((m-1)^2)+m-1)/2 GT 2m,
so:
m GT 5

Requires hybrid code your end though, which complicates matters?

But as a heuristic I think it works?!

SO if m people are interconnected, and they aren’t followd by so many people that you need to do more than eg 2*M API calls to get the lists, you have a saving if:

2m 5?

For small value networks (like – say – the network of Porter Novelli twitter users I did back in January) this is quite feasible. But given my interest in “influence” networks, my research tends to gravitate towards people with lots of followers, which makes the original method really wasteful for most exercises.

As it happens, writing this post cleared my mind, and I’ve now fixed the script in accordance w/ the ‘Where next?’ notes.

I’d never seen either 4store or NetworkX before – thanks! Fascinating. Must have a play with them…

I’ve already got tools to dig into people’s publicly-listed friend networks. As you know, when you send Twitter an API request to do this:

a) it must be authenticated
b) you receive up to 100 results per call
c) authenticated API calls are restricted to 100 per hour.

Let’s say that I have in my list 69 people (#UK MPs currently on Twitter) who mostly have over 900 friends (and some of whom have over 9,000) — I can use up my authenticated calls really very quickly.

It’s great getting all that data. But when all I’m looking for is directed relationships between that set of 69, it makes more sense to use the friendships/show method.

Once I have the matrix, it’s pretty easy to drop it into SNA tools like UCINet and do fun network analysis stuff. TweepDiff — with its A/B comparisons — isn’t up to games like this.

“- make a list of the people who I’m interested in researching [N people]
-for each person on that list, grab the list of all the Twitter people whom they follow”

How about you make those N calls, but rather than process the relationships yourself the hard way, just turn all the X is a friend of Y relationships into triples, and upload them to something like 4store [http://4store.org/] or a graph app like NetworkX [http://networkx.lanl.gov/ ] and then run your queries on those graph native representations?

(I have no idea if this will work – it’s something that’s on my to do list as a ‘maybe try this?’ hunch)

Glad you like DoesFollow. Have you checked out http://tweepdiff.com ? It allows you to easily compare followers versus from friends for two accounts. It also has a screen that allows you to do multiple at one time. It was created by Brian Deterling (a software developer here in Austin). It may be exactly what you need.