Predictive Policing: What is it, How it works, and its Legal Implications
This article reviews literature surrounding big data and predictive policing and provides an analysis of the legal implications of using predictive policing techniques in the Indian context.
For the longest time, humans have been obsessed with prediction. Perhaps the most well-known oracle in history, Pythia, the infallible Oracle of Delphi was said to predict future events in hysterical outbursts on the seventh day of the month, inspired by the god Apollo himself. This fascination with informing ourselves about future events has hardly subsided in us humans. What has changed however is the methods we employ to do so. The development of Big data technologies for one, has seen radical applications into many parts of life as we know it, including enhancing our ability to make accurate predictions about the future.
One notable application of Big data into prediction caters to another basic need since the dawn of human civilisation, the need to protect our communities and cities. The word 'police' itself originates from the Greek word 'polis', which means city. The melding of these two concepts prediction and policing has come together in the practice of Predictive policing, which is the application of computer modelling to historical crime data and metadata to predict future criminal activity. In the subsequent sections, I will attempt an introduction of predictive policing and explain some of the main methods within the domain of predictive policing. Because of the disruptive nature of these technologies, it will also be prudent to expand on the implications predictive technologies have for justice, privacy protections and protections against discrimination among others.
In introducing the concept of predictive policing, my first step is to give a short explanation about current predictive analytics techniques, because these techniques are the ones which are applied into a law enforcement context as predictive policing.
What is predictive analysis
Facilitated by the availability of big data, predictive analytics uses algorithms to recognise data patterns and predict future outcomes. Predictive analytics encompasses data mining, predictive modeling, machine learning, and forecasting. Predictive analytics also relies heavily on machine learning and artificial intelligence approaches . The aim of such analysis is to identify relationships among variables that may not be immediately apparent using hypothesis-driven methods. In the mainstream media, one of the most infamous stories about the use of predictive analysis comes from USA, regarding a department store Target and their data analytics practices . Target mined data from purchasing patterns of people who signed onto their baby registry. From this they were able to predict approximately when customers may be due and target advertisements accordingly. In the noted story, they were so successful that they predicted pregnancy before the pregnant girl's father knew she was pregnant. 
Examples of predictive analytics
- Predicting the success of a movie based on its online ratings
- Many universities, sometimes in partnership with other firms use predictive analytics to provide course recommendations to students, track student performance, personalize curriculum to individual students and foster networking between students.
- Predictive Analysis of Corporate Bond Indices Returns
Relationship between predictive analytics and predictive policing
The same techniques used in many of the predictive methods mentioned above find application into some predictive policing methods. However two important points need to be raised:
First, predictive analytics is actually a subset of predictive policing. This is because while the steps in creating a predictive model, of defining a target variable, exposing your model to training data, selecting appropriate features and finally running predictive analysis  maybe the same in a policing context, there are other methods which may be used to predict crime, but which do not rely on data mining. These techniques may instead use other methods, such as some of those detailed below along with data about historical crime to generate predictions.
In her article "Policing by Numbers: Big Data and the Fourth Amendment", Joh categorises 3 main applications of Big data into policing. These are Predictive Policing, Domain Awareness systems and Genetic Data Banks. Genetic data banks refer to maintaining large databases of DNA that was collected as part of the justice system. Issues arise when the DNA collected is repurposed in order to conduct familial searches, instead of being used for corroborating identity. Familial searches may have disproportionate impacts on minority races. Domain Awareness systems use various computer software and other digital surveillance tools such as Geographical Information Systems  or more illicit ones such as Black Rooms to "help police create a software-enhanced picture of the present, using thousands of data points from multiple sources within a city" . I believe Joh was very accurate in separating Predictive Policing from Domain Awareness systems, especially when it comes to analysing the implications of the various applications of Big data into policing.
In such an analysis of the implications of using predictive policing methods, the issues surrounding predictive technologies often get conflated with larger issues about the application of big data into law enforcement. That opens the debate up to questions about overly intrusive evidence gathering and mass surveillance systems, which though used along with predictive technology, are not themselves predictive in nature. In this article, I aim to concentrate on the specific implications that arise due to predictive methods.
One important point regarding the impact of predictive policing is how the insights that predictive policing methods offer are used. There is much support for the idea that predictive policing does not replace policing methods, but actually augments them. The RAND report specifically cites one myth about predictive policing as "the computer will do everything for you". In reality police officers need to act on the recommendations provided by the technologies.
What is Predictive policing?
Predictive policing is the "application of analytical techniques-particularly quantitative techniques-to identify likely targets for police intervention and prevent crime or solve past crimes by making statistical predictions". It is important to note that the use of data and statistics to inform policing is not new. Indeed, even twenty years ago, before the deluge of big data we have today, law enforcement regimes such as the New York Police Department (NYPD) were already using crime data in a major way. In order to keep track of crime trends, NYPD used the software CompStat to map "crime statistics along with other indicators of problems, such as the locations of crime victims and gun arrests". The senior officers used the information provided by CompStat to monitor trends of crimes on a daily basis and such monitoring became an instrumental way to track the performance of police agencies. CompStat has since seen application in many other jurisdictions .
But what is new is the amount of data available for collection, as well as the ease with which organisations can analyse and draw insightful results from that data. Specifically, new technologies allow for far more rigorous interrogation of data and wide-ranging applications, including adding greater accuracy to the prediction of future incidence of crime.
Predictive Policing methods
Some methods of predictive policing involve application of known standard statistical methods, while other methods involve modifying these standard techniques. Predictive techniques that forecast future criminal activities can be framed around six analytic categories. They all may overlap in the sense that multiple techniques are used to create actual predictive policing software and in fact it is similar theories of criminology which undergird many of these methods, but the categorisation in such a way helps clarify the concept of predictive policing. The basis for the categorisation below comes from a RAND Corporation report entitled 'Predictive Policing: The Role of Crime Forecasting in Law Enforcement Operations' , which is a comprehensive and detailed contribution to scholarship in this nascent area.
Hot spot analysis: Methods involving hot spot analysis attempt to "predict areas of increased crime risk based on historical crime data". The premise behind such methods lies in the adage that "crime tends to be lumpy" . Hot Spot analysis seeks to map out these previous incidences of crime in order to inform potential future crime.
Regression methods: A regression aims to find relationships between independent variables (factors that may influence criminal activity) and certain variables that one aims to predict. Hence, this method would track more variables than just crime history.
Data mining techniques: Data mining attempts to recognise patterns in data and use it to make predictions about the future. One important variant in the various types of data mining methods used in policing are different types of algorithms that are used to mine data in different ways. These are dependent on the nature of the data the predictive model was trained on and will be used to interrogate in the future. Two broad categories of algorithms commonly used are clustering algorithms and classification algorithms:
· Clustering algorithms "form a class of data mining approaches that seek to group data into clusters with similar attributes" . One example of clustering algorithms is spatial clustering algorithms, which use geospatial crime incident data to predict future hot spots for crime.
