Bubble Social, connecting those around you.

What was Bubble?

Bubble Social, or Bubble for short, was a location-based social media app that allowed you to connect with people from 1/4 of a mile to 10 miles around you. Bubble Social was my first true love and brainchild. 

Bubble was open from December 2013 to January 2016. Bubble closed its doors in 2016 after a long-winded round of unsuccessful capital investments. While at Bubble, our team grew to 14 developers all that were working passionately to grow the future of location-based social media.

Bubble Social had big plans to change how social media would work all together by focusing on creating experiences for users by making it easier to create events and reward users for going out to these events.

 

 

We wrote a lot of code…

Lines of Code Written

Code Bases within Project

Hours Spent

The rise & fall of location-based social media.

In 2015, we were convinced that location-based social media was the bleeding edge of how people would connect. After seeing Yik Yak with a $400 million valuation, we were convinced that investors wanted to see more location-based social media apps. We had started our project years before this wave, so we were perfectly positioned to be that player. The social media investment 

In 2015, we were convinced that location-based social media was the bleeding edge of how people would connect. After seeing Yik Yak with a $400 million valuation, we were convinced that investors wanted to see location-based social media. We had started our project years before this wave, so we were perfectly positioned to 

How we rethought social. 

Bubble was a feature-rich platform and we pulled out every stop to ensure that we were innovating on every front. Some of the innovations we were most proud of. Some of the innovations that we attempted weren’t feasible at the scale that we were at, we needed big data to fill a lot of the gaps in data-hungry algorithms that we attempted. Below are those innovations recapped.

Location-based Data

Location-gathering apps were still relatively new at the point at which Bubble started to build location-based features. We used semi-regular pings of the user’s location while they moved throughout their day. While the user was on the app, we essentially streamed the user’s location to the server. We also, to keep things safe, we regularly de-bound data relating to the a specific user’s location over time. We tried to anonymize this information as much as possible to avoid any major privacy issues, even if a data breach occured.

Popularity & Ranking

Popularity & Ranking was one of the most crucial features that made Bubble an exciting platform to use. Our goal was to re-create the same system of high-school popularity, in software. Basically, our idea was that if people were going to use Bubble, they needed incentive to do so. People were only going to be engaged if they were connecting with people that would help them climb the social status ladder. At the same time, we wanted for users that were less “popular” to climb the ladder more quickly than, say, the top 5% of users.

For techies

Popularity & Ranking was a cornerstone of the application, we used this rank to key people into the social experiences that best fit them. Additionally, we were able to gauge the popularity of an event by summing the popularity ranks of each event and then comparing those event popularities with other events nearby. The lower the sum with a greater number of people, the higher the “social impact” of the event was.

The popularity system was a particularly tricky problem to solve because we didn’t want to have to constantly update the rank of each user, this would have been nearly impossible in terms of server-time usage/cost so we knew there needed to be a creative solution applied to this problem that made the rank system stateless. We looked for solutions to this problem for a while but realized that we needed to mould a custom solution for this.

A mathematical approach was ideal for the solution, using the Sigmoid function to model how users grew & shrunk in popularity. The more popular a user got, the more difficult it got to be more popular. Every time that a user made a transaction against the database, we quickly recalculate the popularity of the user and store this to the database. This model made sense because those that were most active on the platform likely had the highest and most accurate ranking.

Events & Clustering

Events & Clustering were a big part of how users interacted with the application. These aspects were also closely related to Popularity & Ranking. Clustering solves when there are too many data points on a map in a small amount of area. When you cluster, you join events that are closest to each other. It became quickly apparent that we were going to need to display lots of data points on a map, we had to create a way of displaying these datapoints concisely. We decided that Events needed to be a main focus of the application, after all, location-based social media should be centered around where you are and where you are going. These events needed to be created automatically when a large number of people all came to a certain place at a certain time.

For techies

We implemented a server-side method to grab map data points for a geobox that was passed as a parameter to the server. We returned a set of the top 10 data points that we then displayed on the map.

Celebrities

In Bubble Social, we knew that celebrities would need to be given star treatment. For that reason, our plans were to give celebrities verified accounts. These verified accounts would be able to be activated at the time of an event to create a Bubble in their area. This would spawn an event on Bubble’s servers automatically.