What is Snowball?
Snowball is best explained with an example:
Imagine an ice cream vendor, cycling to the centre of a park and setting up shop. To promote his service, the vendor activates a client application on his phone with a simple advertisement containing:
Passers by with similar clients on their phone receive this data when they come within Bluetooth range (around 10 meters).
As they move around, they share this information with others in a peer-to-peer fashion, forming a cloud of information well outside of the vendor’s original vicinity.
When a user wants ice cream, they open up an application on their phone and request a search. The application finds services from a database on the device and indexes them according to relevance, normally closest first.
Relative distance is determined from many traceable factors derived from the journey the ad had to take to reach a particular location. In our initial example we know:
- How long an ad’s journey took
- How many hops it needed
- How mobile the intermediaries were while carrying the ad
The user is presented with a list of ice cream vendor advertisements relevant to his or her local area. Using their judgment, they can select the most relevant advertisement and then explore the additional information included with the message (e.g. what flavours of ice cream is he selling?).