· Classification algorithms "seek to establish rules assigning a class or label to events". These algorithms use training data sets "to learn the patterns that determine the class of an observation" The patterns identified by the algorithm will be applied to future data, and where applicable, the algorithm will recognise similar patterns in the data. This can be used to make predictions about future criminal activity for example.
Near-repeat methods: Near-repeat methods work off the assumption that future crimes will take place close to timing and location of current crimes. Hence, it could be postulated that areas of high crime will experience more crime in the near future. This involves the use of a 'self-exciting' algorithm, very similar to algorithms modelling earthquake aftershocks . The premise undergirding such methods is very similar to that of hot spot analysis.
Spatiotemporal analysis: Using "environmental and temporal features of the crime location"  as the basis for predicting future crime. By combining the spatiotemporal features of the crime area with crime incident data, police could use the resultant information to predict the location and time of future crimes. Examples of factors that may be considered include timing of crimes, weather, distance from highways, time from payday and many more.
Risk terrain analysis: Analyses other factors that are useful in predicting crimes. Examples of such factors include "the social, physical, and behavioural factors that make certain areas more likely to be affected by crime"
Various methods listed above are used, often together, to predict the where and when a crime may take place or even potential victims. The unifying thread which relates these methods is their dependence on historical crime data.
Examples of predictive policing:
Most uses of predictive policing that have been studied and reviewed in scholarly work come from the USA, though I will detail one case study from Derbyshire, UK. Below is a collation of various methods that are a practical application of the methods raised above.
Hot Spot analysis in Sacramento: In February 2011, Sacramento Police Department began using hot spot analysis along with research on optimal patrol time to act as a sufficient deterrent to inform how they patrol high-risk areas. This policy was aimed at preventing serious crimes by patrolling these predicted hot spots. In places where there was such patrolling, serious crimes reduced by a quarter with no significant increases such crimes in surrounding areas.
Data Mining and Hot Spot Mapping in Derbyshire, UK: The Safer Derbyshire Partnership, a group of law enforcement agencies and municipal authorities sought to identify juvenile crime hotspots. They used MapInfo software to combine "multiple discrete data sets to create detailed maps and visualisations of criminal activity, including temporal and spatial hotspots" . This information informed law enforcement about how to optimally deploy their resources.
Regression models in Pittsburgh: Researchers used reports from Pittsburgh Bureau of Police about violent crimes and "leading indicator"  crimes, crimes that were relatively minor but which could be a sign of potential future violent offences. The researcher ran analysis of areas with violent crimes, which were used as the dependent variable in analysing whether violent crimes in certain areas could be predicted by the leading indicator data. From the 93 significant violent crime areas that were studied, 19 areas were successfully predicted by the leading indicator data.
Risk terrain modelling analysis in Morris County, New Jersey: Police in Morris County, used risk terrain analysis to tackle violent crimes and burglaries. They considered five inputs in their model: "past burglaries, the address of individuals recently arrested for property crimes, proximity to major highways, the geographic concentration of young men and the location of apartment complexes and hotels."  The Morris County law enforcement officials linked the significant reductions in violent and property crime to their use of risk terrain modelling.
Near-repeat & hot spot analysis used by Santa Cruz Police Department: Uses PredPol software that applies the Mohler's algorithm  to a database with five years' worth of crime data to assess the likelihood of future crime occurring in the geographic areas within the city. Before going on shift, officers receive information identifying 15 such areas with the highest probability of crime. The initiative has been cited as being very successful at reducing burglaries, and was used in Los Angeles and Richmond, Virginia.
Data Mining and Spatiotemporal analysis to predict future criminal activities in Chicago: Officers in Chicago Police Department made visits to people their software predicted were likely to be involved in violent crimes, guided by an algorithm-generated "Heat List". Some of the inputs used in the predictions include some types of arrest records, gun ownership, social networks (police analysis of social networking is also a rising trend in predictive policing) and generally type of people you are acquainted with  among others, but the full list of the factors are not public. The list sends police officers (or sometimes mails letters) to peoples' homes to offer social services or deliver warnings about the consequences for offending. Based in part on the information provided by the algorithm, officers may provide people on the Heat List information about vocational training programs or warnings about how Federal Law provides harsher punishments for reoffending.
Predictive policing in India
In this section, I map out some of the developments in the field of predictive policing within India. On the whole, predictive policing is still very new in India, with Jharkhand being the only state that appears to already have concrete plans in place to introduce predictive policing.
The Jharkhand police began developing their IT infrastructure such as a Geographic Information System (GIS) and Server room when they received funding for Rs. 18.5 crore from the Ministry of Home Affairs. The Open Group on E-governance (OGE), founded as a collaboration between the Jharkhand Police and National Informatics Centre, is now a multi-disciplinary group which takes on different projects related to IT. With regards to predictive policing, some members of OGE began development in 2013 of data mining software which will scan online records that are digitised. The emerging crime trends "can be a building block in the predictive policing project that the state police want to try."
The Jharkhand Police was also reported in 2012 to be in the final stages of forming a partnership with IIM-Ranchi. It was alleged the Jharkhand police aimed to tap into IIM's advanced business analytics skills , skills that can be very useful in a predictive policing context. Mr Pradhan suggested that "predictive policing was based on intelligence-based patrol and rapid response" and that it could go a long way to dealing with the threat of Naxalism in Jharkhand.
However, in Jharkhand, the emphasis appears to be targeted at developing a massive Domain Awareness system, collecting data and creating new ways to present that data to officers on the ground, instead of architecting and using predictive policing software. For example, the Jharkhand police now have in place "a Naxal Information System, Crime Criminal Information System (to be integrated with the CCTNS) and a GIS that supplies customised maps that are vital to operations against Maoist groups". The Jharkhand police's "Crime Analytics Dashboard"  shows the incidence of crime according to type, location and presents it in an accessible portal, providing up-to-date information and undoubtedly raises the situational awareness of the officers. Arguably, the domain awareness systems that are taking shape in Jharkhand would pave the way for predictive policing methods to be applied in the future. These systems and hot spot maps seem to be the start of a new age of policing in Jharkhand.
Predictive Policing Research
One promising idea for predictive policing in India comes from the research conducted by Lavanya Gupta and others entitled "Predicting Crime Rates for Predictive Policing", which was a submission for the Gandhian Young Technological Innovation Award. The research uses regression modelling to predict future crime rates. Drawing from First Information Reports (FIRs) of violent crimes (murder, rape, kidnapping etc.) from Chandigarh Police, the team attempted "to extrapolate annual crime rate trends developed through time series models. This approach also involves correlating past crime trends with factors that will influence the future scope of crime, in particular demographic and macro-economic variables" . The researchers used early crime data as the training data for their model, which after some testing, eventually turned out to have an accuracy of around 88.2%. On the face of it, ideas like this could be the starting point for the introduction of predictive policing into India.
