Exploring the dynamics of protest against National Register of Citizens & Citizenship Amendment Act through online social media: the Indian experience
Souvik Roy, Milan Mukherjee, Priyadarsini Sinha, Sukanta Das, Subhasis Bandopadhyay, Abhik Mukherjee
aa r X i v : . [ c s . C Y ] F e b Exploring the dynamics of protest against NationalRegister of Citizens & Citizenship Amendment Actthrough online social media: the Indian experience
Souvik Roy, Milan Mukherjee, Priyadarsini Sinha, Sukanta Das, SubhasisBandopadhyay, Abhik Mukherjee
Indian Institute of Engineering Science and Technology, ShibpurHowrah, West Bengal, India – 711103
Abstract
The generic fluidity observed in the nature of political protest movementsacross the world during the last decade weigh heavily with the presence ofsocial media. As such, there is a possibility to study the contemporary move-ments with an interdisciplinary approach combining computational analyticswith social science perspectives. The present study has put efforts to un-derstand such dynamics in the context of the ongoing nationwide movementin India opposing the NRC-CAA enactment. The transformative nature ofindividual discontent into collective mobilization, especially with a reflectiveintervention in social media across a sensitive region of the nation state, ispresented here with a combination of qualitative (fieldwork) and quantita-tive (computing) techniques. The study is augmented further by the primarydata generation coupled with real-time application of analytical approaches.
1. Introduction
During the last decade, social media has provided instrumental meansof communication for moulding the nature of political protests in the recentpast[1, 2, 3], such as, mass demonstrations in Moldova [4, 5, 6, 7], Turk-ish protest movement in June, 2013 [8], Egyptian revolution of 2011 [9, 5],Ukrainian protests of 2014 [10], Occupy Wall Street protests [11], mass mobi-lization in Spain in May 2011 [2], protest in Hong Kong in 2014-15 [12] etc..John T. Jost et. al. [1] have summarized evidence from a variety of studiesof protest movements in the United States, Spain, Turkey and Ukraine and
Preprint submitted to – February 23, 2021 resented the following findings about the usage of social media to facilitatepolitical protest - (a) News about the coordination of protest activities (suchas transportation, turnout, police presence, violence etc.) is spread quicklyand seamlessly through social media channels; (b) Social media transmitsemotional and motivational messages about protest (i.e. normativity, so-cial justice and deprivation); and (c) The structure of on-line social networkalso plays an important role in the success and failure of protest movements.However, the research on political participation has also long emphasizedthe debate of ‘raise of political awareness’ and ‘feel good politics with hollowconsequences’ to understand the meaningful and void effect of social mediaon political protest [1, 2, 3, 13].Computer scientist have mainly examined the diffusion process of protestsusing the recruitment patterns in the social media (i.e. Facebook, Twitteretc.) network in the context of mass mobilization [1, 2, 3, 14, 12, 15, 13, 16].According to [3], higher density of ties are present at the core of the net-work, where participants are on average more active in posting messages andretweets. The information has flown largely from the core to the peripherywhere users are significantly less active on a per capita basis but who con-tribute as many messages at the aggregate level. The most interesting factis that removing the lowest five periphery zones results in a dip of slightlymore than 50% in reach. According to [2], the vast majority of contagionchains die soon, with only a very small fraction reaching global dimensions.This result is also supported by other findings [17, 18, 19].However, long before the presence of organizing tool internet, sociologists[20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33] have also discussed the roleof critical mass and their importance in resource mobilization to understandthe success of social movements. This includes historical empirical results,such as, revolt in Paris commune (1871) [34], 1960’s civil right struggles inthe US [35], demonstrations in East Germany prior to the fall of the Berlinwall [36, 37] etc.In a different research effort, computer scientists have also analysed theinformational and motivational content of social media posts during the massmobilization period of political protest [8, 10, 11]. Again, in this context,social scientists have widely discussed about social psychological factors –moral outrage, social identification and group efficacy [38, 39, 40, 41, 42, 43,44, 45, 46, 47, 48, 49, 50, 51] . However, there is a lack of understanding aboutpolitical protest by aggregating both the social media and field experiencetogether which is a limitation identified in the reviewed literature on social2edia analysis.In this backdrop, this article studies the dynamics of protest movementagainst the policy of countrywide National Register of Citizens (NRC) &Citizenship Amendment Act (CAA) in India by analyzing social media aswell as data collected from field work. This article covers the first six monthssince the Central Govt that came to power in June 2019 announced thatthe National Registry of Citizens (NRC) will be prepared in lines with asimilar exercise conducted in Assam [52], a state located in the North Easternzone of India. The exercise in Assam culminated in declaration of a sizablepopulation as de-voters and forcible detention of people in camps pendingforeigner tribunal cases [53]. Hence many peace-loving people across thecountry got scared when the newly elected Govt announced this NRC astheir pan-India policy. In fact, the Govt subsequently passed a legislation toamend the existing Citizenship Act (CAA) along communal lines [54], widelyseen as a precursor to the nationwide implementation of NRC. The grippingfear, with clear evidence from Assam as indication of what lies ahead, resultedin widespread resent that burst into protest throughout the country. Theprotests had an interesting feature of spontaneity and is fast becoming thenew movement cult or norms, with sizeable presence in the social mediafronts. Two modes of social media interactions have been considered, thedeliberations within a group formed with a purpose and individuals tryingto reach out to their followers with some purpose. Based on availability,Facebook and Twitter are respectively chosen for such purpose.In this work, different social media (such as, Facebook and Twitter) ac-tivities are monitored over the first six months of the Contra-NRC & CAAmovement. Group activities and related dynamics is explored from con-tra NRC & CAA Facebook groups that evolved during this time period.The proliferation of such groups, growth of their membership and sharing oftheir categorized contents among people with different social identities havebeen considered. Spatial, temporal and informational dynamics of individ-ual responses to the contra NRC & CAA topics as reflected through Twitteraccounts have also been scrutinized to understand the movement.Somewhat extensive field work has also been conducted in the borderingdistricts of West Bengal and Assam. These districts share border with in-dependent sovereign of Bangladesh and are considered to be the hotspot ofhistorical migration since the partition of Bengal along communal lines. Thisstudy attempts to close the gap between the ground reality and the findings ofsocial media activity through such field work. The work investigates to what3xtent the contra NRC & CAA movement has acquired the characteristicsof mass movement and the associated representation of societal dynamics.In detail, this paper is organized as follows: • In Section 2, the dynamics of the creation of the contra NRC & CAA123 Facebook pages and 79 Facebook groups are depicted. • The growth in membership of the newly created 11 contra NRC & CAAFacebook groups containing total 1 , ,
463 members are described inSection 3. • To understand the structure of the Facebook groups from the per-spective of mass movement, the number of common members in WestBengal based 37 contra NRC & CAA Facebook groups are analyzed inSection 4. • In Section 5, the contagion chain dynamics and religious participationwithin it is depicted for contra NRC & CAA 14 viral Facebook postscontaining total 11 ,
642 Facebook profile users. • The informational and motivational content of 3200 Facebook postsduring the mobilization period of NRC & CAA protest and 500 Face-book posts during spread of COVID-19 period are analyzed in Sec-tion 6. • In Section 7, the dynamics of growth in number of tweets, location andstructure of twitter network are analyzed for 4 , ,
978 tweets duringthe mobilization period of contra NRC & CAA protest. • The topic wise responses and expressiveness of the responders in thefield interviews are analyzed in Section 8. • Finally, Section 9 concludes the paper with critical comparison of dif-ferent signature behaviour between social media and ground reality.
