I got some support on the Twitter Google group and there is in fact a way to get a list of friends and followers for an arbitrary user:
http://twitter.com/followers/ids.xml?screen_name=LowLimitPoker
which will generate a list of numeric ids for all the followers of the LowLimitPoker account (go ahead try it, you know you want to). This would work out great for my purposes, and in fact since I had my original idea to scan through the friends lists of accounts to identify probable high quality accounts I’ve had some additional ideas to improve the quality of the results.
So this is all good news. Well it would be except for the Twitter Rate Limiting which restricts the number of times you can call most API functions which return data (as opposed to those which set values) to 150 requests PER HOUR. Ok you might say wow that’s a lot of requests but consider this. If you use the API function above to return a list of friends, and lets say to make things simple that LowLimitPoker has 150 friends. Then you’ve used up one API call out of your 150 allotment, however if you’d like to convert those 150 numeric IDs to the actual screen names of the friends then you will have to call this API function
http://twitter.com/users/show.xml?id=3754891
150 times (once for each numeric ID) and oops, you’ve just gone over the rate limit and you haven’t really done anything useful yet. Likewise for the application I’m interested in to traverse even one level of the friends tree could use up far more that 150 API requests.
One possibility I’m avoiding but may have to look at more seriously is just parsing the regular Twitter web pages to get lists of friends and/or followers and other information. It will be sad if that turns out to be more efficient (or plausible) than using the API…