The rest of India's law enforcement bodies do not appear to be lagging behind. In the 44th All India police science congress, held in Gandhinagar, Gujarat in March this year, one of the Themes for discussion was the "Role of Preventive Forensics and latest developments in Voice Identification, Tele-forensics and Cyber Forensics".Mr A K Singh, (Additional Director General of Police, Administration) the chairman of the event also said in an interview that there was to be a round-table DGs (Director General of Police) held at the conference to discuss predictive policing. Perhaps predictive policing in India may not be that far away from reality.
CCTNS and the building blocks of Predictive policing
The Ministry of Home Affairs conceived of a Crime and Criminals Tracking and Network System (CCTNS) as part of national e-Governance plans. According to the website of the National Crime Records Bureau (NCRB), CCTNS aims to develop "a nationwide networked infrastructure for evolution of IT-enabled state-of-the-art tracking system around 'investigation of crime and detection of criminals' in real time" 
The plans for predictive policing seem in the works, but first steps that are needed in India across police forces involve digitizing data collection by the police, as well as connecting law enforcement agencies. The NCRB's website described the current possibility of exchange of information between neighbouring police stations, districts or states as being "next to impossible". The aim of CCTNS is precisely to address this gap and integrate and connect the segregated law enforcement arms of the state in India, which would be a foundational step in any initiatives to apply predictive methods.
What are the implications of using predictive policing? Lessons from USA
Despite the moves by law enforcement agencies to adopt predictive policing, one reality is that the implications of predictive policing methods are far from clear. This section will examine these implications on the carriage of justice and its use in law, as well as how it impacts privacy concerns for the individual. It frames the existing debates surrounding these issues with predictive policing, and aims to apply these principles into an Indian context.
Justice, Privacy & IV Amendment
Two key concerns about how predictive policing methods may be used by law enforcement relate to how insights from predictive policing methods are acted upon and how courts interpret them. In the USA, this issue may finds its place under the scope of IV Amendment jurisprudence. The IV amendment states that all citizens are "secure from unreasonable searches and seizures of property by the government". In this sense, the IV amendment forms the basis for search and surveillance law in the USA.
A central aspect of the IV Amendment jurisprudence is drawn from United States v. Katz. In Katz, the FBI attached a microphone to the outside of a public phone booth to record the conversations of Charles Katz, who was making phone calls related to illegal gambling. The court ruled that such actions constituted a search within the auspices of the 4th amendment. The ruling affirmed constitutional protection of all areas where someone has a "reasonable expectation of privacy".
Later cases have provided useful tests for situations where government surveillance tactics may or may not be lawful, depending on whether it violates one's reasonable expectation of privacy. For example, in United States v. Knotts, the court held that "police use of an electronic beeper to follow a suspect surreptitiously did not constitute a Fourth Amendment search". In fact, some argue that that the Supreme Court's reasoning in such cases suggests " any 'scientific enhancement' of the senses used by the police to watch activity falls outside of the Fourth Amendment's protections if the activity takes place in public". This reasoning is based on the third party doctrine which holds that "if you voluntarily provide information to a third party, the IV Amendment does not preclude the government from accessing it without a warrant". The clearest exposition of this reasoning was in Smith v. Maryland, where the presiding judges noted that "this Court consistently has held that a person has no legitimate expectation of privacy in information he voluntarily turns over to third parties".
However, the third party has seen some challenge in recent time. In United States v. Jones, it was ruled that the government's warrantless GPS tracking of his vehicle 24 hours a day for 28 days violated his Fourth Amendment rights. Though the majority ruling was that warrantless GPS tracking constituted a search, it was in a concurring opinion written by Justice Sonya Sotomayor that such intrusive warrantless surveillance was said to infringe one's reasonable expectation of privacy. As Newell reflected on Sotomayor's opinion,
"Justice Sotomayor stated that the time had come for Fourth Amendment jurisprudence to discard the premise that legitimate expectations of privacy could only be found in situations of near or complete secrecy. Sotomayor argued that people should be able to maintain reasonable expectations of privacy in some information voluntarily disclosed to third parties".
She said that the court's current reasoning on what constitutes reasonable expectations of privacy in information disclosed to third parties, such as email or phone records or even purchase histories, is "ill-suited to the digital age, in which people reveal a great deal of information about themselves to third parties in the course of carrying out mundane tasks".
Predictive policing vs. Mass surveillance and Domain Awareness Systems
However, there is an important distinction to be drawn between these cases and evidence from predictive policing. This has to do with the difference in nature of the evidence collection. Arguably, from Jones and others, what we see is that use of mass surveillance and domain awareness systems, drawing from Joh's categorisation of domain awareness systems as being distinct from predictive policing mentioned above, could potentially encroach on one's reasonable expectation of privacy. However, I think that predictive policing, and the possible implications for justice associated with it, its predictive harms, are quite distinct from what has been heard by courts thus far.
The reason for distinct risks between predictive harms and privacy harms originating from information gathering is related to the nature of predictive policing technologies, and how they are used. It is highly unlikely that the evidence submitted by the State to indict an offender will be mainly predictive in nature. For example, would it be possible to convict an accused person solely on the premise that he was predicted to be highly likely to commit a crime, and that subsequently he did? The legal standard of proving guilt beyond a reasonable doubt  can hardly be met solely on predictive evidence for a multitude of reasons. Predictive policing methods could at most, be said to inform police about the risk of someone committing a crime or of crime happening at a certain location, as demonstrated above.
Predictive policing and Criminal Procedure
It may therefore pay to analyse how predictive policing may be used across the various processes within the criminal justice system. In fact, in an analysis of the various stages of criminal procedure, from opening an investigation to gathering evidence, followed by arrest, trial, conviction and sentencing, we see that as the individual gets subject to more serious incursions or sanctions by the state, it takes a higher standard of certainty about wrongdoing and a higher burden of proof, in order to legitimize that particular action.
Hence, at more advanced stages of the criminal justice process such as seeking arrest warrants or trial, it is very unlikely that predictive policing on its own can have a tangible impact, because the nature of predictive evidence is probability based. It aims to calculate the risk of future crime occurring based on statistical analysis of past crime data. While extremely useful, probabilities on their own will not come remotely close meet the legal standards of proving 'guilt beyond reasonable doubt'. It may be at the earlier stages of the criminal justice process that evidence predictive policing might see more widespread application, in terms of applying for search warrants and searching suspicious people while on patrol.
In fact, in the law enforcement context, prediction as a concept is not new to justice. Both courts and law enforcement officials already make predictions about future likelihood of crimes. In the case of issuing warrants, the IV amendment makes provisions that law enforcement officials show that the potential search is based "upon probable cause" in order for a judge to grant a warrant. In US v. Brinegar, probable cause was defined as existing "where the facts and circumstances within the officers' knowledge, and of which they have reasonably trustworthy information, are sufficient in themselves to warrant a belief by a man of reasonable caution that a crime is being committed" . Again, this legal standard seems too high for predictive evidence meet.
