The discussion on bias automatically brings the question of what is the truth? To argue that a newspaper favors a particular topic or person (i.e. It is biased), one has to know what is the {\it expected} number of times that a topic or person should be mentioned.

Although politics are a multi-facet subject and politicians many times become popular on the media due to characteristics other than their performance as a politician, we decided to evaluate our proposed method entirely from the angle of ``collaborativeness''. We define the ``collaborativeness'' of a politician (in our case a senator) as the number of times this politician has collaborated with other senators (absolute collaborativeness) and the rate of each of those collaborations (weighted collaborativeness). The true level of collaboration of a senator is given by the concept of co-sponsorships. In the US Senate, bills can be introduced by any of the senators and others generally demonstrate their support by co-signing the introduced bill. Furthermore, according to Campbell \cite{campbell1982cosponsoring}, co-sponsorship are actively sought by the bill proponent because they can use the number of co-sponsorships on speeches as an indication of broad support for the bill's idea. Given this market for co-sponsorships, one could be inclined to say that they take place quite often, but Fowler \cite{fowler2006connecting} has observed that the average legislator cosponsors only 2-3\% of the bills---meaning they are quite selective.

The data related to the co-sponsorship from the 103$^{\text{th}}$ US Senate was collected by the New York Times Congress API\footnote{http://developer.nytimes.com/docs/read/congress\_api}. Both senator's name and number of co-sponsorship in the bill were available there. We collected data for 104 senators which is 4 over the normal 100. The additional names are: Jeffrey Chiesa (1), Republican, in replacement to Frank Lautenberg (died), then replaced by Democrat Cory Booker (2); William Cowan (3), Democrat, interim of John Kerry (Secretary of State), then replaced by Democrat Edward Markey (4).

One of the challenges when it comes to doing analysis of newspapers bias relates to the data collected. Most newspapers are not keen on giving their data for analysis and request us not to use Web crawlers to collect data either. Therefore, we decided to abandon our initial thought of using Web crawlers in lieu of using information from Google search queries because Google has already done the parsing of the pages on each site and can provide us with just the count we need.

Media bias in politics has always been a topic of interest because it is a subject everyone feels a little passionate about. Most people who pay attention to politics have an opinion about media outlets and how they possibly sway public opinion. Interestingly, no matter who you select, they believe the media always favors the other candidate, party, ideology, etc. In the USA, liberals say the media is conservative \cite{Dennis:Liberal:1996} and conservatives say the media is liberal \cite{Bozell:How:1992}. Who is right?

In 2000, D'Alessio and Allen \cite{DAlessio:2000te} set out to answer the question above. They investigated 3 kinds of bias namely: {\it gatekeeping bias}, the preference for selecting stories about one party or the other; {\it coverage bias}, or the amount of coverage each party gets; and {\it statement bias} which deals with the favorability of coverage towards one party. The authors looked at bias in newspapers, magazines and television and of these only the television had a slight coverage bias and statement bias.

Druckman and Parkin \cite{druckman2005impact} showed that the effect of a newspaper tone towards a candidate can impact the voters' decision and they found some good evidence that the slanted editorials indeed can achieve this goal. They also raised the question of whether readers are to blame or the victims for not looking for diversity in their media activities (reading, watching TV, etc.)

Groseclose and Mulyo \cite{Groseclose:2005wg} have proposed an interesting measure of media bias and applied it to print and television media outlets. Their metric is based on a count of mentions to specific think tanks. Think tanks themselves are assigned a score from conservative to liberal based on the mentions of their names by members of congress. The starting point is the ADA metric (Americans for Democratic Actions) of members of congress, which is used to calculate an ADA-like for each think tank which will then be used to calculate another ADA-like score for the media outlets. The authors found that most media outlets are left-leaning (liberal), except for the Fox News' Special Report and the Washington Times.

Media bias may also be related to voting patterns. DellaVigna and Kaplan \cite{dellavigna2007fox} have done a study specifically with the Fox News Channel in more than 9,000 towns in the USA and found that their pro-right approach accounted for a significant gain of the republican party between the years 1996 and 2000. On average the party gained 0.4 to 0.7 percentage points in the towns where the Fox News Channel was available.

Nowadays we live in the decade of the big data. Scientists today have the ability to collect data about virtually anything and certainly politics is one area of interest. This availability should yield more reliable results on bias, be it on politics or any other issue worth studying.

The use of networks to model and study relationships between seismic events has been used in the past for small areas of the globe. Here we demonstrate that similar techniques could also be used at the global level. More importantly, many of the techniques used in complex networks analysis were used here to show that there seem to exist long-distance relations between seismic events.

First we argued in favor of the long-distance relation hypothesis by showing that the network has small-world characteristics. Given the small-world characteristics of high clustering and low average path length, we were able to argue that seisms around the world appear not to be independent of each other. To strengthen this argument, we decided to do a temporal analysis of our network. Plotting the probability distribution for the time intervals between successive earthquakes, we have found that this distribution is well fitted by a $q$-exponential, indicating a behavior described by the non-extensive statistical mechanics, which obtain $q$-exponential distributions from the generalized Tsallis entropy. This non-extensive behavior also contributes to the long-distance relation hypothesis, since the non-extensive statistical has been used to explain many complex systems with long-range interactions and long-range temporal memory.

\green{Another interesting approach we intend to do in the future relates to using community analysis or community detection to understand how seismic locations are grouped. We believe that given the long-range relations that we found here it is unlikely that the globe would be well organized around local communities of nodes (geographical locations).}