2. Creation of groups and pages in social media
This section studies the growth of Facebook pages and Facebook groupsin numbers during mobilization period of contra NRC & CAA protest. Todo so, we have collected dataset from random but purposive 123 Facebook4ages and 79 Facebook groups during the period of mid-September, 2019 tomid-January, 2020. To facilitate this study, we have used following notationsin the analysis. • N ( k ) indicates total number of Facebook pages/groups present at in-stant k . Total time of the study is divided into slots of 3-days interval. • N ( n ) depicts the total number of pages/groups at the end of our timewindow, i.e. January 15, 2020. Here, k runs from 1 to n . Therefore, N ( n ) = 123 (resp. 79) for Facebook pages (resp. Facebook groups).When we consider both pages and groups together, then N ( n ) = 202. Let us now plot the data, to understand the accelerating rate of growthof Facebook pages and Facebook groups in number during the mobilizationperiod. Fig. 1(a) depicts the growth of N ( k )/ N ( n ), which represent thenormalized number of activated pages/groups, over time. Next we plot thenormalized rate of increase in actual number of pages/groups created duringeach time slot k . That is, ( N ( k + 1) - N ( k ))/ N ( n ) is to be plotted against k . Fig. 1(b) shows such graphs. Now, the above results depict the followingdynamics of the page / group creation. N ( k ) / N ( n ) GroupPageGroup & Page ( N ( k + ) - N ( k )) / N ( n ) GroupPageGroup & Page (a) (b)
Figure 1: The growth of (a) N ( k )/ N ( n ) over time; (b) ( N ( k + 1) - N ( k ))/ N ( n ) against k .
1. There are sharp rises in the graphs of Fig. 1 after k = 20, which rep-resents the rise of page/group creation after December 12, 2019. The5itizenship Amendment Bill was passed in Parliament of India on 12December, 2019. Further, the new group creation counter reaches itspeak during the first half of January, 2020 (see Fig. 1(b)). This is an in-dication that the people are increasingly joining the protest movementafter December 12, 2019.2. It is observed that before December 12, 2019, Facebook groups aremostly West Bengal and Assam based. That is, people of these stateswere more concerned about the implication of NRC & CAA. In As-sam, the procedure of NRC dragged on for six years and finally a listof citizens was published in August, 2019 [53] excluding more that 1 . impact of Assam NRC (14 . migrationafter and related problems (13 . , ,
179 likes dur-ing the entire period of time. Moreover, the 79 groups contain total4 , ,
294 members with highest membership of 1 , ,
547 in a singlegroup. The total number of posts per day in the 79 Facebook groupsis 44 , , ,
951 (7000/day) times over a three month period. Similarly,the most popular Facebook page on Ukrainan protests of 2014 was liked6ore than 1 , ,
000 times during two weeks period of time [10].
3. Membership proliferation in the groups
This section studies the growth in membership of the newly created Face-book groups during mobilization period of CAA & NRC protest. To do so,we have chosen 11 groups out of 79 groups of Section 2. Choice of such 11group is diverse: based on number of members, creation date (before/afterDecember 12, 2019) of the group, participation of parliamentary politicalparty in the group and location (state in India) of the group. Note that,although the data of original 79 group had been collected for the periodfrom mid-September, 2019 to mid-January, 2020, now, we have slightly mod-ified the period due to unavailability of membership data for the month ofSeptember. The modified period is October 11, 2019 to January 21, 2020.These 11 groups contain total 1 , ,
463 members with highest (resp. low-est) number of member in the group being 1 , ,
547 (resp. 1 , ,
000 (resp. 100). Here, fivegroups out of 11 are West Bengal based and are created in the month ofSeptember, October and November. However, the rests (outside of WestBengal) are created after 12 December, 2019 which was the passing date ofCAA in Parliament of India.This Section shares similar notations with Section 2. Here, M ( k ) indicatesnumber of members in the corresponding Facebook group at instant k . Totaltime of the study is divided into slots of 2-days interval. Here, M ( n ) depictsthe total number of member in the corresponding Facebook group at the endof our time window, i.e. January 21, 2020. To understand the dynamics of member joining rate in groups, we haveplotted number of member in groups against time during the period of Octo-ber 11, 2019 to January 21, 2020. Fig. 2(a) depicts the growth of M ( k )/ M ( n ),which represents the normalized number of joined members over time. InFig. 2(b), we plot ( M ( k + 1)- M ( k ))/ M ( n ) against k . Now, the above re-sults depict the following dynamics of the number of members joining in thegroups. • The plots show following two sharp rises in the member joining in thegroups: 7 M ( k ) / M ( n ) G1G2G3G4G5G6G7G8G9G10G11 ( M ( k + ) - M ( k )) / M ( n ) G1G2G3G4G5G6G7G8G9G10G11 (a) (b)
Figure 2: The growth of (a) M ( k )/ M ( n ) over time; and (b) ( M ( k + 1)- M ( k ))/ M ( n )against k . – Firstly, for the groups located at West Bengal, a sharp rise isobserved within k = 0 to k = 5, which represents starting ofOctober, 2019. Note that, this corresponds to the time whenHome Minister of India Government had announced during a visitto kolkata that the procedure of NRC would take place in WestBengal too [52]. This reason is definitely responsible for the risein membership and this observation is also reflected from the fielddata, for example, see part of field interviews . In this context,we noticed that 11 .
34% of those interviewed responded citing theproblem of confusing statements by political leadership . – A second sharp rise in new membership for all the groups has been “After the speech of Amit Shah on 1 st October, 2019 in Kolkata the scenario haschanged drastically. There was no space to walk around the court area; mothers with their4-6 months child, elderly people were seen in the court for the process. There has beenlong queue for about 15 days in the court.” : GF-RM-EM-AU-DNM-L4-O2-A2. “There used to be a queue of about thousands and thousands (4000-5000) of peopleoutside the Block office. As there has been a deadline there was a huge crowd. A personhas also lost his life due to stampede in the crowd. Everything started after the speech ofAmit Shah in Kolkata on 1st October, 2019 and then Dilip Ghosh said that the Muslimswould be thrown out ruthlessly. And now some other political parties in their conductedmeetings and rallies said that they would not allow NRC to get implemented so that thefear among the people has decreased.” : GM-RM-EM-AR-DNM-L3-O2-A3. k = 20, which represents December 12, 2019, thedate of passage of CAA in the parliament.The most popular Facebook group, based in West Bengal, has touched1 , ,
547 members within only six month after its creation (6 September,2019), which is remarkable. Moreover, Contra-NRC/CAA/ NPR (NationalPopulation Register) physical protest rallies, meetings, street meeting, cyclerallies were organized by these Facebook groups. Moreover, new movementalorganization was created through these Facebook groups by common peo-ple without presence of traditional political leadership. According to oneof the ‘admins’ (administrators) of this Facebook group, interviewed on 22February, 2020, in Kolkata,“
Around six months ago we created a Facebook messenger groupagainst the NRC, CAA and NPR and then a Facebook groupthrough it. With the creation of the group the member countstarted to increase at an exponential rate. We had completelyno idea about how the member count could increase in such amanner. The members were continuously pressing for an Anti-NRC, CAA, NPR movement programme on the road so that thelocal people could be involved. To understand the process of howto involve local people in the movement, we arranged a conven-tion meeting including the members. At the meeting we reacheda decision that on 19th December we will be organising an Anti-NRC, CAA, and NPR rally in Kolkata, announcing it throughthe Facebook group.On 19th December, 2019 in Kolkata an anti-NRC, CAA, NPRrally was organised on behalf of “NO NRC Movement” Face-book group, according to the police report there was estimationof around 60-70 thousand of people.People have built up a social media centric organisation, whichstarted initially through a chat of 2-3 people. And now the peoplethemselves are organising their committee, arranging programmesagainst NRC.”