However, the police also have an important role to play in preventing crimes by looking out for potential crimes while on patrol or while doing surveillance. When the police stop a civilian on the road to search him, reasonable suspicion must be established. This standard of reasonable suspicion was defined in most clearly in Terry v. Ohio, which required police to "be able to point to specific and articulable facts which, taken together with rational inferences from those facts, reasonably warrant that intrusion". Therefore, "reasonable suspicion that 'criminal activity may be afoot' is at base a prediction that the facts and circumstances warrant the reasonable prediction that a crime is occurring or will occur". Despite the assertion that "there are as of yet no reported cases on predictive policing in the Fourth Amendment context", examining the impact of predictive policing on the doctrine of reasonable suspicion could be very instructive in understanding the implications for justice and privacy .
Predictive Policing and Reasonable Suspicion
Ferguson's insightful contribution to this area of scholarship involves the identification of existing areas where prediction already takes place in policing, and analogising them into a predictive policing context. These three areas are: responding to tips, profiling, and high crime areas (hot spots).
Tips are pieces of information shared with the police by members of the public. Often tips, either anonymous or from known police informants, may predict future actions of certain people, and require the police to act on this information. The precedent for understanding the role of tips in probable cause comes from Illinois v. Gates. It was held that "an informant's 'veracity,' 'reliability,' and 'basis of knowledge'-remain 'highly relevant in determining the value'" of the said tip. Anonymous tips need to be detailed, timely and individualised enough to justify reasonable suspicion . And when the informant is known to be reliable, then his prior reliability may justify reasonable suspicion despite lacking a basis in knowledge.
Ferguson argues that whereas predictive policing cannot provide individualised tips, it is possible to consider reliable tips about certain areas as a parallel to predictive policing. And since the courts had shown a preference for reliability even in the face of a weak basis in knowledge, it is possible to see the reasonable suspicion standard change in its application. It also implies that IV protections may be different in places where crime is predicted to occur .
Despite the negative connotations and controversial overtones at the mere sound of the word, profiling is already a method commonly used by law enforcement. For example, after a crime has been committed and general features of the suspect identified by witnesses, police often stop civilians who fit this description. Another example of profiling is common in combating drug trafficking, where agents keep track of travellers at airports to watch for suspicious behaviour. Based on their experience of common traits which distinguish drug traffickers from regular travellers (a profile), agents may search travellers if they fit the profile. In the case of United States v. Sokolow, the courts "recognized that a drug courier profile is not an irrelevant or inappropriate consideration that, taken in the totality of circumstances, can be considered in a reasonable suspicion determination" . Similar lines of thinking could be employed in observing people exchanging small amounts of money in an area known for high levels of drug activity, conceiving predictive actions as a form of profile.
It is valid to consider predictive policing as a form of profiling, but Ferguson argues that the predictive policing context means this 'new form' of profiling could change IV analysis. The premise behind such an argument lies in the fact that a prediction made by some algorithm about potential high risk of crime in a certain area, could be taken in conjunction observations of ordinarily innocuous events. Read in the totality of circumstances, these two threads may justify individual reasonable suspicion . For example, a man looking into cars at a parking lot may not by itself justify reasonable suspicion, but taken together with a prediction of high risk of car theft at that locality, it may well justify reasonable suspicion. It is this impact of predictive policing, which influences the analysis of reasonable suspicion in a totality of circumstances that may represent new implications for courts looking at IV amendment protections.
Profiling, Predictive Policing and Discrimination
The above sections have already brought up the point that law enforcement agencies already utilize profiling methods in their operations. Also, as the sections on how predictive analytics works and on methods of predictive policing make clear, predictive policing definitely incorporates the development of profiles for predicting future criminal activity. Concerns about predictive models generate potentially discriminatory predictions therefore are very serious, and need addressing. Potential discrimination may be either overt, though far less likely, or unintended. A valuable case study of which sheds light on such discriminatory data mining practices can be found in US Labour law. It was shown how predictive models could be discriminatory at various stages, from conceptualising the model and training it with training data, to eventually selecting inappropriate features to search for . It is also possible for data scientists to (intentionally or not) use proxies for identifiers like race, income level, health condition and religion. Barocas and Selbst argue that "the current distribution of relevant attributes-attributes that can and should be taken into consideration in apportioning opportunities fairly-are demonstrably correlated with sensitive attributes" . Hence, what may result is unintended discrimination, as predictive models and their subjective and implicit biases are reflected in predicted decisions, or that the discrimination is not even accounted for in the first place. While I have not found any case law where courts have examined such situations in a criminal context, at the very least, law enforcement agencies need to be aware of these possibilities and guard against any forms of discriminatory profiling.
However, Ferguson argues that "the precision of the technology may in fact provide more protection for citizens in broadly defined high crime areas" . This is because the label of a 'high-crime area' may no longer apply to large areas but instead to very specific areas of criminal activity. This implies that previously defined areas of high crime, like entire neighbourhoods may not be scrutinised in such detail. Instead, police now may be more precise in locating and policing areas of high crime, such as an individual street corner or a particular block of flats instead of an entire locality.
Courts have also considered the existence of notoriously 'high-crime areas as part of considering reasonable suspicion. This was seen in Illinois v. Wardlow , where the "high crime nature of an area can be considered in evaluating the officer's objective suspicion". Many cases have since applied this reasoning without scrutinising the predictive value of such a label. In fact, Ferguson asserts that such labelling has questionable evidential value. He uses the facts of the Wardlow case itself to challenge the 'high crime area' factor. Ferguson cites the reasoning of one of the judges in the case:
"While the area in question-Chicago's District 11-was a low-income area known for violent crimes, how that information factored into a predictive judgment about a man holding a bag in the afternoon is not immediately clear."
Especially because "the most basic models of predictive policing rely on past crimes", it is likely that the predictive policing methods like hot spot or spatiotemporal analysis and risk terrain modelling may help to gather or build data models about high crime areas. Furthermore, the mathematical rigour of the predictive modelling could help clarify the term 'high crime area'. As Ferguson argues, "courts may no longer need to rely on the generalized high crime area terminology when more particularized and more relevant information is available" .
Ferguson synthesises four themes to which encapsulate reasonable suspicion analysis:
- Predictive information is not enough on its own. Instead, it is "considered relevant to the totality of circumstances, but must be corroborated by direct police observation".
- The prediction must also "be particularized to a person, a profile, or a place, in a way that directly connects the suspected crime to the suspected person, profile, or place".
- It must also be detailed enough to distinguish a person or place from others not the focus of the prediction .
- Finally, predicted information becomes less valuable over time. Hence it must be acted on quickly or be lost .
