Update 1/6: Wow, this made top of HackerNews! I'm honored. Hope you enjoy this post - let me know what you think in the comments. Besides this blog, I can also be followed on Twitter @hunterwalk
Michael Arrington wrote a compelling piece suggesting Facebook's continued success is dependent upon allowing users to wipe their friend slate clean and rebuild their graph. This second time around he suggests, we'd do a better job of curating the relationships to create a high signal, low noise set of friends. Mike even pinky swears that he'll stick to his guns. Unfortunately, and this is no statement about the credibility of Mike's pinky swears, I believe most social communication products - from Facebook to Twitter to Path are designed in a way which causes us to all slowly ruin the quality of our experience.
Basically we say we love high signal, low noise services but then proceed to bring the noise. When people say something like "Oh Path is great, it's just my close friends" it's because we start out any product by using it with just our close friends. When he invented the telephone, Alexander Graham Bell called Watson, not someone he barely met at the town bar. However for a set of reasons I'll outline below, we keep adding relationships and activity until it no longer resembles the initial value proposition.
So what compels us to ruin our own social communication tools? And what are the product design implications?
1) Vanity & Ego
Hunter's rule: Any communication service which publicly displays a metric serving as a proxy for popularity will cause users to take steps to increase that number. Number of friends on Facebook, number of connections on LinkedIn are two easy examples. First order derivatives are just as problematic: if you friend more people on Path (up to the 150 limit) you increase the number of people likely to see your posts which will increase the visible "viewed by" count and number of emoticons. On Twitter, you are tempted to follow back in order to solidify them as a follower. Or RT someone so that they'll RT you later. And these pressures are occurring just within the services themselves. If social influence metascores such as Klout gain prominence it will increase the problem as users seek to grow more popular to increase their score. Unfortunately publishing volume seems to be a compelling way to accomplish this as opposed to just quality
Product Solution: Hide those damn numbers!!! Let the account owner see them but not the general public. Imagine Twitter without public follower/following counts, Facebook without a friend count. I've argued before that Path's design choice here is counter to their stated goal of intimacy.
2) Boredom & Pleasure
You fire up Path/Instagram/Facebook/Twitter/whatever and, gasp, there aren't any new updates from friends. Hmm, that was unrewarding, so you decide to friend some more folks to decrease the chances of shooting blanks. It's equivalent to random reward cycles in game design - each time you log-in there might be a prize, and it keeps you coming back. In this case you can influence the rate of "prizes" by following more folks. The result becomes a noisy feed but feeds the addiction to "new" content.
Product Solution: Never leave the main feed empty of new items to review. Create an ephemeral experience for me when necessary. For example, instead of putting "popular" photos only behind a separate tab on Instagram, if I open the app and none of my friends have updated you might give me an element that shows popular photos I might enjoy (say four smaller photo tiles which allow me to select and click over to that person's photostream or to a trending tag). Or in G+, if there aren't any updates, popular posts or relevant clusters from Google News (they do a variation of this today but it's intersecting the middle of my feed which I don't like -- it's really hard to mix info types, my suggestion here is not to put at bottom of new feed items, but rather to only insert this element if there aren't any new items next time you log in - state-aware UX).
3) Reciprocation & Conflict Avoidance
C'mon, admit it, you've followed someone just because they followed you first. And you're afraid to unfriend/unfollow because you don't want to have THAT conversation with them. These issues are most prevalent in bidirectional systems such as Facebook but most services have this issue - I'm guilty a few times of DM'ing someone who then reminds me they can't DM me back because I DON'T FOLLOW THEM. Sheepish "oops." [Note, I know Mike probably has no problem with this one]
Product Solution: This is tough. Obviously in bidirectional systems, providing a unidirectional fallback is useful (such as "I won't friend you on Facebook but you're free to subscribe to me"). And environments like Twitter don't notify you when someone unfollows (although there are services which will). And G+'s Circles do a nice job of allowing me to pseudo-follow (ie you see that you are in one of my circles but it could be a circle that I never look at - just a holding pen for people I don't care about but don't want to offend).
The only other suggestion I have revolves around the service making it easier to decline by creating norms -- such as if Facebook had prefilled decline friendship selection which said "hey, I like you but I'm trying to keep my group here small. please feel free to subscribe." That would allow you to make more standard the denial of friendship and make it seem like something which happens every day plus give an option to stay connected.
4) I Like You But Not Everything About You
Social graph != interest graph. So I friend someone but eventually see noise from them - where noise is defined as "jeez, you really like to post dog photos." If it's a particular data source you can block it at the app level (no more castleville invites!) but when it's just generic media types, you can't block.
Product Solution: Another tough one because the act of trying to filter social graph items by some measure of your interest in them is (a) difficult technically and (b) confusing to users ("why didn't i see that"). Better is to design your feeds in a way which make the sharer more prominent than the item they are sharing - so it feels like you are learning something about your friend through the context of what they shared, even if you are not consuming the item they are sharing. The perceived value is in building stronger ties to your friend, not using them as a discovery mechanism for new content.
Obviously most of these services are also engineered for growth and expansion of the graph rather - "find and invite your friends" get built way before quality scoring of feed items. And API/platform integrations allowing folks to more easily publish into the feed, increasing the noise (game invites! new Oinks! I just weighed myself!). But for purposes of responding to Mike's piece, I wanted to focus on human factors, not inherent conflicts between ad-supported business' growth strategies and user goals.
"This work is licensed under a Creative Commons Attribution 3.0 Unported License."