4. Structure of the groups
In this section, we concentrate on the West Bengal (specifically Kolkata)based Facebook groups to understand the administrative hierarchy within9he groups that may be conducive to understand the dynamics of a fledglingmass movement. The reason behind this choice for the study is that WestBengal was one of the most affected states during the Partition of India in1947 [55, 56, 57]. Therefore, the ingredients of mass movement against NRC& CAA is already present in West Bengal. Altogether 37 Facebook groupshave been identified, out of the 79 groups discussed earlier in Section 2,those are demographically located within Kolkata and its suburbs. Datasetof unique Facebook member identification has been collected for these 37Facebook groups till January 21, 2020. Total member count of these 37groups is 46 , In this section, number of common members for these 37 Facebook groupsare computed considering every possible combination. Here, m = 37 is thetotal number of Facebook groups. Now, the total number of possible com-binations after considering n arbitrary groups, out of m , is m C n where thecombination index can be represented by i = { , , · · · , m C n } . Let us repre-sent, the member sets of n groups for i th index combination by q i , q i , · · · , q in .Therefore, the total number of common members of n groups together in the i th index combination is S n,i = { q i ∩ q i ∩ · · · ∩ q in } (1)Now, S n = m C n X i =1 S n,i (2)depicts the total number of common members for all m C n combinations of n groups together. Therefore, the average number of common member for n groups is A n ( m ) = 1 m C n S n (3)10here n = { , , · · · , m } . As an example, A (37) indicates the average num-ber of common members after considering all possible combinations of 2groups taken together. A graph representation of the members, with edgesformed based on the groups they belong to, can depict a qualitative associ-ation among the members. Fig. 3 depicts the average number of common members A n (37) as a func-tion of n . According to Fig. 3, the average number of common membersconsidering all combinations of two groups together, i.e. A (37), is 13 . . A (37) ≈ A n ( ) Figure 3: Average number of common members A n (37) as a function of n where n = { , , · · · , } . Fig. 4 depicts the membership association of 6 Facebook groups togethercontaining total 30 ,
699 members. In Fig. 4, the common members are du-plicated in respective groups and hence associated with more than one edge.Therefore, plotting all members together across the groups as nodes reflectson the number of common members in these groups where A (6) = 64 . A (6) = 7 . A (6) = 2 . A (6) = 0 .
66; and A (6) = 0.At first glance, less common members in the groups indicate towardslocalized mass movement situation with less centralized control. Whereas,large count of common members across the groups depict activist dependent11 igure 4: Dynamics of 6 large Facebook groups containing total 30 ,
699 members where A (6) = 64 . A (6) = 7 . A (6) = 2 . A (6) = 0 .
66; and A (6) = 0. Here, thenodes (in blue) represents the unique member Facebook identities/unique Facebook groupidentities. Moreover, the nodes (representing unique member Facebook identities) connectswith corresponding nodes (representing unique Facebook group identities) via edges. A (37) = 13 . A (37)= 1 . A (37) ≈ .
37% of the chosen Facebookgroups are associated with the posters of chosen protest events. This has atilt towards higher sharing of protest information among the active groupsin spite of differences in member orientation. This in our opinion indicatesa tendency towards mass-movement dynamics leaving aside the baggage ofsectarian identity.
5. Contagion chain dynamics
One important aspect of the study is to consider how the contra NRC& CAA Facebook posts flow to the members and see whether any dynamics13f religious participation exists therein. This section studies the contagionchain (‘like/share’) dynamics for somewhat randomly chosen 14 viral Face-book posts related with protest against NRC & CAA. These posts have aninformation catchment containing total 11 ,
642 Facebook profile users. Allthese chosen posts belong to one single contra NRC & CAA Facebook group.Here, seven posts which involve 5 ,
138 Facebook profile users were posted be-fore December 12, 2019 which was the passing date of CAA in Parliament ofIndia. An equal number (7) of posts involving 6 ,
504 Facebook profile usersare chosen from those posted after December 12, 2019. Here, this studyfocuses on the following aspects.1. This study compares the participation of Muslim and Non-MuslimFacebook users in the ‘like/share’ dynamics of the chosen Facebookposts. Such comparison reflects the community participation aroundFacebook posts before and after December 12, 2019. This religious an-gle is important to understand the movement dynamics as the CAA isbiased with religious overtones in determining citizenship of India.2. Moreover, the study can bring out religious dependency in the ‘like/share’dynamics, i.e. whether Facebook users tend to like/share posts fromFacebook users of their own religion can reflect upon the secular atti-tude prevalent in the society.3. In addition to the above objectives, following the methodology of Sec-tion 4, the overlap of Facebook profile users in this ‘like/share’ datasetcan give an idea about the role of activist or any sort of centralizeddependency in shaping the movement dynamics.Here, the raw dataset contains the source Facebook profile user (i.e. fromwhom the post originated), the destination Facebook profile user (i.e. wholiked/shared the post). The members of our research team hand-coded thereligion of profile user from the user’s profile name and surname as Muslim orNon-Muslim. Note that, in case of quantitative and qualitative analysis, toidentify the number of common user profiles in the like/share dynamics forthese 14 Facebook posts, this study follows the same methodology of earlierSection 4.
Table 1 depicts the like/share statistics for the 14 Facebook posts. Thisresult indicates that the participation of Muslim community in the movement14ncreases (16%) as a community after the passage of CAA in the parliament.It may be observed that, the fear within Muslim community is also reflectedfrom the field data, for evidence, see selected part of field responses . Before CAA After CAA TotalSource-Destination pairs Number % Number % Number %Muslim origin 3078 60% 4951 76% 8029 69%Non-Muslim origin 2060 40% 1553 24% 3613 31%M-M 1852 36.04% 3158 48.55% 5010 43.03%NM-NM 1566 30.47% 984 15.12% 2550 21.90%Co-religion total 3418 66.52% 4142 63.68% 7560 64.93%NM-M 1227 23.88% 1794 27.58% 3021 25.94%M-NM 493 9.59% 568 8.73% 1061 9.11%Cross-religion total 1720 33.47% 2362 36.31% 4082 35.06%
Table 1: Religion based like share dynamics in the chosen 14 Facebook posts. Here, M-M (resp. NM-NM; NM-M; M-NM) represents Muslim (resp. Non-Muslim; Non-Muslim;Muslim) profile users post liked/shared by Muslim (resp. Non-Muslim; Muslim; Non-Muslim) profile user.
According to Table 1, for 64 .
93% cases Facebook user profile liked/sharedFacebook post from Facebook user profile of same religion, i.e Non-Muslimprofile user liked/shared Non-Muslim profile user’s post or Muslim profileuser liked/shared Muslim profile user’s post. This statistics remain almostsame if we separately calculate for Facebook post before and after 12 Decem-ber, 2019. It is very interesting to note that, Muslim profile user liked/sharedNon-Muslim profile user’s Facebook post for 25 .
94% cases, however, Non-Muslim profile user liked/shared Muslim profile user’s Facebook post for9 .