Conclusions from America
The main conclusion to draw from the analysis of the parallels between existing predictions in IV amendment law and predictive policing is that "predictive policing will impact the reasonable suspicion calculus by becoming a factor within the totality of circumstances test". Naturally, it reaffirms the imperative for predictive techniques to collect reliable data  and analyse it transparently. Moreover, in order for courts to evaluate the reliability of the data and the processes used (since predictive methods become part of the reasonable suspicion calculus), courts need to be able to analyse the predictive process. This has implications for the how hearings may be conducted, for how legal adjudicators may require training and many more. Another important concern is that the model of predictive information and police corroboration or direct observation may mean that in areas which were predicted to have low risk of crime, the reasonable suspicion doctrine works against law enforcement. There may be less effort paid to patrolling these other areas as a result of predictions.
Implications for India
While there have been no cases directly involving predictive policing methods, it would be prudent to examine the parts of Indian law which would inform the calculus on the lawfulness of using predictive policing methods. A useful lens to examine this might be found in the observation that prediction is not in itself a novel concept in justice, and is already used by courts and law enforcement in numerous circumstances.
Criminal Procedure in Non-Warrant Contexts
The most logical way to begin analysing the legal implications of predictive policing in India may probably involve identifying parallels between American and Indian criminal procedure, specifically searching for instances where 'reasonable suspicion' or some analogous requirement exists for justifying police searches.
In non-warrant scenarios, we find conditions for officers to conduct such a warrantless search in Section 165 of the Criminal Procedure Code (Cr PC). For clarity purposes I have stated section 165 (1) in full:
"Whenever an officer in charge of a police station or a police officer making an investigation has reasonable grounds for believing that anything necessary for the purposes of an investigation into any offence which he is authorised to investigate may be found in any place with the limits of the police station of which he is in charge, or to which he is attached, and that such thing cannot in his opinion be otherwise obtained without undue delay, such officer may, after recording in writing the grounds of his belief and specifying in such writing, so far as possible, the thing for which search is to be made, search, or cause search to be made, for such thing in any place within the limits of such station." 
However, India differs from the USA in that its Cr PC allows for police to arrest individuals without a warrant as well. As observed in Gulab Chand Upadhyaya vs State Of U.P, "Section 41 Cr PC gives the power to the police to arrest without warrant in cognizable offences, in cases enumerated in that Section. One such case is of receipt of a 'reasonable complaint' or 'credible information' or 'reasonable suspicion'"  Like above, I have stated section 41 (1) and subsection (a) in full:
"41. When police may arrest without warrant.
(1) Any police officer may without an order from a Magistrate and without a warrant, arrest any person-
(a) who has been concerned in any cognizable offence, or against whom a reasonable complaint has been made, or credible information has been received, or a reasonable suspicion exists, of his having been so concerned"
In analysing the above sections of Indian criminal procedure from a predictive policing angle, one may find both similarities and differences between the proposed American approach and possible Indian approaches to interpreting or incorporating predictive policing evidence.
Similarity of 'reasonable suspicion' requirement
For one, the requirement for "reasonable grounds" or "reasonable suspicion" seems to be analogous to the American doctrine of reasonable suspicion. This suggests that the concepts used in forming reasonable suspicion, for the police to "be able to point to specific and articulable facts which, taken together with rational inferences from those facts, reasonably warrant that intrusion" may also be useful in the Indian context.
One case which sheds light on an Indian interpretation of reasonable suspicion or grounds is State of Punjab v. Balbir Singh. In that case, the court observes a requirement for "reason to believe that such an offence under Chapter IV has been committed and, therefore, an arrest or search was necessary as contemplated under these provisions" in the context of Section 41 and 42 in The Narcotic Drugs and Psychotropic Substances Act, 1985. In examining the requirement of having "reason to believe", the court draws on Partap Singh (Dr) v. Director of Enforcement, Foreign Exchange Regulation Act, where the judge observed that "the expression 'reason to believe' is not synonymous with subjective satisfaction of the officer. The belief must be held in good faith; it cannot be merely a pretence….."
In light of this, the judge in Balbir Singh remarked that "whether there was such reason to believe and whether the officer empowered acted in a bona fide manner, depends upon the facts and circumstances of the case and will have a bearing in appreciation of the evidence" . The standard considered by the court in Balbir Singh and Partap Singh is different from the 'reasonable suspicion' or 'reasonable grounds' standard as per Section 41 and 165 of Cr PC. But I think the discussion can help to inform our analysis of the idea of reasonableness in law enforcement actions. Of importance was the court requirement of something more than mere "pretence" as well as a belief held in good faith. This could suggest that in fact the reasoning in American jurisprudence about reasonable suspicion might be at least somewhat similar to how Indian courts view reasonable suspicion or grounds in the context of predictive policing, and therefore how we could similarly conjecture that predictive evidence could form part of the reasonable suspicion calculus in India as well.
Difference in judicial treatment of illegally obtained evidence - Indian lack of exclusionary rules
However, the apparent similarity of how police in America and India may act in non-warrant situations - guided by the idea of reasonable suspicion - is only veneered by linguistic parallels. Despite the existence of such conditions which govern the searches without a warrant, I believe that Indian courts currently may provide far less protection against unlawful use of predictive technologies. The main premise behind this argument is that Indian courts refuse to exclude evidence that was obtained in breaches of the conditions of sections of the Cr PC. What exists in place of evidentiary safeguards is a line of cases in which courts routinely admit unlawfully or illegally obtained evidence. Without protections against unlawfully gathered evidence being considered relevant by courts, any regulations on search or conditions to be met before a search is lawful become ineffective. Evidence may simply enter the courtroom through a backdoor.
In the USA, this is by and large, not the case. Although there are exceptions to these rules, exclusionary rules are set out to prevent admission of evidence which violates the constitution. "The exclusionary rule applies to evidence gained from an unreasonable search or seizure in violation of the Fourth Amendment ". Mapp v. Ohio  set the precedent for excluding unconstitutionally gathered evidence, where the court ruled that "all evidence obtained by searches and seizures in violation of the Federal Constitution is inadmissible in a criminal trial in a state court" .
Any such evidence which then leads law enforcement to collect new information may also be excluded, as part of the "fruit of the poisonous tree" doctrine, established in Silverthorne Lumber Co. v. United States . The doctrine is a metaphor which suggests that if the source of certain evidence is tainted, so is 'fruit' or derivatives from that unconstitutional evidence. One such application was in Beck v. Ohio, where the courts overturned a petitioner's conviction because the evidence used to convict him was obtained via an unlawful arrest.
However in India's context, there is very little protection against the admission and use of unlawfully gathered evidence. In fact, there are a line of cases which lay out the extent of consideration given to unlawfully gathered evidence - both cases that specifically deal with the rules as per the Indian Cr PC as well as cases from other contexts - which follow and develop this line of reasoning of allowing illegally obtained evidence.
