In writing this post, I wanted to give a little background on data science, introduce a little perspective on what shoto does from the back-end data side of things, and explain some of the challenges it helps us overcome. I look after all data science efforts at shoto, and gathering insight from shoto’s data is an integral step to delivering our users an enjoyable and fun visual experience that’s different from what’s currently out there.
For all of recent memory we’ve seen big data for big social media: Twitter, Facebook, Instagram, etc. For those of us in the data science field, we all remember this famous graph visual of Twitter soon after Osama Bin Laden’s death
A rumor going viral about Osama Bin Laden’s death (http://www.visualizing.org/stories/visualizing-bin-laden)
I find this visualization interesting because it shows how quickly a tweet can spread out endogenously within a network given the right circumstances. It is directed and spreads out as we would expect a rumor should spread. Twitter is great for this directed and very public form of communication. We can observe power users with a significant number of followers (Keith Urbahn has over 1,000) and a high degree of one-way activity from each node as it trickles from power users to more common (lower degree) users. This is what Twitter is good for: blasting out some news quickly down a series of one-way streets. Now what if we could see how everyday people interact with their real, everyday, personal contacts as opposed to their social network connections? We would probably see something like this:
Subset of a shoto user’s network
Each circle represents a node/user while the lines represent their connection to another user, i.e. sharing photos together. In this case we see shoto users sharing with a small number of people, usually less than seven, and in most cases it’s two or three (we’d have to zoom in to see the detail because there are a lot of binary and tertiary relationships that are not visible from this distance as each line is weighted by activity). And unlike the Twitter graph visual, this is not directed, meaning the edges/lines represent back and forth sharing.
This all makes sense in the real world. Highly gregarious people may socialize regularly with 20 or more people, while a more introverted person may regularly socialize with significantly less. And couples are likely to look very binary, even though they may be sharing with lots of people, because the weight of the connection between the couple will drown out the others.
While Twitter and Facebook will continue to serve their purpose in the future, we are seeing this large trend among younger demographics to move away from a Facebook/Twitter megablast platform toward more private networking and sharing services. This is not surprising. After all, in the real world, how many of us actually share our everyday photos or special events to our extended friend network? For most people most of the time, you probably only physically interact with about 50 or less of your Facebook friends. Of those, you probably are really close and in very frequent physical contact with less than 10.
More generally, apps are looking to be more in line with real people in the real world. Instead of expecting humans to follow an app’s behavior, our app should mirror and enhance human behavior, as it is, a priori. Current photo sharing apps all follow the Facebook/Instagram paradigm.
And herein lies the premise behind shoto and its private photo sharing technology: shoto is for real friends to share photos. And when I say this I mean any two or more people who do things together in the same physical locality…doing something where they can actually touch each other, share food across a table, give each other a hug or a kiss, or anything else that people do when they are together. Unlike other photo sharing apps, shoto is not just for those who sit at a computer all day or night.
shoto shares photos automatically, with albums of these photos syndicated based on location, time, and other context. When you physically share the same space and time as one of your friends on shoto, you get to see each others photos automatically. You can literally watch their photos populate your phone in real time when the time is right for such sharing. You may see that silly pose or face you made perhaps unexpectedly, and your friends can see your photos of them. All in real time and in the same privately shared photo album.
This, however, is the challenge: looking at people’s behavior in a location and deciding when they are together and when they are apart. Without giving away too much secret sauce, here’s a look at what a series of photos, taken by myself standing in the same exact position in the real world, look like on a map:
These points represent photos taken at a location with wifi turned on
Not too bad. It’s mostly clustered at the same residence almost. Here is the same person having taken a similar series of photos at the same location without wifi:
These points represent photos taken at a location with wifi turned off
It’s not hard to see the challenge. Even with Wifi, there’s a big spread in the location data all from the same exact position. This is not an issue in most cases, especially large open areas. But it is especially important to make this distinction in dense urban areas where two neighbors that are separated, as far as the users are concerned, can easily overlap, or in a large apartment building where they can easily be on top of each other. Add in a road trip or a sail around the bay with friends and the challenge is even more noticeable.
shoto looks at these considerations and determines if two people are together or not. If they are, they get to see each others photos taken during that sharing moment. The key is to look at context. The contextual information necessary for making this happen falls naturally from peoples behavior in the real world. shoto is for sharing and by sharing we can see a lot of how people interact not just the when/where.
People are visual, the Web is visual, mobile app networking is increasingly visual. And shoto does and will always strive to capture the visuals of your everyday and your not-so-everyday memories with those who matter, beautifully.
Frank Taylor, Lead Context & Insight Hacker @shoto
(Haven’t read Frank’s introduction to the shoto team yet? Check it out here)