11% cases. Fig 5 depicts the group attribute layout based on religion forthe chosen viral Facebook posts before and after December 12, 2019 sepa-rately. In Fig. 5, two Facebook users are connected via edges if one likedor shared another’s post. Such social-media result indicates co-existence ofboth the secular and communal identity in the society. “The leaders of RSS are saying that new bill is going to come and that it will help therefugee Hindus in the process of getting citizenship. It will not even create problems forMuslims but their right to vote will be taken away and hence will deprive them of severalopportunities.”: GM-RH-EHR-AU-DM-L4-O2-A3 “There was a great fear among the Muslims. A Muslim man gives me his LIC premiumof rupees 1.5 Lakh in a month, he owns a rice mill. After all this issues of NRC he hasstopped to do so, saying what will happen by giving premium, they want to chase us fromthe country. Let’s save money, when they will chase us we can sell our houses and go.”:GM-RH-EHR-AU-DM-L3-O4-A2 Figure 5: Like-share dynamics of 7 Facebook posts containing 5138 (resp. 6504) Facebookprofile users before (resp. after)
CAA was passed in Indian Parliament. This Figure showsthe group attribute layout based on religion. The Non-Muslims are marked by red andMuslims are marked by green, i.e. the Red circle is for Non-Muslim users and Green circleis for Muslim users. The edges among them shows the relationship. A n ( m ) Before CAA postsAfter CAA postsAll posts
Figure 6: Average number of common profiles A n (all) (resp. A n (before)/ A n (after)) as afunction of n where n = { , , · · · , } for all post (resp. n = { , , · · · , } for before/afterCAA Facebook posts separately). Now, from a completely different perspective, Fig. 6 depicts the averagenumber of common profiles for all possible permutations of the 14 Facebookposts, for before CAA 7 posts, and for after CAA 7 posts. In a more tranquilsocio-political situation, a Facebook post by an activist traditionally gets likeor share only from activist community. However, for a mass movement dy-namics, it is not true. Therefore, less common profiles in these posts indicatetowards localized mass movement gravitating with less centralized control.Whereas, large count of common profiles across the posts depict activist de-pendent centralized control in the movement. To identify this dynamics, wehave collected all the 14 contra NRC & CAA posts from same Facebookgroup. According to Fig. 6, the average number of common members profileconsidering all combinations of two Facebook posts together approaches 1which indicates again towards the phenomenon of mass movement with lessactivist dependency. Fig. 7 depicts like-share dynamics of 6 viral Facebookposts containing total 7980 Facebook profile users where the nodes depictthe unique Facebook profile user identification. Therefore, two nodes repre-senting Facebook profile user are connected via edges if one liked or sharedanother’s post. This qualitative result shows the visual evidence of the quan-titative result. Moreover, the qualitative result of Fig. 7 depicts that vastmajority of contagion chain die soon which shows similar signature behaviourwith the findings of [2, 17, 18, 19]. 17 igure 7: Like-share dynamics of 6 Facebook posts (3 from before and 3 from after
CAA )containing 7980 Facebook profile users where the non-Muslims are marked by red andMuslims are marked by green. . Analysis of group post content Informational and motivational content of social media posts during protestmovements have been briefly investigated in the literature of social media &social protests [8, 10, 11]. Social scientists have widely discussed about socialpsychological factors like – moral outrage (i.e. anger or indignation at per-ceived injustice) [38, 39, 40, 41, 42, 43, 44], social identification (i.e. a strongsense of group belonging and shared interests) [45, 46, 47, 48, 49, 39] andgroup efficacy (i.e. beliefs about group efficacy or empowerment) [50, 51, 42].Here, this section reflects upon the content analysis of Facebook posts dur-ing the mobilization period of NRC & CAA Protest. To do so, we havecollected a quadrilingual dataset of 3200 random Facebook posts from eightcontra NRC & CAA Facebook groups which are in English, Bangla, Hindiand Asomiya language. We have conducted a quantitative analysis of these3200 Facebook posts.
To investigate information and motivational contents of these Facebookposts during the mobilization period of NRC & CAA protest, we have cate-gorized the Facebook posts into following classes.
Class 1:
Information about contra CAA and NRC meeting and protest,posters, slogans. Moreover, we have further divided this class intofollowing sub-classes. – Participation of Female along with Children and trans-gendercommunity in the protest; and – Participation of Student community in the protest.
Class 2:
Emotional and motivational content like poems, songs in theanti CAA and NRC protest event;
Class 3:
Knowledge information about CAA and NRC which is aboutlegal and document related complication;
Class 4:
General anti - government post about economy, employmentand communal politics of government;
Class 5:
Miscellaneous content.19embers of our research team hand - coded the classification for the se-lected posts. Here, every post was coded by three different research assistantsand the category of the Facebook post was decided from the decision of themajority.
Here, 48 .
40 % of the Facebook posts contained exchange of protest infor-mation, such as details about date, time, location of protest rally, picturesof mass mobilization in the protest rallies. Moreover, 23 .
92 % of the Face-book posts within this class contained information related to participationof Females in the contra NRC & CAA protest. In reality, the participationof women in the protest, especially in Shaheen Bagh movement, have beenwidely discussed in the literature [58, 59]. Women in the Shaheen Bagh area,Delhi began their peaceful protest on December 16, 2019 which has spread tomore that 115 places all over India. According to [58], from around 20 , . . Moreover, 21 . Problems are more in cases of Muslim women, whose life has three parts, beforemarriage they have surnames like ‘Khatun/Begum’ after marriage they can use ‘Bibi’.In maximum cases Muslim women according to Muslim personal law ‘Bibi’ surnames arethere. Now from modern ideologies many do not use ‘Bibi’ but traditionally people usedit. Automatically people used to put ‘Bibi’ according to Muslim Marriage Act. Afterthese Muslim women became widow their surname would change into ‘Bewara’. And thiswas similar in case of my grandmother at first it was ‘Khatun’ then to ‘Bibi’ and laterwhen my grandfather expired voter card was changed and written ‘Bewara’. This is avery big problem and so everyone had to stand in the line for correction of documents. :GM-RM-EM-AR-DNM-L3-O3-A2 The problem that came to me was that a woman was married and had a child ofaround 2 years, after which she had problems with her husband and he did not give herany documents. Then the lady returned back to the village and to get new documents sheapplied for panchayat certificates. These types of problems were seen among women and
In cities mainly Kolkata, there is a place for movement to de-velop because of the few universities of Kolkata. Those universi-ties have got many educated people who have a better understand-ing about Indian political background. From that basis of politi-cal background they hold their protest. There have been protestsin Jadavpur University, but how many such protests were therein Jadavpur Lok Sabha Constituency? There have been no suchprotests related to NRC-NPR-CAA in Jadavpur Lok Sabha Con-stituency. The protest that happened mainly involved students ofJadavpur University. This is main advantage in Kolkata, and sothe protests go on. ”On the other hand, 25.18% of Facebook posts contained emotional andmotivational content, such as news of death of protester during protest, newsof death of people in the Detention camps, poems and songs about why oneshould take part in the protest movement, why this process is harmful forboth the original inhabitants and the migrant people, why this procedureis threat to the Muslim community etc.. In this class, a sizable number ofFacebook posts contained the heartbreaking incidents of Assam after NRCimplementation. In this regard, we observed that impact of Assam NRC(14 .
94 %) are discussed by people during the field interviews. Here, 7 .
15 % ofFacebook posts contained knowledge information about the Government pro-cedure of NPR, NRC and CAA. This is corroborated by the field interviews,where the detail knowledge about complex technicalities of implementationon NRC & CAA (4 .
63 %) are responded by very few people which depictssimilar behaviour. Moreover, we are not able to categorize 14.53% Facebookposts in any category which remains as miscellaneous posts. However, theseFacebook posts are not always explicitly ’irrelevant’ or ’spam’. Sometime,these Facebook posts are associated with anti government feelings. even for that child. They are still unable to get the corrected documents.: GM-RM-EM-AR-DNM-L3-O2-A2
Figure 8: % wise result of the Facebook posts (a) for the classes 1-5 and (b) result withinclass 1.
