Signal to noise in social communications
May 21st, 2008 by Benjamin KuoI’m showing my technical background here, but I was recently browsing through my ever growing backlog of “friend updates” from Facebook, LinkedIn, Twitter, etc. and started thinking about the poor signal to noise ratio in social communications.
The signal to noise ratio–in electrical engineering–is a measure of how much useful signal you are actually getting from a communications channel, versus the background noise in the signal. For those poor souls who grew up in the world of analog, the noise is the snow on a bad TV set, and the buzzing you hear on a remote radio station. In the world of social communications, signal is the useful information you get from that information, and noise is the extra, useless stuff you just really don’t need to be filtering through.
Facebook, Twitter, and other social sites are great, but they’ve got a critical problem, which is the amount of noise is almost staggering; without some kind of filter, it’s very, very difficult to pull out the most critical information you need to know from the constant data stream. In figuring out signal-to-noise, engineers use the amount of power in the actual signal divided by the amount of power in the noise to determine how much useful signal you have. You could almost determine a numeric ratio by using the number of messages and updates in the feeds which are are non-useful information to actual informative information.
In a social sense, signal to me might be:
- Insightful pointers to interesting news or commentary
- Information about upcoming events or happenings
- Breaking news
- People joining or leaving a company
- Questions to the community, or trading of useful information
- Anything actionable - that you can act on in some way
Noise, on the other hand, would be:
- Details on lunch/dinner/breakfast or some other random detail of daily life
- System messages about who you have added as a friend (LinkedIn is particularly bad about this: once you get past a few hundred contacts this becomes an un-usable flood).
- Random stream-of-consciousness posts
- Auto-generated, GPS-powered “I’m here” information
- Micro traffic updates (everyone in LA on Twitter seems to be stuck somewhere in the city in traffic)
- Updates on all the contacts I have who have changed their profile picture
- etc.
In order to make these services more useful, there needs to be more signal than noise — and tools to filter out those signals from the massive data stream they’ve created. Those tools need to be built into the services (there’s plenty of random hacks for doing some kind of text-based filtering for Facebook feeds, twitter updates, etc.), and they need to be automatic, easy to use, and tunable. Social communications has lots of promise, but without some help improving the signal quality, it’s very easy to get completely inundated by the noise.
