1. Central hub

jimsciuttoABC wrote: #iranelection Mousavi asking clerics to issue fatwas against Ahmadinejad as president

A number of the users who retweeted this message (NYTimesKristof, HegedusEricC, calindrome) were retweeted themselves by their followers. In this structure, it is evident that there is one major hub who is also the source of the communication. Nodes that appear with no connection to any other nodes on the screen are similar enough to be part of the same thread, yet either did not include an attribution to another user or did not use an RT format that my script could detect.

2. Multiple hubs

The Huffington post was so shocked by the news that the EU had accepted Ahmadinejad's victory so fast that three identical posts were made from Huff-po's three accounts: huffingtonpost, thehuffpost and huffpolitics.

Melitaz (top right node) writes: "@huffpolitics I agree - surprising to see that the EU has already accepted Ahmadinejad's victory". In this case, the message is an immediate reaction to the original conversation starter, and the user Melitaz responded by using the @reply convention (and not RT'ing), even though the majority of the texts match.

3. Massive Dispersal

In this example users were passing around information about the functioning proxy IP addresses that could be used from within Iran. mandelbrot5 was the very first one my script caught sending out these proxy numbers (June 14th, 19:03). In the early morning of June 15th, a variety of users (ivan007, yishym, doctorow) retweeted a very similar message describing the proxy addresses, finally reaching stephenfry who posted the proxy addresses the following afternoon. The line of nodes along the outskirts represent either tweets that have no RT attribution (but clearly belong to this thread) or use syntax that my algorithm cannot understand.

4. Non Attribution (people not attributing the story to its source - mostly on purpose)

Starting mid-day June 16th, the following thread appeared: "RT from Iran: Email and call #CNNFAIL Stop Posting Twitter Names on TV! Coverage is good, treachery isn't!" From the nature of this message, it is evident that the users who care to pass this message onwards will be very careful about making an attribution pointing to the source. As suspected, the network structure here is spread out, as users hardly point to each other.

By June 19th, Twitter users were already trained not to mention the source of their message, and instead, labeled their source as "RT from iran:". The thread above displays post around the organization of a memory march. In this example, it is impossible to make out any central source or authority from which this message originated.