One case to pay attention to is State of Maharastra v. Natwarlal Damodardas Soni - in this case, the Anti-Corruption Bureau searched the house of the accused after receiving certain information as a tip. The police "had powers under the Code of Criminal Procedure to search and seize this gold if they had reason to believe that a cognizable offence had been committed in respect thereof". Justice Sarkaria, in delivering his judgement, observed that for argument's sake, even if the search was illegal, "then also, it will not affect the validity of the seizure and further investigation". The judge drew reasoning from Radhakishan v. State of U.P. This which was a case involving a postman who had certain postal items that were undelivered recovered from his house. As the judge in Radhakishan noted:
"So far as the alleged illegality of the search is concerned, it is sufficient to say that even assuming that the search was illegal the seizure of the articles is not vitiated. It may be that where the provisions of Sections 103 and 165 of the Code of Criminal Procedure, are contravened the search could be resisted by the person whose premises are sought to be searched. It may also be that because of the illegality of the search the court may be inclined to examine carefully the evidence regarding the seizure. But beyond these two consequences no further consequence ensues." 
Shyam Lal Sharma v. State of M.P. was also drawn upon, where it was held that "even if the search is illegal being in contravention with the requirements of Section 165 of the Criminal Procedure Code, 1898, that provision ceases to have any application to the subsequent steps in the investigation".
Even in Gulab Chand Upadhyay, mentioned above, the presiding judge contended that even "if arrest is made, it does not require any, much less strong, reasons to be recorded or reported by the police. Thus so long as the information or suspicion of cognizable offence is "reasonable" or "credible", the police officer is not accountable for the discretion of arresting or no arresting".
A more complete articulation of the receptiveness of Indian courts to admit illegally gathered evidence can be seen in the aforementioned Balbir Singh. The judgement aimed to:
"dispose of one of the contentions that failure to comply with the provisions of Cr PC in respect of search and seizure even up to that stage would also vitiate the trial. This aspect has been considered in a number of cases and it has been held that the violation of the provisions particularly that of Sections 100, 102, 103 or 165 Cr PC strictly per se does not vitiate the prosecution case. If there is such violation, what the courts have to see is whether any prejudice was caused to the accused and in appreciating the evidence and other relevant factors, the courts should bear in mind that there was such a violation and from that point of view evaluate the evidence on record."
The judges then consulted a series of authorities on the failure to comply with provisions of the Cr PC:
- State of Punjab v. Wassan Singh: "irregularity in a search cannot vitiate the seizure of the articles".
- Sunder Singh v. State of U.P: 'irregularity cannot vitiate the trial unless the accused has been prejudiced by the defect and it is also held that if reliable local witnesses are not available the search would not be vitiated."
- Matajog Dobey v.H.C. Bhari: "when the salutory provisions have not been complied with, it may, however, affect the weight of the evidence in support of the search or may furnish a reason for disbelieving the evidence produced by the prosecution unless the prosecution properly explains such circumstance which made it impossible for it to comply with these provisions."
- R v. Sang: "reiterated the same principle that if evidence was admissible it matters not how it was obtained." Lord Diplock, one of the Lords adjudicating the case, observed that "however much the judge may dislike the way in which a particular piece of evidence was obtained before proceedings were commenced, if it is admissible evidence probative of the accused's guilt "it is no part of his judicial function to exclude it for this reason".  As the judge in Balbir Singh quoted from Lord Diplock, a judge "has no discretion to refuse to admit relevant admissible evidence on the ground that it was obtained by improper or unfair means. The court is not concerned with how it was obtained."
The vast body of case law presented above provides observers with a clear image of the courts willingness to admit and consider illegally obtained evidence. The lack of safeguards against admission of unlawful evidence are important from the standpoint of preventing the excessive or unlawful use of predictive policing methods. The affronts to justice and privacy, as well as the risks of profiling, seem to become magnified when law enforcement use predictive methods more than just to augment their policing techniques but to replace some of them. The efficacy and expediency offered by using predictive policing needs to be balanced against the competing interest of ensuring rule of law and due process. In the Indian context, it seems courts sparsely consider this competing interest.
Naturally, weighing in on which approach is better depends on a multitude of criteria like context, practicality, societal norms and many more. It also draws on existing debates in administrative law about the role of courts, which may emphasise protecting individuals and preventing excessive state power (red light theory) or emphasise efficiency in the governing process with courts assisting the state to achieve policy objectives (green light theory) .
A practical response may be that India should aim to embrace both elements and balance them appropriately, although what an appropriate balance again may vary. There are some who claim that this balance already exists in India. Evidence for such a claim may come from R.M. Malkani v. State of Maharashtra, where the court considered whether an illegally tape-recorded conversation could be admissible. In its reasoning, the court drew from Kuruma, Son of Kanju v. R. , noting that
" if evidence was admissible it matters not how it was obtained. There is of course always a word of caution. It is that the Judge has a discretion to disallow evidence in a criminal case if the strict rules of admissibility would operate unfairly against the accused. That caution is the golden rule in criminal jurisprudence".
While this discretion exists at least principally in India, in practice the cases presented above show that judges rarely exercise that discretion to prevent or bar the admission of illegally obtained evidence or evidence that was obtained in a manner that infringed the provisions governing search or arrest in the Cr PC. Indeed, the concern is that perhaps the necessary safeguards required to keep law enforcement practices, including predictive policing techniques, in check would be better served by a greater focus on reconsidering the legality of unlawfully gathered evidence. If not, evidence which should otherwise be inadmissible may find its way into consideration by existing legal backdoors.
Risk of discriminatory predictive analysis
Regarding the risk of discriminatory profiling, Article 15 of India's Constitution states that "the State shall not discriminate against any citizen on grounds only of religion, race, caste, sex, place of birth or any of them" . The existence of constitutional protection for such forms of discrimination suggests that India will be able to guard against discriminatory predictive policing. However, as mentioned before, predictive analytics often discriminates institutionally, "whereby unconscious implicit biases and inertia within society's institutions account for a large part of the disparate effects observed, rather than intentional choices". As in most jurisdictions, preventing these forms of discrimination are much harder. Especially in a jurisdiction whose courts are already receptive to allowing admission of illegally obtained evidence, the risk of discriminatory data mining or prejudiced algorithms being used by police becomes magnified. Because the discrimination may be unintentional, it may be even harder for evidence from discriminatory predictive methods to be scrutinised or when applicable, dismissed by the courts.
Conclusion for India
One thing which is eminently clear from the analysis of possible interpretations of predictive evidence is that Indian Courts have had no experience with any predictive policing cases, because the technology itself is still at a nascent stage. There is in fact a long way to go before predictive policing will become used on a scale similar to that of USA for example.
But, even in places where predictive policing is used much more prominently, there is no precedent to observe how courts may view predictive policing. Ferguson's method of locating analogous situations to predictive policing which courts have already considered is one notable approach, but even this does not provide complete answer. One of his main conclusions that predictive policing will affect the reasonable suspicion calculus, or in India's case, contribute to 'reasonable grounds' in some ways, is perhaps the most valid one.
