2What is it?Everyone talks about the small world phenomenon, but truly what is it? There are three landmark papers:Stanley Milgram (1967)Duncan Watts & Steve Strogatz (1998)Jon Kleinberg (2001 )
3Milgram’s experimentA person P in Nebraska was given a letter to deliver to another person Q in Massachusetts. P was told about Q’s address and occupation, and instructed to send the letter to someone she knew on a first-name basis in order to transmit the letter to the destination as fast as possible.
4Milgram’s experimentOver many trials, the average number of intermediate steps in a successful chain was found to lie between 5 and 6.
5Milgram’s experimentInitial success rate was very low (5%). The follow-up experiments made some modifications of the original experiment.The outcome of the experiment led to the term:six degrees of separation
6Other example of small world Small world graphs are highly clustered like regular lattices, yet paths of short length exist between random peers. Example of such graphs arePower grid of western USCollaboration graph of movie actorsNeural network of worm C-elegans
7Watts and Strogatz 1998Research originally inspired by Watt’s efforts to understand the synchronization of cricket chirps, which show a high degree of coordination over long ranges, as though the insects are being guided by an invisible conductor.Disease spreads faster over a small-world network.
8Questions not answered Why six degrees of separation? Any scientific reason? What properties do these social graphs have? Are there other situations in which this model is applicable?Time to reverse engineer this.
9A characterization of graphs Completely regularSmall-world graphs (N >> k >> ln (N) >>1)Completely random
10Completely regular N=20 K= 4 (each node has k neighbors) High clustering coefficientand high diameter.C = 3/6 = 1/2, L ~ N/kThe clustering coefficient for a vertexis given by the proportion of links betweenthe vertices within its neighborhooddivided by the number of links that couldpossibly exist between them.A ring lattice
11Completely random LOW clustering coefficient and LOW diameter. C ~ k/n, L ~ log N
12Small world graphs With probability p rewire each link in a regular graph to a randomly selected node
13Small world graphs Such a rewiring results in a a graph that has a a high clustering coefficient but low diameter …
14Small world graphs L= typical distance between two nodes C= Clustering coefficientN >> k >> ln(N) >> 1CGuarantees that the graphIs connectedL10.01p
15Kleinberg’s questionWatts and Strogatz’s research only showed the existence of short paths between arbitrary pair of nodes. What is the guarantee that one can find such a path for communication?(If there is no algorithm for finding the short path, then it is not of much value!)
16Kleinberg’s Small-World Model Embed the graph into a grid. Each node has links to every node at lattice distance p(short range neighbors) & q long range links. Choose long-range links s.t. theprob. to have a long range contact is proportional to 1/dr (r is a new parameter)p = 1, q = 2r = 2
17ResultsTheorem 1.When r = 0, no decentralized algorithm can find the short chains (even if they exist).
18ResultsTheorem 2.When r=0, the expected no of hops needed to connect with a peer is O(n2/3).[Uniform distribution prevents a decentralized algorithm from using any clue from the geometry of the grid].