Table 2: Examples of the Facebook posts for each class. ig. 8 depicts the summarized results of 3200 Facebook posts. Moreover.Table 2 shows the example of Facebook post for every class. In comparisonwith literature, the result of Fig. 8 shows similar signature behaviour withtweet activity during Occupy Wall Street protest. Langer et. al. [11] havequalitatively analysied more than 7000 tweets on Occupy Wall Street proteststhat occurred in New Yark city where 44% of tweets contained informationalcontent, i.e details about protest location, safety, police presence.Note that, here, this study conducted on Facebook posts during the mo-bilization period of contra NRC & CAA protest starting from mid-December,2019 to mid-February, 2020. However, the situation drastically changed inMarch after commencement of the lock-down due to the spread of COVID-19.For evidence, we have investigated the informational contents of Facebookposts in the contra NRC & CAA Facebook groups during the lock-down pe-riod following the same methodology. In this study, we have collected 500posts randomly during the last week of March (1st week of lock-down periodin India). It is interesting to note that only 10 .
8% Facebook posts containedinformation about NRC & CAA. Along with that, 29 .
4% Facebook posts, outof total 500 Facebook posts, contained issues related with NRC & CAA andsituation of COVID-19 both. However, 53 .
6% of Facebook posts containedinformation about issues related with only COVID-19, that is, general andhealth information related with COVID-19, situation of working class peopleand failures of government during the lock down period.
7. Spatiotemporal analysis of individual activity
This section studies the Twitter activity during the mobilization periodof NRC & CAA protest. To do so, we have collected tweets for 48 hash-tagsduring the period from December 10,2019 to February 10, 2020. The 48hash-tags contain total 5,05,792 tweets during the two month mobilizationperiod of NRC & CAA protest. The study deals with 4,90,978 tweets, out of5,05,792 tweets and the rest are excluded from the study due to incompleteinformation. Here, “ .1. Dynamics of number of tweets
To understand the dynamics of number of tweets per day, we have plottednumber of tweets against time during the period of December 10, 2019 toFebruary 10, 2020. This section shares similar notation with Section 2 where N ( k ) indicates total number of tweets present at instant k . Total time of thestudy is divided into slots of 2-days interval. Here, N ( n ) depicts the totalnumber of tweets at the end of our time window, i.e. February 10, 2020.Therefore, N ( n ) = 4 , , N ( k ) / N ( n ) ( N ( k + ) - N ( k )) / N ( n ) (a) (b) Figure 9: The growth of (a) N ( k )/ N ( n ) over time; (b) ( N ( k + 1) - N ( k ))/ N ( n ) against k . As a result, Fig. 9(a) depicts the growth of the normalized number oftweets. Fig. 9(b) shows the acceleration in tweet count. These plots depictthe following dynamics of tweets.1. There is sudden rise (highest peak) in the plots of Fig. 9 after k = 13,which represents the rise in number of tweets after January 6, 2020.Note that, in the first week of January 2020, the goons attacked stu-dents in Jawahar Lal Nehru University. In fact, the higher educationhas becomes a battle field in the war of culture and ideologies [60]. If wefollow the rise of tweets with respect to total number of tweets, we findthat the peak is touched during that time (see Fig. 9). This is an in-dication that the people are increasingly joining the protest movementafter January first week.2. Moreover, the plots of Fig. 9 show the first peak after k = 3, whichrepresents December 16, 2019. In this context, the most popular Sha-25een Bagh protest began on December 16, 2019 following the violentbrutal attack by Delhi police on the students of Jamia Millia IslamiaUniversity on December 15, 2019 [58, 59].The plots show the last peak after k = 25 which represents date afterJanuary 29, 2020. Here also, on January 30, 2020, an armed Hindufundamentalist man fired at a crowd gathered in Jamia Millia IslamiaUniversity.3. It is further observed that 10 hash-tags, out of 48, are associated withprotest movement by students. 1 , ,
588 tweets are related with these10 hash-tags which is 24 .
63% of total number of tweets. On the otherhand, 4 hash-tags, out of 48, are associated with protest by Muslimwomen which is 7 .
6% (38,472) of total number of tweets. This is an in-dication of participation of students and Muslim women in the protestmovement. The details of Facebook posts on female and student par-ticipation in the protest and their relationship with field responses arediscussed in Section 6. The twitter data also reflects similar signaturebehaviour.
The dataset contains the location of twitter user profile (where availablein the twitter profile). Here, the dataset depicts the list of 82 countriesacross world and 47 cities within India along with number of tweets from thecorresponding country/city. The location of tweets can be able to give aninsight about the relationship between ground reality and online activism.In literature, in case of Egyptian revolution of 2011, less than 30% of tweetsoriginated in Egypt [9]. However, most of the tweets, out of 30 milliontweets, were sent in Turkish from inside the country during Turkish protestof 2013-14 [8].Fig. 10 depicts countries (except India) across the world with number oftweets as parameter during the mobilization period December 10, 2019 toFebruary 10, 2020 of contra NRC & CAA protest. Note that, out of 4,90,978tweets, 2,64,071 tweets are associated with the specific location of the usertwitter profile. Out of these 2,64,071 tweets, 2,33,960 tweets (88.60%) arefrom India, whereas 12,992 (4.92%) (resp, 9,716(3.68%); 5,045(1.91%)) tweetsare respectively associated with location in Middle east countries, USA andEurope. That is, most of the tweets ( ≈ igure 10: Countries across the world from which tweets originated during the mobilizationperiod of contra NRC & CAA protest (outside India). India which reflects similar signature with Turkish protest of 2013-14 [8], butnot with Egyptian revolution of 2011 [9].However, in India also most of the tweets are associated with location inMetro cities. Here, 61,353 (resp. 28,399; 10,016; 7,098), i.e. 26.22% (resp.12.14 %; 4.28 %; 3.03 %), tweets are associated with metro city locationDelhi (resp. Mumbai; Bangalore; Kolkata). Fig 11 depicts location wisetweets density in Indian cities during the mobilization period of contra NRC& CAA protest. In conclusion, this is an indication that the people ontwitter network of contra NRC & CAA protest are mostly from metro citiesand belongs to elite class.
Structure of twitter network during mass mobilization period of socialprotests have been widely discussed in the literature, in case of mass mobi-lization in Spain 2011 [2], ‘united for global change’ demonstration in May2012 [3], Istanbul’s Gazi park protest in May 2013 [1, 3]. These studiesmainly focus on dynamics of protest recruitment and information flow withrespect to the role of active and moderately active twitter users. In this study,27 igure 11: Tweets density in Indian cities during the mobilization period of contra NRC& CAA protest.