However, what provides more cause for concern in India's context are the limited protections against use of unlawfully gathered evidence. The lack of 'exclusionary rules' unlike those present in the US amplifies the various risks of predictive policing because individuals have little means of redress in such situations where predictive policing may be used unjustly against them.
Yet, the promise of predictive policing remains undeniably attractive for India. The successes predictive policing methods seem to have had In the US and UK coupled with the more efficient allocation of law enforcement's resources as a consequence of adapting predictive policing evidence this point. The government recognises this and seems to be laying the foundation and basic digital infrastructure required to utilize predictive policing optimally. One ought also to ask whether it is the even within the court's purview to decide what kind of policing methods are to be permissible through evaluating the nature of evidence. There is a case to be made for the legislative arm of the state to provide direction on how predictive policing is to be used in India. Perhaps the law must also evolve with the changes in technology, especially if courts are to scrutinise the predictive policing methods themselves.
 Joh, Elizabeth E. "Policing by Numbers: Big Data and the Fourth Amendment." SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, February 1, 2014. http://papers.ssrn.com/abstract=2403028.
 Tene, Omer, and Jules Polonetsky. "Big Data for All: Privacy and User Control in the Age of Analytics." Northwestern Journal of Technology and Intellectual Property 11, no. 5 (April 17, 2013): 239.
 Datta, Rajbir Singh. "Predictive Analytics: The Use and Constitutionality of Technology in Combating Homegrown Terrorist Threats." SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, May 1, 2013. http://papers.ssrn.com/abstract=2320160.
 Johnson, Jeffrey Alan. "Ethics of Data Mining and Predictive Analytics in Higher Education." SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, May 8, 2013. http://papers.ssrn.com/abstract=2156058.
 Duhigg, Charles. "How Companies Learn Your Secrets." The New York Times, February 16, 2012. http://www.nytimes.com/2012/02/19/magazine/shopping-habits.html.
 Lijaya, A, M Pranav, P B Sarath Babu, and V R Nithin. "Predicting Movie Success Based on IMDB Data." International Journal of Data Mining Techniques and Applications 3 (June 2014): 365-68.
 Johnson, Jeffrey Alan. "Ethics of Data Mining and Predictive Analytics in Higher Education." SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, May 8, 2013. http://papers.ssrn.com/abstract=2156058.
 Sangvinatsos, Antonios A. "Explanatory and Predictive Analysis of Corporate Bond Indices Returns." SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, June 1, 2005. http://papers.ssrn.com/abstract=891641.
 Barocas, Solon, and Andrew D. Selbst. "Big Data's Disparate Impact." SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, February 13, 2015. http://papers.ssrn.com/abstract=2477899.
 Joh, supra note 1.
 US Environmental Protection Agency. "How We Use Data in the Mid-Atlantic Region." US EPA. Accessed November 6, 2015. http://archive.epa.gov/reg3esd1/data/web/html/.
 Joh, supra note 1, at pg 48.
 Perry, Walter L., Brian McInnis, Carter C. Price, Susan Smith and John S. Hollywood. Predictive Policing: The Role of Crime Forecasting in Law Enforcement Operations. Santa Monica, CA: RAND Corporation, 2013. http://www.rand.org/pubs/research_reports/RR233. Also available in print form.
 Ibid, at pg 2.
 Chan, Sewell. "Why Did Crime Fall in New York City?" City Room. Accessed November 6, 2015. http://cityroom.blogs.nytimes.com/2007/08/13/why-did-crime-fall-in-new-york-city/.
 Bureau of Justice Assistance. "COMPSTAT: ITS ORIGINS, EVOLUTION, AND FUTURE IN LAW ENFORCEMENT AGENCIES," 2013. http://www.policeforum.org/assets/docs/Free_Online_Documents/Compstat/compstat%20-%20its%20origins%20evolution%20and%20future%20in%20law%20enforcement%20agencies%202013.pdf.
 1996 internal NYPD article "Managing for Results: Building a Police Organization that Dramatically Reduces Crime, Disorder, and Fear."
 Bratton, William. "Crime by the Numbers." The New York Times, February 17, 2010. http://www.nytimes.com/2010/02/17/opinion/17bratton.html.
 RAND CORP, supra note 16.
 RAND CORP, supra note 16, at pg 19.
 Joh, supra note 1, at pg 44.
 RAND CORP, supra note 16, pg 38.
 RAND CORP, supra note 16, at pg 39.
 RAND CORP, supra note 16, at pg 41.
 Data-Smart City Solutions. "Dr. George Mohler: Mathematician and Crime Fighter." Data-Smart City Solutions, May 8, 2013. http://datasmart.ash.harvard.edu/news/article/dr.-george-mohler-mathematician-and-crime-fighter-166.
 RAND CORP, supra note 16, at pg 44.
 Joh, supra note 1, at pg 45.
 Ouellette, Danielle. "Dispatch - A Hot Spots Experiment: Sacramento Police Department," June 2012. http://cops.usdoj.gov/html/dispatch/06-2012/hot-spots-and-sacramento-pd.asp.
 Pitney Bowes Business Insight. "The Safer Derbyshire Partnership." Derbyshire, 2013. http://www.mapinfo.com/wp-content/uploads/2013/05/safer-derbyshire-casestudy.pdf.
 Daniel B Neill, Wilpen L. Gorr. "Detecting and Preventing Emerging Epidemics of Crime," 2007.
 RAND CORP, supra note 16, at pg 33.
 Joh, supra note 1, at pg 46.
 Paul, Jeffery S, and Thomas M. Joiner. "Integration of Centralized Intelligence with Geographic Information Systems: A Countywide Initiative." Geography and Public Safety 3, no. 1 (October 2011): 5-7.
 Mohler, supra note 30.
 Moses, B., Lyria, & Chan, J. (2014). Using Big Data for Legal and Law Enforcement
Decisions: Testing the New Tools (SSRN Scholarly Paper No. ID 2513564). Rochester, NY: Social Science Research Network. Retrieved from http://papers.ssrn.com/abstract=2513564
 Gorner, Jeremy. "Chicago Police Use Heat List as Strategy to Prevent Violence." Chicago Tribune. August 21, 2013. http://articles.chicagotribune.com/2013-08-21/news/ct-met-heat-list-20130821_1_chicago-police-commander-andrew-papachristos-heat-list.
 Stroud, Matt. "The Minority Report: Chicago's New Police Computer Predicts Crimes, but Is It Racist?" The Verge. Accessed November 13, 2015. http://www.theverge.com/2014/2/19/5419854/the-minority-report-this-computer-predicts-crime-but-is-it-racist.
 Moser, Whet. "The Small Social Networks at the Heart of Chicago Violence." Chicago Magazine, December 9, 2013. http://www.chicagomag.com/city-life/December-2013/The-Small-Social-Networks-at-the-Heart-of-Chicago-Violence/.
 Lester, Aaron. "Police Clicking into Crimes Using New Software." Boston Globe, March 18, 2013. https://www.bostonglobe.com/business/2013/03/17/police-intelligence-one-click-away/DzzDbrwdiNkjNMA1159ybM/story.html.
