I used my Google Maps location history to see if my boyfriend and I could have met before we did

Map showing all the places my boyfriend and I had both been at different times before we met. Main roads around the city were very common.

When I met my boyfriend at the end of 2020, I sometimes felt a tiny tinge of regret that we never got to make an adorable meet-cute story, despite living in the same city for five years and attending the same university for three. Our story isn’t totally unique, we started speaking on Bumble, and we had a lot in common so, in a trough between COVID waves, we decided to have a coffee. Not exactly the story you tell for generations.

But on a deeper level than that the question plagued me, why is it that we never connected despite spending a lot of theoretical time in the same places at similar times? We even did the same anthropology course as an interesting elective, although a year apart.

As dedicated introverts, I realise the answer is probably that neither of us would have struck up a conversation with a stranger in the wild, so even if we had been in the same room at the same time we probably wouldn’t have connected on the same level as in an explicit dating environment like an online dating app. However, I still wanted to know just how many opportunities we passed up to find each other years before we actually did.

Thanks to Google’s timeline collecting geo-location data I thought that I could potentially find a real answer to how fatefully our paths crisscrossed before we met.

Dates which Google had recorded locations for, starting in 2015, and ending in 2020 when we met. Each column represents one week. The number of records refers to how many 10-minute intervals had at least one record of coordinates with a reported accuracy of at least 111m.

Only around 43.5% of the time between 2015 when I moved to Cape Town and 2020 when we met had records for the both of us. The times without records could have been due to breaks in Google having permission to record location history or a lack of mobile data (which is notably expensive in South Africa). This means that there were still to be blind spots no matter how many (if any) path-crosses I found.

This chart shows the distance from each other over time for periods where we both had records. Blue dots indicate coordinates that were 0m apart, but the accuracy buffer on that is 111m.

With this in mind, I started by measuring the distance from each other at any given time. This showed that we had been recorded at the same coordinates (with a reported accuracy of less than 111m) at the same time a total of 29 times in the periods where we did have records.

It also confirmed that most of our chances did occur in the period when we both studied at the same university campus around 2015–2017, but most notably in 2016. After that, we drifted apart before we would ultimately drift back together.

Map showing the locations of the closest brushes, coloured by the time in minutes in which we were both recorded at those coordinates. Most of the 0-minute (in the same 10-minute time interval) encounters took place on the university campus.

When plotting the locations of the closest encounters the majority of them do occur on the university campus, with another small cluster down at the shopping centre nearest to campus. I mapped our movements for each of the days we could have met on campus in more detail.

Often in the 43.5% of the time that was recorded, we not only cross paths for brief seconds but spent plenty of time swirling around each other moving between buildings on campus. It’s hard to say how many times we could have attended lectures in adjacent theatres, queued for lunch at our favourite stall, or sat down at computers next to each other in the same labs, but it is evident that this likely happened semi-regularly. So we had many opportunities to meet, but with unseeing eyes and heads in the clouds, the conclusion I draw is that it just wasn’t the right time yet.

Many path-crosses occurred under the trees outside the social sciences building. We both often got coffee from the shop just off to the right.

Want to recreate this yourself?

• Full-length analysis and code: https://chan.co.za/how-fateful
• GitHub: https://github.com/channon036/how_fateful

Helpful resources:

• D. Kahle and H. Wickham. ggmap: Spatial Visualization with ggplot2. The R Journal, 5(1), 144–161. URL http://journal.r-project.org/archive/2013-1/kahle-wickham.pdf
• Macarulla Rodriguez, Andrea & Tiberius, Christian & Bree, Roel & Geradts, Zeno. (2018). Google timeline accuracy assessment and error prediction. Forensic Sciences Research. 3. 240–255. 10.1080/20961790.2018.1509187.
• movable-type.co.uk/scripts/latlong.html#https://www.geeksforgeeks.org/program-distance-two-points-earth/#:~:text=For%20this%20divide%20the%20values,is%20the%20radius%20of%20Earth.