28e explore the survival dynamics of online activism during the mobilizationperiod of contra NRC & CAA protest.The 4 , ,
978 tweets during the two month mobilization period of contraNRC & CAA protest are tweeted by 1 , ,
019 unique twitter user. That is,on average each user has twitted four times during the corresponding twomonth mobilization period of contra NRC & CAA protest. To understandmore detail dynamics, let us consider, X , Y are sets with twitter users havingtweeted in time step t and t + 1. The size of the sets are x = | X | and y = | Y | . Therefore, we define,Survive = | X ∩ Y | (Twitter users having tweeted);Birth = | Y − X | (Newly active twitter users);Death = | X − Y | (Twitter users active only in previous).Here, “ ∩ , − ” respectively notes the notion of intersection and differencefollowing classical set theory notation. Different size of time step is consid-ered where D = n depicts time step of size n days ( n ×
24 hours). Now,Fig. 12 (a), (b), (c) respectively shows change in actual number of survive,death, birth for D = { } . That is, we plot Survive(t), Death(t),Birth(t) against t in Fig. 12. S e r v i v e ( t ) D=1D=2D=3D=4D=5D=6D=8 D e a t h ( t ) D=1D=2D=3D=4D=5D=6D=8 B i r t h ( t ) D=1D=2D=3D=4D=5D=6D=8
Figure 12: Change in Survive, Death, Birth with time for D = { } . To understand frequencies involved, fast Fourier transformation (FFT)[61] has been computed using standard subroutines. The FFT of a time seriescan provide an idea about the dominant frequencies associated with the timeseries data. In the parlance of signal processing, the time domain signal isexpressed as sum of signals having known frequencies with varying amplitudethat depends on the actual data values. Here the dataset comprises the29ount of unique individuals tweeting on a given day. Hence the FFT analysiscomputes the amplitude (modulus used for handling the complex number)for the defined frequencies. It resembles the count of unique individuals whotweet as frequently as daily, on alternate days, twice a week and so on. Inconclusion, it can be seen that there is no dominant group of individual userswho tweeted regularly on the issue, as is expected in case automated faketweets are fired which contradict with the widespread usage of (fake/bot)accounts in Indian political twitter network [62]. Fig. 13 depicts amplitudeof survive over frequency (days) for D = { } . A m p li t u d e o f s u r v i v e A m p li t u d e o f s u r v i v e A m p li t u d e o f s u r v i v e D = 1 D = 2 D = 3
Figure 13: Amplitude of survive over frequency (days) for D = { } . To summarize, the peaks of number of tweets tally with ground eventsduring the mobilization period of contra NRC & CAA protest. However, thetwitter activity only reflects from elite class of metro cities. Moreover, thoughthere is no evidence of presence of (fake/bot) twitter accounts in contra NRC& CAA twitter network, the survival rate of twitter users depicts no evidenceof recruitment patterns for online activism as well. Therefore, It may be truethat the intention of twitter activity is to make ‘contra NRC & CAA’ hashtagsas only twitter trends to seek the attention of global spectator.
8. Field interview findings
During the field visit, we received 72 field interviews (with a gender ra-tio of 5 M : 1 F ) from 6 districts (Dakshin Dinajpur, Uttar Dinajpur, CoochBehar, Alipurduar of West Bengal and Dhubri, Kokrajhar of Assam). How-ever, most of the respondents were from West Bengal. Here, 72 .
22 % of30he respondents are Hindus, whereas 25 % are Muslims. In terms of eth-nicities, 45.83% (resp. 18.05 %; 22.22 %; and 13.88 %) responding peoplebelongs from Hindu refugee (resp. Ancient (non-migrant) Hindu; Muslim;and Rajbanshi) community. The demography related information of the re-spondents are summarized in Fig 14. Most of the responding people werefrom middle age group, belonged either to the age group 25-40 (36 . . To understand the field interviews, we classify the content of field inter-views into following nine classes:T1 As a community, Muslims are in Problem;T2 Impact of Assam NRC;T3 Confusing statements by political leadership/social media;T4 Migration after 1971 and related problems;T5 Problem faced by son of the soil;T6 Business/harassment in the name of Document correction;T7 Sectarian and Communal content;T8 Technicalities of implementation; andT9 Fallout of Government Failure - political/economic and protests.Now, we again put these nine classes under following three major classes:(C1)
Documents [class: T1, T4, T5]; (C2)
Political [class: T3, T7, T9];and (C3)
Implementation [class: T2, T6, T8]. Fig. 15 depicts the % wisetopic responses in the field interviews for the nine classes and three majorclasses. We noticed that problem faced by son of the soil (19.58%), impact ofAssam NRC (14.94%), migration after 1971 and related problems (13.91%)are responded by people with significantly higher importance.
Bar chart can highlight association between the different demographicfactors and topic of responses in the field interview. Fig. 16 depicts Barchart for different demographic factors. It is interesting to note that Muslim31 igure 14: Demographic details of the respondents and district of West Bengal and Assamto which the people are responding in field interviews. igure 15: % wise topic responses in the field interviews for the classesT1, T2, T3, ,T4, T5, T6, T7, T8, T9 and major classes C1, C2, C3. people, i.e. ethnic Muslim, express a significantly higher level of concernabout the problems of son of the soil than the problems of Muslim community.Moreover, we found that Muslims (religion wise) have no association with themigration after 1971 of Hindus and related problems. On the other hand,Rajbanshi community (ethnicity wise) strictly speaks about their own “son ofthe soil” sentiments. We noticed that the Assam NRC has high impact on thefemales (gender wise), illiterate people (level of literacy wise) and workingpeople from the informal sector (occupation wise). However, Assam NRCreflects almost same impact for Urban and Rural people (area of residencewise); and migrants and non-migrants people (demography wise). Again, it isvery interesting to note that urban and educated people express significantlyhigher concern about economic fallout of Government in association withNRC & CAA issue.Next, to understand the expressiveness of the field interviews, we haveplotted a Bipartite graph G = (U,V,E), where U = { R , R , · · · , R n } indicatesthe set of field responders, V = { T , T , · · · , T } depicts the set of nine topicclasses and E denotes the set of edges of the bipartite graph. Here, node R i and T j are connected via edge if R i responder in the field interview speaksabout T j topic. Therefore, the ‘expressiveness’ property of responder in thefield interview indicates the number of cycles in the bipartite graph startingfrom the corresponding responder node. As an example, R j → T → R j → T → R j → T → R j indicates a length three cycle covering T , T , T topics. Therefore, different length cycle is possible following this procedure.33 % MaleFemale
T1 T2 T3 T4 T5 T6 T7 T8 T905101520253035 % HinduMuslim
T1 T2 T3 T4 T5 T6 T7 T8 T9051015202530354045 % MuslimHindu RefugeeAncient HinduRajbanshi
T1 T2 T3 T4 T5 T6 T7 T8 T9051015202530 % UrbanRural
T1 T2 T3 T4 T5 T6 T7 T8 T9051015202530 % MigrantsNon-Migrants
T1 T2 T3 T4 T5 T6 T7 T8 T905101520253035 % IlliterateUpto class 10GraduateHigher Education
T1 T2 T3 T4 T5 T6 T7 T8 T90510152025 % StudentGovt. JobBusinessUnorganized Sector
T1 T2 T3 T4 T5 T6 T7 T8 T9051015202530 % <2525-4040-6060+ Figure 16: Association between the different demographic factors and topic of responses.