 Stanley, Jay. "Chicago Police 'Heat List' Renews Old Fears About Government Flagging and Tagging." American Civil Liberties Union, February 25, 2014. https://www.aclu.org/blog/chicago-police-heat-list-renews-old-fears-about-government-flagging-and-tagging.
 Rieke, Aaron, David Robinson, and Harlan Yu. "Civil Rights, Big Data, and Our Algorithmic Future," September 2014. https://bigdata.fairness.io/wp-content/uploads/2015/04/2015-04-20-Civil-Rights-Big-Data-and-Our-Algorithmic-Future-v1.2.pdf.
 Edmond, Deepu Sebastian. "Jhakhand's Digital Leap." Indian Express, September 15, 2013. http://www.jhpolice.gov.in/news/jhakhands-digital-leap-indian-express-15092013-18219-1379316969.
 Jharkhand Police. "Jharkhand Police IT Vision 2020 - Effective Shared Open E-Governance." 2012. http://jhpolice.gov.in/vision2020. See slide 2
 Edmond, supra note 49.
 Edmond, supra note 49.
 Kumar, Raj. "Enter, the Future of Policing - Cops to Team up with IIM Analysts to Predict & Prevent Incidents." The Telegraph. August 28, 2012. http://www.telegraphindia.com/1120828/jsp/jharkhand/story_15905662.jsp#.VkXwxvnhDWK.
 See supra note 49.
 Lavanya Gupta, and Selva Priya. "Predicting Crime Rates for Predictive Policing." Gandhian Young Technological Innovation Award, December 29, 2014. http://gyti.techpedia.in/project-detail/predicting-crime-rates-for-predictive-policing/3545.
 Gupta, Lavanya. "Minority Report: Minority Report." Accessed November 13, 2015. http://cmuws2014.blogspot.in/2015/01/minority-report.html.
 See supra note 59.
 India, Press Trust of. "Police Science Congress in Gujarat to Have DRDO Exhibition." Business Standard India, March 10, 2015. http://www.business-standard.com/article/pti-stories/police-science-congress-in-gujarat-to-have-drdo-exhibition-115031001310_1.html.
 National Crime Records Bureau. "About Crime and Criminal Tracking Network & Systems - CCTNS." Accessed November 13, 2015. http://ncrb.gov.in/cctns.htm.
 Ibid. (See index page)
 See supra note 1, at pg 60.
 See supra note 1, at pg 60.
 Villasenor, John. "What You Need to Know about the Third-Party Doctrine." The Atlantic, December 30, 2013. http://www.theatlantic.com/technology/archive/2013/12/what-you-need-to-know-about-the-third-party-doctrine/282721/.
 Newell, Bryce Clayton. "Local Law Enforcement Jumps on the Big Data Bandwagon: Automated License Plate Recognition Systems, Information Privacy, and Access to Government Information." SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, October 16, 2013. http://papers.ssrn.com/abstract=2341182, at pg 24.
 See supra note 72.
 Dahyabhai Chhaganbhai Thakker vs State Of Gujarat, 1964 AIR 1563
 See supra note 16.
 See supra note 66.
 Ferguson, Andrew Guthrie. "Big Data and Predictive Reasonable Suspicion." SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, April 4, 2014. http://papers.ssrn.com/abstract=2394683, at pg 287. See also supra note 79.
 See supra note 80.
 See supra note 80.
 See supra note 80.
 See supra note 80, at pg 289.
 See supra note 80, at pg 291.
 See supra note 80, at pg 293.
 See supra note 80, at pg 308.
 Larissa Cespedes-Yaffar, Shayona Dhanak, and Amy Stephenson. "U.S. v. Mendenhall, U.S. v. Sokolow, and the Drug Courier Profile Evidence Controversy." Accessed July 6, 2015. http://courses2.cit.cornell.edu/sociallaw/student_projects/drugcourier.html.
 See supra note 80, at pg 295.
 See supra note 80, at pg 297.
 See supra note 80, at pg 308.
 See supra note 80, at pg 310.
 See supra note 11.
 See supra note 11.
 See supra note 80, at pg 303.
 See supra note 80, at pg 300.
 See supra note 80, at pg 301.
 See supra note 1, at pg 42.
 See supra note 80, at pg 303.
 See supra note 80, at pg 303.
 See supra note 80, at pg 312.
 See supra note 80, at pg 317.
 See supra note 80, at pg 319.
 See supra note 80, at pg 321.
 Gulab Chand Upadhyaya vs State Of U.P, 2002 CriLJ 2907
 Section 41 Indian Criminal Procedure Code
 See supra note 79
 State of Punjab v. Balbir Singh. (1994) 3 SCC 299
 Partap Singh (Dr) v. Director of Enforcement, Foreign Exchange Regulation Act. (1985) 3 SCC 72 : 1985 SCC (Cri) 312 : 1985 SCC (Tax) 352 : AIR 1985 SC 989
 Ibid, at SCC pg 77-78.
 See supra note 121, at pg 313.
 Carlson, Mr David. "Exclusionary Rule." LII / Legal Information Institute, June 10, 2009. https://www.law.cornell.edu/wex/exclusionary_rule.
 Busby, John C. "Fruit of the Poisonous Tree." LII / Legal Information Institute, September 21, 2009. https://www.law.cornell.edu/wex/fruit_of_the_poisonous_tree.
 State of Maharashtra v. Natwarlal Damodardas Soni, (1980) 4 SCC 669, at 673.
 Radhakishan v. State of U.P. [AIR 1963 SC 822 : 1963 Supp 1 SCR 408, 411, 412 : (1963) 1 Cri LJ 809]
 Ibid, at SCR pg 411-12.
 Shyam Lal Sharma v. State of M.P. (1972) 1 SCC 764 : 1974 SCC (Cri) 470 : AIR 1972 SC 886
 See supra note 135, at page 674.
 See supra note 119, at para. 10.
 See supra note 121, at pg 309.
 State of Punjab v. Wassan Singh, (1981) 2 SCC 1 : 1981 SCC (Cri) 292
 See supra note 121, at pg 309.
 Sunder Singh v. State of U.P, AIR 1956 SC 411 : 1956 Cri LJ 801
 See supra note 121, at pg 309.
 Matajog Dobey v.H.C. Bhari, AIR 1956 SC 44 : (1955) 2 SCR 925 : 1956 Cri LJ 140
 See supra note 121, at pg 309.
 R v. Sang, (1979) 2 All ER 1222, 1230-31
 See supra note 121, at pg 309.
 Harlow, Carol, and Richard Rawlings. Law and Administration. 3rd ed. Law in Context. Cambridge University Press, 2009.
 R.M. Malkani v. State of Maharashtra, (1973) 1 SCC 471
 Kuruma, Son of Kanju v. R., (1955) AC 197
 See supra note 154, at 477.
 See supra note 11.