... so this is the list for my ads. I'd love to see an advert that brought together the first four.

Ads Topics

• Welsh language
• Non-metropolitan district
• Heavy metal music
• Steam engine
• Councillor
• The Printworks
• Boy
• Steak
• Paris Saint-Germain Rugby League
• West Sussex
• Major League Baseball
• Sculpture
• Internet forum
• Community (TV series)
• American rock
• Caregiver
• Christian music
• US Boulogne
• Eastbourne
• Wigan Warriors
• BBC Radio
• Visual arts
• Sports
• Royal Academy of Dance
• Wales national rugby league team
• Viola
• Vitis
• Boulogne-sur-Mer
• Eating
• Royal National Lifeboat Institution
• Performing arts
• SV Werder Bremen
• Rugby league
• Genesis Bikes
• Women's rugby league
• Financial Times
• R.E.M. discography
• National Autistic Society
• Buckinghamshire
• Institution
• Country
• S4C
• Football team
• Irn-Bru
• England
• Music festivals
• Hay
• Brian Lara
• Art museum
• Rugby Football League
• Fine art
• Diners
• Education
• Zanzibar
• Beer
• Regions of England
• Zoom (TV channel)
• Love (John Lennon song)
• Rugby League World Cup
• San Francisco Giants
• Wales
• Bliss (2007 film)
• Seafood
• Handball at the Summer Olympics
• Labyrinth
• Cake
• European Handball Federation
• North America
• Amersham
• Fringe (TV series)
• Songs of Leonard Cohen
• Beach handball
• Association football (Soccer)
• Music
• Atlantic Ocean
• Nature
• Folk music
• Newspapers
• St Leonards-on-Sea
• Sovereign state
• Loch Ness
• TV
• USA Rugby League
• Gothamist
• Black Books
• Food
• Search and rescue
• Entertainment
• Africa
• Roaring Twenties
• Cognition
• Alternative rock
• Freeview (UK)
• Coffeehouses
• Group action
• Medicine
• Europe
• East Sussex
• Liberal Democrats
• Royal Opera House
• Yorkshire County Cricket Club
• City of London
• Atlas Obscura
• Time Out (magazine)
• Physical exercise
• Nestlé
• The Core
• Magazines
• Soft rock
• Technology
• Virtue
• Choir
• Opera
• Peep Show (TV series)
• Singer-songwriter
• Popular music
• Phil Spector
• Physical fitness
• Vintage Everyday
• TIDAL
• BBC Radio Cymru
• Cricket
• Sussex
• Violin
• Biology
• Culture
• Cycling
• Motorcycle speedway
• Leonard Cohen
• Seed
• South East England
• Hamburger
• Budapest
• Tea
• Patient
• The Daily Mash
• Computers
• International Handball Federation
• Satire
• Arts and music
• Dance
• Owen Jones (writer)
• Nerds (candy)
• Hastings, New Zealand
• Property
• UK Cycling Events
• Rhythm and blues music
• Nespresso
• Ballet
• Veterinary physician
• Bleachers
• Halite
• National Eisteddfod of Wales
• Super League
• Learning
• Bradford City A.F.C.
• Coastal trading vessel
• Cambridgeshire
• Liverpool
• The Observer
• Sound
• Super League I
• Bicycle
• World music
• Chicago Cubs
• National Theatre Live
• Intelligence
• Chiltern Hills
• Richard Dawkins
• Jerwood Gallery
• Friendship
• Trattoria
• Rock music
• English National Ballet
• YMCA
• St George's Hospital
• Recreation
• The Vines (band)
• Her Majesty's Coastguard
• Heart (novel)
• Team sport
• Police
• Love
• Futsal
• National Rugby League
• Bexhill-on-Sea
• Charity and causes
• BBC Radio 3
• Home video game console
• BT Sport
• Father Ted
• Seinfeld
• Pelican
• The Beach Boys
• Parkrun
• Hospice
• Blues music
• Coffee
• Kino (band)
• Travel
• Concerts
• R.E.M.
• ClickHole
• Journalism
• Rother
• Orchestra
• Friends (The Beach Boys album)
• The Royal Ballet
• Sussex County Cricket Club
• Castle (TV series)
• Sega Genesis
• St Leonards, New South Wales
• Nutrition
• World Triathlon Corporation
• Young Vic
• America (band)
• Website
• Music genre
• Cape Verde
• Hastings
• MLB.com
• Writing
• BBC
• Waikato Bay of Plenty Magic
• Lifeboat (rescue)
• EHF Champions League
• Health care
• Welcome (2007 film)
• Netball
• Saskatchewan Roughriders
• Slate (magazine)
• Aesthetics
• Online
• Classical music
• Canadian football
• In Time: The Best of R.E.M. 1988–2003
• Video games
• England national rugby league team
• British Cycling
• Les Diables Rouges
• England Netball
• Sussex Police
• Ship
• Chesham
• Bollywood movies
• Emotion

[Farmduck]

Ginger, you must live on FB. Do you click a lot of likes? I never click likes. I've been on FB since 2008 and the only reason I used it was because my family was spread out across OZ and the Chatbox was so much cheaper than phone calls. I've never had a news feed so I guess that gives them less material because I'm not clicking news items.

[gingerjon]

41 minutes ago, Farmduck said:

Ginger, you must live on FB. 

I don't live near anyone I grew up with, none of my uni friends are local to me now and my family are miles away, or in other countries, so a lot of my interactions are via FB. So it's not remotely surprising that there's a lot of stuff.

