Apparently iPhones running iOS4 periodically record the device’s Latitude and Longitude and timestamp in one of the numerous SQLite databases in phone’s backup and configuration system. Several sites have provided methods of extracting the filename from encrypted backup filesystem. Once located, the contents of the file in question can be investigated using any number of techniques including the SQLite Manager plugin for Firefox.

This system of logging the location of the iPhone to file has been in use for some time, as pointed out in this excellent blog post by Alex Levinson, co-author of the book iOS Forensic Analysis.

Many news reports have claimed that this information is not transmitted back to Apple, but this revealing post from F-Secure Security Labs suggests that by default this location information is sent back to Apple twice a day so that Apple may collate a database of Wi-Fi locations.

The co-ordinates stored in the location database are also readily available to any iPhone application installed on the phone through the iOS API. About halfway down the page in the What’s New in iOS 4 page on the apple website we find this quote:

Background location – Navigation apps can now continue to guide users who are listening to their iPods, or using other apps. iOS 4 also provides a new and battery-efficient way to monitor location when users move between cell towers. This is a great way for your social networking apps to keep track of users and their friends’ locations.

Visualising The Data

Since I had access to the raw data in a convenient SQLite database format, I decided to create my own visualisation. I wrote a PHP script that selected some of the rows from the tracking database and generated a series of images using PHP’s GD Image Library. The location of each point was highlighted orange and allowed to fade gradually over time, creating a “heatmap” effect.

The resulting animation shows the location of my iPhone during the period July 2010 to April 2011. During that period I made several interstate trips, but this animation focuses on the Sydney, Newcastle, Hunter Valley and Port Stephens areas.

At the beginning of the tracking period I lived in Sydney, but traveled to Newcastle regularly for work. Numerous hotspots can be seen in the Wyong area, this is because I usually stop at the Twin Service Stations at Wyong to break up the journey along the F3. Here I usually check my email and make a few phone calls. Later parts of the animation show more activity in Newcastle and ends with a recent holiday to Port Stephens.

The animation is also available in High Definition on YouTube.

In total there were 4181 rows of co-ordinates in my tracking file, each one with a timestamp and estimated horizontal accuracy. The information was inside a table called called “cellLocation”. The dataset went back some 275 days. Presumably the collection of gelocation information in this file began when I installed iOS4 in the middle of last year.

Clearly, storage of such information in an unencrypted format by default on every iPhone in the world could be a potential privacy issue, but it’s not like this information hasn’t been collected before. Google, Foursquare, Facebook and Twitter all collect information on the whereabouts of it’s users whenever they can. These companies gain access to the more accurate information as they have access to the device’s built-in GPS. Each time you check in on foursquare, tag yourself in a Facebook place, or geotag your tweets you are explicitly opting in to these information collection services.

What’s different this time is that now the end user has the ability to actually view the information that is being collected about their location by their mobile devices.

This one of the first times that this information has been so readily accessible. Over 100 Million iPhones have been sold, making this one of the most widely and publicly available geolocation datasets. Unfortunately, this information is fragmented across tens of millions of computer in the form of iPhone’s backup files.

Various open source applications are appearing that allow users to visualise their movements based on the contents of the iPhone tracking file. Form example, an OSX application released by Alasdair Allan and Pete Warden, and a Java applicationn that outputs the waypints in .kml format ready for viewing in Google Earth.

I can say from my own experience that reviewing the location information from my own phone was a thought provoking exercise. I started wondering who collects information about my location. I’ve already mentioned Twitter, Foursquare, Facebook etc. but there’s many more less obvious examples of tracking going on.

The geolocation information has been derived from the location of nearby cell towers. Information of this nature is collected by mobile providers so they know which cell tower to route the current call through. The Latitude and Longitude seems to be accurate to within a few kilometers at worst.

My current mobile carrier, Optus, would have a wealth of information about my whereabouts at any point in time. The data they have would be almost exactly the same as the information I extracted from the iPhone database.

Mobile carriers collect this information routinely

We can’t stop the mobile carriers from collecting this information because they need it for the the robust and efficient operation of their network. What we should be doing however is questioning is what their policy with respect to this information once they collect it.

For example:

• Exactly what information is collected and retained
• How long will they retain this information.
• What is the carriers policy on giving up this information to third parties (Advertisiers, partner companies, law enforcement).

The location of a users phone at a particular time location is routinely used in criminal cases. This is how former Justice Einfeld was convicted of lying under oath in early 2009. Police then used mobile phone records to trace Einfeld as he drove south over the Spit Bridge.

If the mobile phone companies are willing to give up this data for case involving a speeding fine, then they are very likely to be willing to divulge the same information to authorities under more serious circumstances.

I’m hoping this post raises awareness of the fact that location information can be, and is being collected from every mobile phone user. Perhaps in the near future it will be routine to ask more about your mobile carrier’s privacy and data retention policies when choosing a network.

Occasionally, I use twitter to broadcast information about local events through the @newcastlemusic account. For example, one of my recent tweets was about the Newcastle Improvised Music Association (NIMA). The association is organising weekly jazz nights at the Grand Hotel in Bolton St, Newcastle. That’s Newcastle, NSW, Australia, not just any Newcastle.

