[Photo credit: Cristina Castro]
Now that we’ve gotten Gander’s baseline functionality running nicely, we’re tackling prioritization. To cop one of my grandfather’s favorite phrases, it’s a real bear of a problem, because there are many methods of assigning importance, and none of them is right for everyone. As such, our goal is to figure out which method is the most right for the most people. Is it dyadic reciprocity? Subject-based categorization? Sentiment-based categorization? Contextual (temporal and geographic) prioritization? Today's post dives* into the first.
Robert Hanneman and Mark Riddle explain dyadic reciprocity in full detail in Introduction to Social Networks, but the gist is that, given two people, Gillian and Lutz, there are four possible relationships between them:
- Gillian and Lutz do not know one another (no tie).
- Gillian likes Lutz, but Lutz does not like Gillian (one-way tie).
- Gillian does not like Lutz, but Lutz likes Gillian (one-way tie).
- Gillian and Lutz like each other (reciprocated tie).
Substitute “emails” for “likes,” and you have a very basic framework for ranking email, with emails between reciprocated ties ranking highest and emails between strangers ranking lowest.
However, if your inbox is anything like Gillian’s, this framework would still produce a large number of chronologically sorted emails, as she has 227 reciprocated ties associated with her work address.
To make it more effective, you could look at the rate at which Gillian replies to each of her reciprocated ties, and how far each deviate from her average and overall correspondence frequencies. If you want to get even more precise, you could calculate the momentum of a given correspondence. Here, you’d be wise to tie in temporal data as well, as Gillian is positively slammed at the end of each quarter, but tends to have plenty of free time in August.
We’re testing variations of this framework out on our own inboxes now. If it ends up the winner, I’ll be sure to let you know!
*A one-knee dive, as this is all fairly new to me.