Let us consider, c nR j indicates number of n length cycle for R j responder.Therefore, total number of cycle for responder R j is, C R j = c R j + c R j + · · · + c nR j (4)Moreover, W C R j = 1 × c R j + 2 × c R j + · · · + n × c nR j (5)Here, expressiveness of a responder is depicted by parameter C R j and W C R j which indicates the local connection of a responder with other re-sponder.Next, we examined expressiveness ( C R j ) of the responders for differentdemographic category. The violin plots [63] in Fig. 17 highlight the expres-siveness score over different subgroups of demographic factors. We noticedthat the mean expressiveness level of the people belonging to the ethnic Hindurefugee group significantly higher (Kruskal-Wallis [63], p = 0 . . . { T , · · · , T } , C R j and W C R j , we used34 igure 17: Expressiveness (number of cycle in the bipartite graph) against different de-mographic factors. Here, for gender, Male: GM, Female: GF; for Religion, Hindu: RH,Muslim: RM; for Ethnicities, Rajbanshi: ER, Hindu Refugee: EHR, Muslim: EM, An-cient Hindu: EAH; for Area of Residence, Urban: AU, Rural: AR; for Demographics,Migrants: DM, Non-Migrants: DNM; for Level of Literacy, Illiterate: L1, Upto class 10:L2, Graduate: L3, Higher Education: L4; for Occupation, Student: O1, Govt Job: O2,Business: O3, Unorganized Sector: O4; for Age, less than 25: A1, (25-40): A2, (40-60):A3, (more than 60): A4. igure 18: Alluvial diagram to realize the relationship between (A) ethnicity of responderwhere Rajbanshi: ER, Hindu Refugee: EHR, Muslim: EM, Ancient Hindu: EAH; (B)association of a responder with number of topics ( { T , · · · , T } ), (C) number of cycle inthe Bipartite graph C R j and (D) W C R j . C R j and connectionwith others W C R j . Fig. 18 highlights that most the people from Rajbanshicommunity has association with only one topic and very less expressiveness.To summarize, this section depicts the topic wise responses in the fieldinterviews; association between different demographic factors and topic of re-sponses; and expressiveness of field interviews against different demographicfactors. Moreover, the correlation between social media dynamics and fieldresponses is already discussed in previous sections.
9. Conclusion
This work has explored the contra NRC based content in social media,mainly from the forums of Facebook groups and Twitter activists. The dy-namical nature of the groups with respect to their formation and membershipgrowth has been analyzed in detail. The textual content of the deliberationsin English, Bangla, Hindi and Asomiya posted by the members has beensubject to scrutiny and classified into relevant categories. The like and sharedynamics of some viral posts have been characterised using graph theory al-gorithms. The study also considers the spatial dynamics of individual Twit-ter activists reacting to the NRC / CAA policies of the central government.Throughout all this analysis, the goal has been to study how and whether thesocial media is able to build up a mass response (or movement ) against thedraconian and sectarian policies of granting and retaining Indian citizenship.Field work has been conducted to generate evidence for the social mediabased analysis. The following points could be established from the availableevidence. • There is a perceptible dichotomy in people’s choice between abiding bynew requirements and joining the protest movement. Looking at theaffidavit / document correction crowd size and the timings as reportedby the interviewed people in the court areas of both Uttar and DakshinDinajpur, this becomes evident. The dates when the crowd increasedin number also saw huge growth in membership of social media groupsagainst NRC. This means that people are becoming part of the protestmovement and supportive of protests, but everyone is not doing soconvincingly, because at the same time the crowd for taking precautionskept swelling. 37
Influence of Assam situation and NRC conducted there is quite per-ceptible in social media . The ground truth as evidenced by field visitsto the bordering districts of Kokrajhar and Dhubri indicate similar na-ture. The timing of the field visit coincided with the protest againstCAA in the state of Assam. The topic was being highly debated amongthe people. There was dissent regarding the nature of implementationof NRC in Assam and at the same time dissent was growing about thefact that the whole effort of six years was getting nullified and people’ssacrifice becoming fruitless • Participation of Muslims before and after CAA or Ayodhya ruling bysupreme court has been studied from social media datasets. Similarnature of Muslim participation is evidenced by the field visits. Thedates of visit incidentally fell during two crucial rulings of Ayodhyaruling and CAA passage. The feeling of alienation and need for protestwas voiced by those interviewed by us. The community seemed to berelying on protest movement to a great extent. • Nature of Leadership of the contra NRC movement has shown somecontradiction. The social media analysis shows non-hierarchical or flatleadership for movement in social media. But the findings during thefield work has given evidence of hierarchical party presence. Manyof the interviewees have talked about the different political parties,especially their local leadership operating under instructions from thecentral party leadership. • The partition of undivided Bengal is another area of departure. Thisaspect is relatively scarce in the discussion topics of the social mediagroups. But during field visit it appears that the partition has left deepwound in people’s mind. • Another aspect that shows contradictory nature is with regard to thequality of the topics being discussed. Social media groups appear tobe more progressive than some field visit findings which brought outintensive communal nature of the participants.The intersection between the two sets, the members of the social mediaand the field interviewees, is almost null set. Hence the contradictory natureof the findings can be explained to an extent. With time, it is expected38hat when the government starts to push ahead with the implementationof NRC / CAA, the contradiction will either lead to withering away of theprotest movement or produce one of the still unfolding mass movement withemerging non-hierarchical leadership.
Acknowledgements
This research is partially supported by ImpactfulPolicy Research in Social Science (IMPRESS) under Indian Council of SocialScience Research, Govt. of India (P3271/220).The authors are grateful to Swarnava Chakraborty for collecting socialmedia dataset as an undergraduate intern student. The authors are alsograteful to Sk Arafat Zaman for his work as an Asamiya language translator.
References [1] J. T. Jost, P. Barber´a, R. Bonneau, M. Langer, M. Metzger, J. Nagler,J. Sterling, J. A. Tucker, How social media facilitates political protest:Information, motivation, and social networks, Political Psychology 39(2018) 85–118.[2] S. Gonz´alez-Bail´on, J. Borge-Holthoefer, A. Rivero, Y. Moreno, Thedynamics of protest recruitment through an online network, ScientificReports (2011).[3] P. Barber´a, N. Wang, R. Bonneau, J. T. Jost, J. Nagler, J. Tucker,S. Gonz´alez-Bail´on, The critical periphery in the growth of socialprotests, PLOS ONE 10 (2015) 1–15.[4] D. Faris, Revolutions without revolutionaries? socail media networksand regime responce in egypt, (Unpublished Doctoral Dissertation)(2010).[5] G. Lotan, E. Graeff, M. Ananny, D. Gaffney, I. Pearce, danah boyd, Thearab spring— the revolutions were tweeted: Information flows duringthe 2011 tunisian and egyptian revolutions, International Journal ofCommunication 5 (2011).[6] V. V. Lysenko, K. C. Desouza, Moldova’s internet revolution: Analyzingthe role of technologies in various phases of the confrontation, Techno-logical Forecasting and Social Change 79 (2012) 341 – 361. ContainsSpecial Section: Emerging Technologies and Inequalities.397] A. Mungiu-Pippidi, I. Munteanu, Moldova’s ”twitter revolution”, Jour-nal of Democracy 20 (2009) 136–142.