There are a few there which show that the system isn't perfect though.

For example, Kino (band). I've never heard of Kino (band). What I do do though is follow a local arts venue called kino-teatr who put their listings on Facebook.

[Farmduck]

Machine learning is slowly getting a clue. About 6 years ago I bought a lawnmower on ebay and every week for the next 3 months I got one of those "Things you might be interested in" emails including a range of lawnmowers.

In my pre-Ad Block days I used to get lots of ads from Jewish dating sites. I'm not one of the Chosen People but it was back in the Myspace peak. I spent a lot of time on the Myspace News and Politics forum and read a lot of links to Israeli newspaper articles (obviously pre-paywall, too.)

[ckn]

5 minutes ago, Farmduck said:

Machine learning is slowly getting a clue. About 6 years ago I bought a lawnmower on ebay and every week for the next 3 months I got one of those "Things you might be interested in" emails including a range of lawnmowers.

In my pre-Ad Block days I used to get lots of ads from Jewish dating sites. I'm not one of the Chosen People but it was back in the Myspace peak. I spent a lot of time on the Myspace News and Politics forum and read a lot of links to Israeli newspaper articles (obviously pre-paywall, too.)

I don't agree.  I get those same adverts.  I buy something on Amazon then get adverts for the next month for the same stuff.  Why would I want more of the same thing I've just bought when I'd expect it to last for a good while?  I bought it, surely that means I don't need another.

I remember a couple of years ago that my gran wanted the new Russell Watson CD for Christmas, I got adverts for the same CD until nearly Easter!

[JonM]

Interesting. Mine is pretty much spot on about me, other than the following Ad topics:

Abington Township, Montgomery County, Pennsylvania (there's a village a few miles from me in Cambs called Abington, where I've taken pictures while out running)

Saxophone  - (my least favourite instrument)

Ladies' Gaelic football (no idea?)

Phoenix (band) (who?)

Only two advertisers have my details - Paypal & KLM.

VPN seems to be doing a pretty good job of obfuscating where I am.

[JonM]

Are they claiming that this is all of the data they keep about you? I would have expected that they tracked & recorded which videos you watch, which photos you look, whose profiles you view etc etc.

[gingerjon]

4 minutes ago, JonM said:

Are they claiming that this is all of the data they keep about you? I would have expected that they tracked & recorded which videos you watch, which photos you look, whose profiles you view etc etc.

It's obviously just a skim off of some bits. You can also download every comment you've ever made. Which is ... as dull as you'd expect.

[Shadow]

2 hours ago, JonM said:

Ladies' Gaelic football (no idea?)

Yeah right 

[JonM]

1 hour ago, Shadow said:

Yeah right 

I'm in various Irish music facebook groups and I watched some videos of the Women's RLWC on facebook, so maybe it's put 2+2 together? If it was based on random lechery, it's far more likely to show up as ladies' track & field  

[heartofGold]

 

God mine is boring:

Oxford English Dictionary

Bromley

Anonymous

Books

Grapevine, Texas (??)

eBay

o2

[gingerjon]

2 hours ago, JonM said:

I'm in various Irish music facebook groups and I watched some videos of the Women's RLWC on facebook, so maybe it's put 2+2 together? If it was based on random lechery, it's far more likely to show up as ladies' track & field  

"College athetics (female)"

[Futtocks]

My Ad Topics

Sports (wow, that's specific!)
Rugby League (but of course)
Michael J. Nelson (I have been a fan of his work for a long time, so okay)

And that's the entire lot.

[ckn]

6 minutes ago, Saint 1 said:

I had 790 ad topics, so a list over treble the size of Gingerjon's! I'll save everyone the trouble of scrolling through it. Lots of  accurate stuff such as specific trance DJs, strength and conditioning and rugby league stuff, politics and economics too.

Some less accurate stuff (as expected) such as:

Hardcore punk

English Defence League

Tibetan Buddhism 

Vegetarianism

Homosexuality

Breastfeeding

Nick Griffin

Far-right politics

I found some of the other stuff more interesting. I've deleted 650 friends since getting Facebook in 2008. It's not even like I'm particularly exclusive with my friends - I still have over 900 friends. Other bits are more creepy - being able to see who I have called and messaged, at what time and date, for seemingly everyone in quite a substantial period of 2016 and 2017. Also seeing the contacts I have saved on my phone, even if I have since deleted them.

I'd be interested to dig deeper into the data - probably because my job is as an analyst for a data company. I'd love to see confidence estimates for each topic, or to see which topics are most predictive of responses etc. 

That's the thing though, Facebook don't have to work hard to convince Marketing people who don't "get" data mining that you really are a hardcore punk loving, EDL supporting buddhist vegetarian homosexual with breastfeeding fetishes. 

I learned a lot about it when I worked with a telecoms manufacturer's fraud detection division where they sold neural network based solutions that were scarily accurate at flagging fraud within one call from a user.  It's easy flagging lots of stuff up, the hard bit is flagging the right stuff.  For example, my wife sometimes uses this machine when she can't be bothered firing her own one up (I'm confident that my auto-complete in my browser won't get me a smack round the back of the head) so I get advertising based on a mix of her stuff and mine, I'm convinced that google thinks I have a shoe fetish because of my wife's Imelda Marcos like obsession with shoes.

[graveyard johnny]

lockjaw