Within a few hours that information had been retweeted by @NewcastlePlaces. All well and good you may well say … except that they had linked to a map for the event that placed it somewhere in the UK. The Schmap indicated that the jazz nights would be taking at The Grand Hotel in Newcastle Upon Tyne.

I replied to their tweet and, to their credit, they removed the tweet and the map immediately, however the incident made me curious. How did this happen? Had it happened to anyone else? So I took a look at their account to see what their other tweets were like.

Places and Events in Newcastle. Indeed, but which Newcastle?

The @NewcastlePlaces account had tweeted nearly 800 times. The bulk of the tweets were regurgitating someones elses information about an upcoming event, then adding a handy map for the event on the Schmap.it service. Their twitter icon featured an image of Newcastle Upon Tyne in the UK, so I assume the operators of the account are based there. The majority of their followers seem to be from the North East of England.

Schmap’s about page describes the service as ” … a location technology service provider and local publisher, with cutting-edge expertise at the intersection of local, social, commercial and the real-time web”.

If they’d made a mapping mistake with my tweet, perhaps they’d made others too.

So, using a little local knowledge, I decided to read through their most recent 400 tweets to see if I could find some more #schmapfail. To my astonishment I identified 60 tweets that contained a Schmap showing an incorrect location.

Here is an example of one of their more recent mapping failures:

RT @1233newcastle A protest is being held at the top of King Edward Park against the redevelopment of th… http://schmap.it/3QTQrx?a <--MAP
5:44 PM Feb 23rd via schmap.it

Schmap showing incorrect location for the King Edward Park protest

This tweet was posted by 1233 ABC Newcastle, a radio station located in Newcastle, NSW, Australia. For some inexplicible reason, @NewcastlePlaces had mapped the protest as taking place at the King Edward VI School in Morpeth, UK, over 15,000 km away from it’s actual location.

Another recent mapping failure involved a retweet of @mediahunter‘s callout for interns for autumn 2011 at Sticky. @Mediahunter is a prominent Twitter user from Newcastle, NSW, Australia. His twitter profile clearly states that he is indeed based in Newcastle, Australia.

Map for Sticky: http://schmap.it/Sn2WQv?a – RT @mediahunter Closing applications today for Autumn internships at Sticky. Hurry!!! http://…
Tuesday, 15 February 2011 9:58:56 AM via schmap.it

Map showing incorrect location for Sticky

The Schmap incorrectly indicates that Sticky Ads is located at 152 Hawthorn Road, Ashington, a small suburb located about 20km North of Newcastle Upon Tyne in the UK. This is in fact the location of Sticky Fingers, a takeaway food shop.

Here’s yet another example that should have raised some eyebrows. The tweet specifically mentions that the poster is at “Australia’s most famous beach”. Yet somehow the Schmap ends up pointing to a small frozen food supplier in Gateshead, UK.

Icebergs: http://schmap.it/RIqs3H?a – RT @Maya_Abeille It was lunch at Icebergs at Australia’a most famous beach (baked herb encrusted se…
Wednesday, 5 January 2011 3:40:51 PM via schmap.it

Lunch at Icebergs at Australia's most famous beach

One final example, this time involving the location of a dead body.

Body in Singleton pool: Police are investigating the discovery of a body in a backyard p… http://schmap.it/JstXbT?a (via @newcastleherald)
Friday, 22 October 2010 9:59:35 AM via schmap.it

A dead body, located at Singleton, Winn, Saunders Solicitors by schmap

So the #schmapfail suggests that there was a body at 232 Shields Road, Newcastle Upon Tyne. This is in fact the location of the offices of Singleton Winn Saunders Solicitors. I’m sure they appreciate the insinuation that they are connected with a murder case.

So we can see a pattern emerging here. I imagine the process that @NewcastlePlaces follows goes like this:

1. Search through the tweets of anyone who’s twitter location contains the word “Newcastle”
2. Determine if they are tweeting about an event
3. Search for a business in the Newcastle Upon Tyne area that matches the event location
4. Retweet and include a Schmap that points to the UK business

It’s not hard to see why this process fails so often. Someone from Newcastle Australia is hardly going to be tweeting about an event in the North East of the UK.

In addition to the 60 incorrectly mapped tweets I identified there were also five apologies. Presumably these were responding to people who had complained about their events being incorrectly mapped.

All five apologies are listed below:

@newcastlemusic Oh, thanks for the info. We’ve already removed the tweet.
Thursday, 17 February 2011 2:03:39 PM via schmap.it

@SCOLLEYart thanks for letting us know, we have removed the RT
Wednesday, 16 February 2011 1:47:17 AM via schmap.it

@dmilor whoops! thanks for letting us know, we have removed the RT
Saturday, 12 February 2011 2:25:40 AM via schmap.it

@BacchusHeather we have removed the RT
Wednesday, 26 January 2011 3:09:54 AM via schmap.it

@Bibbleq oh dear, well if you give us the correct address we can fix the schmap for you … or we can also delete the tweet if you like
Wednesday, 15 September 2010 7:38:28 PM via schmap.it

Sixty-five Schmaps out of four hundred means a SchmapFail rate of 16%.

Is this a twitter account you want to be relying on for your local event information?