[8] J. A. Tucker, J. Nagler, M. M. Metzger, P. Barber´a, D. Penfold-Brown,R. Bonneau, Big data, social media, and protest: Foundations for a re-search agenda, in: R. M. Alvarez (Ed.), Computational Social Science:Discovery and Prediction, Analytical Methods for Social Research, Cam-bridge University Press, 2016, p. 199–224.[9] K. Starbird, L. Palen, (how) will the revolution be retweeted? infor-mation diffusion and the 2011 egyptian uprising, in: Proceedings ofthe ACM 2012 Conference on Computer Supported Cooperative Work,CSCW ’12, Association for Computing Machinery, New York, NY, USA,2012, p. 7–16.[10] O. Onuch, Social networks and social media in ukrainian “euromaidan”protest, Washigton Post (2014).[11] M. Langer, J. Jost, R. Bonneau, M. Metzger, S. Noorbaloochi,D. Penfold-Brown, Digital dissent: An analysis of the motivational con-tents of tweets from an occupy wall street demonstration., MotivationScience (2018).[12] E. Parker, Social media and the hong knong protests, The New Yorker(2014).[13] M. Gladwell, Small change, why the revolution will not be tweeted, TheNew Yorker (2010).[14] R. M. Bond, C. J. Fariss, J. J. Jones, A. D. I. Kramer, C. Marlow, J. E.Settle, J. H. Fowler, A 61-million-person experiment in social influenceand political mobilization, Nature (2012).[15] C. Shirky, The political power of social media technology, the publicsphere, and political change, Foreign Affairs 90 (2011).[16] E. Morozov, The Net Delusion: How Not to Liberate The World, AllenLane, London, 2011.[17] E. Bakshy, J. Hofman, W. Mason, D. Watts, Everyone’s an influencer:Quantifying influence on twitter, pp. 65–74.4018] H. Kwak, C. Lee, H. Park, S. Moon, What is twitter, a social networkor a news media?, in: Proceedings of the 19th International Conferenceon World Wide Web, WWW ’10, Association for Computing Machinery,New York, NY, USA, 2010, p. 591–600.[19] E. Sun, I. Rosenn, C. Marlow, T. M. Lento, Gesundheit! modelingcontagion through facebook news feed, in: ICWSM.[20] J. C. Jenkins, Resource mobilization theory and the study of socialmovements, Annual Review of Sociology 9 (1983) 527–553.[21] M. Granovetter, Threshold models of collective behavior, AmericanJournal of Sociology 83 (1978) 1420–1443.[22] M. Biggs, Strikes as forest fires: Chicago and paris in the late nineteenthcentury, American Journal of Sociology 110 (2005) 1684–1714.[23] T. C. Schelling, Micromotives and macrobehavior, Norton, London,1978.[24] G. Marwell, P. Oliver, The Critical Mass in Collective Action, Studiesin Rationality and Social Change, Cambridge University Press, 1993.[25] M. W. Macy, Learning theory and the logic of critical mass, AmericanSociological Review 55 (1990) 809–826.[26] M. Granovetter, R. Soong, Threshold models of diffusion and collectivebehavior, The Journal of Mathematical Sociology 9 (1983) 165–179.[27] D. J. Watts, A simple model of global cascades on random networks,Proceedings of the National Academy of Sciences 99 (2002) 5766–5771.[28] D. J. Watts, P. S. Dodds, Influentials, networks, and public opinionformation, Journal of Consumer Research 34 (2007) 441–458.[29] G. Marwell, P. E. Oliver, R. Prahl, Social networks and collective action:A theory of the critical mass. iii, American Journal of Sociology 94(1988) 502–534.[30] M. W. Macy, Chains of cooperation: Threshold effects in collectiveaction, American Sociological Review 56 (1991) 730–747.4131] R. V. Gould, Collective action and network structure, American Socio-logical Review 58 (1993) 182–196.[32] D. Centola, M. Macy, Complex contagions and the weakness of longties, American Journal of Sociology 113 (2007) 702–734.[33] D. A. Siegel, Social networks and collective action, American Journalof Political Science 53 (2009) 122–138.[34] R. V. Gould, Multiple networks and mobilization in the paris commune,1871, American Sociological Review 56 (1991) 716–729.[35] D. McAdam, Recruitment to high-risk activism: The case of freedomsummer, American Journal of Sociology 92 (1986) 64–90.[36] S. Lohmann, The dynamics of informational cascades: The mondaydemonstrations in leipzig, east germany, 1989–91, World Politics 47(1994) 42–101.[37] K.-D. Opp, C. Gern, Dissident groups, personal networks, and spon-taneous cooperation: The east german revolution of 1989, AmericanSociological Review 58 (1993) 659–680.[38] J. Barbalet, Emotin, social theory, and social structure: A macrosociological approach, Cambridge University Press, 1998.[39] J. T. Jost, V. Chaikalis-Petritsis, D. Abrams, J. Sidanius, J. van derToorn, C. Bratt, Why men (and women) do and don’t rebel: Effectsof system justification on willingness to protest, Personality and SocialPsychology Bulletin 38 (2012) 197–208.[40] K. Kawakami, K. L. Dion, Social identity and affect as determinantsof collective action - toward an integration of relative deprivation andsocial identity theories, Theory & Psychology 5 (1995) 551–577.[41] S. St¨urmer, B. Simon, Pathways to collective protest: Calculation, iden-tification, or emotion? a critical analysis of the role of group-based angerin social movement participation, Journal of Social Issues 65 (2009) 681–705. 4242] N. Tausch, J. C. Becker, R. Spears, O. Christ, R. Saab, P. Singh, R. N.Siddiqui, Explaining radical group behavior: Developing emotion andefficacy routes to normative and nonnormative collective action., Journalof personality and social psychology 101 (2011) 129–48.[43] C. Wakslak, J. Jost, T. Tyler, E. Chen, Moral outrage mediates thedampening effect of system justification on support for redistributivesocial policies, Psychological science 18 (2007) 267–74.[44] M. Van Zomeren, T. Postmes, R. Spears, Toward an integrative socialidentity model of collective action: A quantitative research synthesis ofthree socio-psychological perspectives, Psychological bulletin 134 (2008)504–35.[45] L. G. E. Smith, E. F. Thomas, C. McGarty, “we must be the changewe want to see in the world”: Integrating norms and identities throughsocial interaction, Political Psychology 36 (2015) 543–557.[46] C. McGarty, E. F. Thomas, G. Lala, L. G. E. Smith, A.-M. Bliuc, Newtechnologies, new identities, and the growth of mass opposition in thearab spring, Political Psychology 35 (2014) 725–740.[47] B. Klandomans, The social psychology of protest, Oxford, United King-dom, 1997.[48] C. Kelly, The social psychology of collective action, Taylor & Francis,1996.[49] J. Drury, S. Reicher, Collective psychological empowerment as a modelof social change: Researching crowds and power, Journal of Social Issues65 (2009) 707–725.[50] D. Mazzoni, M. van Zomeren, E. Cicognani, The motivating role ofperceived right violation and efficacy beliefs in identification with theitalian water movement, Political Psychology 36 (2015) 315–330.[51] M. van Zomeren, C. W. Leach, R. Spears, Protesters as “passionateeconomists”: A dynamic dual pathway model of approach coping withcollective disadvantage, Personality and Social Psychology Review 16(2012) 180–199. 4352] Amit shah in kolkata: Nrc to be implemented in west bengal, infiltratorswill be evicted, The Economic Times (01 October 2019).[53] Assam nrc: What next for 1.9 million ‘stateless’ indians?, BBC NEWS(31 August 2019).[54] B. Kuchay, What you should know about india’s ‘anti-muslim’ citizen-ship law, ALJAZEERA (16 December 2019).[55] J. Chatterji, The Spoils of Partition: Bengal and India, 1947-1967,Cambridge Studies in Indian History and Society, Cambridge Univer-sity Press, 2007.[56] J. Chatterji, Partition’s Legacies, Hedgehog and fox, Permanent Black,2019.[57] D. Sengupta, The Partition of Bengal: Fragile Borders and New Identi-ties, Cambridge University Press, 2015.[58] S. Nigam, Many dimensions of shaheen bagh movement in india, SSRN(2020).[59] S. Mustafa (Ed.), Shaheen Bagh And The Idea Of India, Speaking Tiger,2020.[60] Protect india’s universities, Nature (14 January 2020).[61] C. V. Loan, Computational Frameworks for the Fast Fourier Transform,SIAM, 1992.[62] N. Ganguly, P. Kumaraguru, The positive and negative effects of socialmedia in india, Communications of the ACM 62 (2019) 98–99.[63] J. L. Hintze, R. D. Nelson, Violin plots: A box plot-density trace syn-ergism, The American Statistician 52 (1998) 181–184.[64] M. Rosvall, C. Bergstrom, Mapping change in large networks. plos one,5(1), e8694, PloS one 5 (2010) e8694.44 N ( t ) / N ( N ( t + ) - N ( t )) //