Why a walk starts at your front door
A hike starts at a car park. A dog walk starts at your door. It goes round the block, through the park or along the river, and ends where it began. On one map you hardly notice. Lay three or four on top of each other and every line meets at the same spot, and a map search does the rest.
That applies to any app that shows routes publicly, not one in particular. Most offer a privacy zone that removes the first and last few metres before anyone else sees the route. How well that works has been studied properly.
How much does a hidden start point protect you?
Two research teams tested it on public data from a large fitness platform. Dogs had nothing to do with it, but the mechanics are the same: the route starts at home, and the app cuts a circle out around the start.
In 2018, researchers at the University of Illinois looked at one month of activity, around 21 million uploads from 3 million people (Hassan et al., USENIX Security 2018). Their tool was geometry. When a route is cut at the edge of the circle, its visible end sits exactly on that edge. A handful of such ends is enough to rebuild the circle, and with it the centre. For people with more than one protected activity they found 84 per cent of protected locations, and 95.1 per cent from three activities upwards. Even with the largest radius the platform offered, roughly a kilometre, they still found 44 per cent.
A team at KU Leuven went further in 2022, adding 1.4 million activities, the street network and the distance shown for each outing (Dhondt et al., ACM CCS 2022). If you know how long the whole route was, you can work out how far it is from the visible end to the door. That way they located up to 85 per cent of zones with a 200-metre radius, the most common setting, and 55 per cent at one kilometre. Zones whose centre had been deliberately shifted did not hold up either. Where streets were sparse, the attack did better still.
Both teams recommend the same things: bigger zones, a randomised edge instead of a fixed one, and a rounded distance instead of an exact one. The smallest zones in either study had a radius of about 200 metres. Dog-walking apps often trim far less.
When you are not at home
The address is only half of it. From one walk, someone learns a neighbourhood. From twelve, they learn that on weekdays you head out at about 6:40 for half an hour, that the evening walk is longer, and that one week in February nothing was recorded at all. No single entry gives much away. Together they do.
How far this can go was shown in early 2018 by an Australian student. On a sports service’s worldwide activity map he spotted military bases in Syria and Iraq, traced by the running loops of the soldiers stationed there (ABC News, 29.01.2018). None of the individual recordings looked unusual.
What actually helps when you share
- Don’t share in public. With a private account, where only people you have chosen to give access can see your walks, no stranger can collect them. That deals with the address and the routine at once. Also check who already follows you.
- Set the privacy zone to the largest your app allows. According to both studies, radius helps most. In the countryside it helps less, because only a few paths lead into the zone.
- Don’t share every walk, and not always the same day’s. A pattern needs repetition.
- Look at your own walk from the outside. Open a shared walk from a second account, or on the phone of someone who follows your dog, and check where the line begins. Then you know what others see.
- Mind the gaps. Two weeks without a walk tell anyone reading along that you are probably away.
- Starting the recording a few streets away helps against the address. It does nothing about the times.
Two checks for any app
Whichever app you use, two looks at your phone’s settings are worth the time.
- Location “Always” or “While using the app”? “Always” can have a reason, for instance so recording carries on with the screen locked. A good app tells you that reason and also records with “While using”. Try it on a short walk before you allow more.
- Which other permissions does it want? Each should match something you can see in the app. A microphone with no voice feature, or access to your contacts with nothing to invite anyone to, is worth asking about.
What Barnaby covers
By default, Barnaby trims the start and end of every walk before anyone else sees it, from the very first one. The default is 50 metres, adjustable from 25 to 125; you can also trim just one end or switch the protection off. The cut falls exactly at the chosen radius, with no randomness, and distance and duration count the whole walk. That is exactly what both studies exploit. The protection removes the point, not the pattern.
New accounts are public, except for users under 18. If you would rather yours wasn’t, switch your account to private; the FAQ explains how. What the cut looks like on the map is shown in the home zone section of the home page.
Sources
- Wajih Ul Hassan, Saad Hussain, Adam Bates, “Analysis of Privacy Protections in Fitness Tracking Social Networks -or- You can run, but can you hide?”, USENIX Security 2018. Data set, circle attack, 84 and 95.1 per cent, 44 per cent at the largest radius.
- Karel Dhondt, Victor Le Pochat, Alexios Voulimeneas, Wouter Joosen, Stijn Volckaert, “A Run a Day Won't Keep the Hacker Away: Inference Attacks on Endpoint Privacy Zones in Fitness Tracking Social Networks”, ACM CCS 2022 (open-access PDF). Distance-based attack, up to 85 and 55 per cent, countermeasures.
- ABC News (Australia), 29 January 2018, report on the military bases visible on the activity map.
We read both studies in full on 29 September 2026. They are based on a fitness platform, not on dog-walking apps. What carries over is the mechanism, not every figure.
