Algorithm Governance Framework for Media Regulation in Nigerian Media System

The world of media and information, from the time internet became popular, has been in constant rapid changes to such an extent that older models of conventional media system are being challenged if not replaced. One of the changes being experienced in the contemporary media environment is the use of computer codes or algorithms to perform gatekeeping functions that used to be done solely by human agents. This paper reviews the state and challenges of media regulations and the use of algorithms in Nigerian media system. The review showed a gap in media regulation in Nigeria where journalism is compartmentalised in contrast to media convergence and that algorithmic-based journalism may not be effectively regulated. To close this gap, the study used risk-based analysis as a theoretical framework and library research as method to design a framework for algorithmic media governance in the country. The result is the Converged Media Governance Framework for Algorithmic and Mixed Journalism. The framework reconceptualizes the way and manner media regulation is framed and organised as involving only human agents and alsothat journalism in the country should not be compartmentalised in the light of media convergence. The paper recommended the adoption of the framework by stakeholders in the Nigerian media system.


INTRODUCTION
In the second week of December 2018, a story broke out from the United States Congress investigation on social media misuse during their own last general election. A congressman asked a CEO why is it that if the word 'idiot' is searched on google image (or photo) search, the picture of President Donald Trump appears. Also, negative news stories appear whenever 'health care bill' is searched. Google chief executive Sundar Pachai was asked the reason for this anomaly. His response is that google search engine is not done by human beings (it is done by algorithm) and that it brings out result based 200 parameters including relevance and popularity. Because of that there is little or nothing they can do about it. Facebook had exhibited similar issue when fake news posts are directed or targeted to particular individuals or groups obviously to influence their views on some issues. Still Facebook officials argue that it is not personal or done by their staff but rather selected by their algorithm. How true is it that there is nothing they can do and as such, the world is at the mercy of algorithm?
Analysts in the computer world have predicted that a lot of the activities that human beings used to do manually are likely to be done by computerized machines. This is largely due to advancement in artificial intelligence, big data analytics, mobile telephony and IT in general. Many activities have moved online as internet penetration is becoming more of a necessity of 21 st century life. Not only are financial and business services conducted largely online but nowadays many domains of human activities are moving online. This massive shift to online environment has a number of implications for individuals organisations and the society.
In this current age of internet, the importance of data cannot be over emphasized and data which has been described as the 'new gold' of the 21 st century. WEF 2011 is necessary for innovation and economic success as London Economics 2010. To manipulate our data to make money, these online platforms create and use powerful algorithms behind the scene and for searching,

| T h e J o u r n a l o f S o c i e t y and M e d i a 4 ( 1 )
aggregation, surveillance, forecast, filtering, recommendations, scoring, content production and allocation (Saurwein, Just, and Latzer 2015). These algorithms have taken the place of earlier less intrusive physical, mechanical and electronic technologies (Manovich 2014) which relied on human decision making and therefore have no governing powers in themselves. So, instead of some workers sitting behind a desk to monitor, approve, control, manage, etc it is machines that do these (Lazer, Kennedy, King, and Vespignani 2014) identified the various types of algorithms based on their functions and their common examples (see Table 1).
While the above are generalised domains, algorithm journalism, also called software generated journalism, robot journalism, artificial intelligence news, software generated news, automated journalism, bot driven or data driven journalism have widely diffused in the field of journalism especially in north America and Europe (Lindén 2017) and thereby performing secondary agenda setting and gatekeeping roles Bakker 2012 (Just and Latzer 2017). Associated Press, Forbes, Los Angeles Times and Pro Publica are among the popular media organisations using algorithmic journalism (Graefe 2016). Accordingly, algorithm journalism which widely deployed in sports and business news domains is deployed either as sole (automated) agents or mixed human computer journalism.
Algorithms is not only diffusing into the newsrooms but according to Linden, is having disruptive effect on the normative foundations of journalism. One of the areas that algorithm is disrupting is the regulatory framework with which journalism is governed. B a s h i r : A l g o r i t h m G o v e r n a n c e F r a m e w o r k f o r M e d i a R e g u l a t i o n i n N i g e r i a n M e d i a S y s t e m | 183 Table 1 Types Example

Functional Typology of Algorithmic Selection Applications
Looking at the functions that algorithms perform and the power they are amassing, many analysts are questioning and even challenging their power over our lives and our professions. We have seen how Facebook algorithm was used to manipulate message British citizens were exposed to social media during the Brexit debate, the last United States general election and the Nigerian 2015 general elections. There were lots of manipulations and trending fake news because some third parties were able to access and tweak algorithms of various social media platforms. Criminal hackers too can take advantage of these algorithms by hacking into them to inflict damages to users.
While the concerns about the enormous power of algorithms is well known across the globe especially as result of the rise of populism, fake news, hate speech and numerous abuses of users' data on numerous platforms like Google and Facebook, the regulatory response to such misuse andabuses especially in developing countries are not well articulated to suit the current challenge. The conventional mass media in mostcountries is regulated through brick and mortar regulatory mechanisms that works on the assumption that human agency (as individuals and organisations) alone are the actors. However, with algorithmic journalism, computer algorithms (or computer software) are increasing playing the role of journalists or complementing them.
According to (Apuke 2016) ICTs has taken roots in Nigerian mass sector in present day Nigerian journalism with many organisations using computer and computer networks. Large number of Nigerian media are also using internet with content and activities done online (Talabi 2011). The adoption of ICTs by the media is a prelude to the introduction of algorithmic journalism in every society.
In Nigeria the earliest known use of algorithm or robot to generate news stories in public domain was when Fintel Envoya botwas developed in 2018 by an organization known as Orodata Science for financial information reporting using structured data collected by the Nigerian Inter-Bank Settlement System (Aboh 2020). The bot generates short news stories on payment channels and banking sector credits. The same organization also developed an algorithm called Arria to generate hundreds of stories from structured election data in Nigeria.
Another use of algorithm or robot for information in a quasi-journalistic way in Nigeria by BudgiT, an organisation aimed at promoting financial accountability in public domain. The organisation autogenerate some reports adding graphics and interactive features to make government budgets in Nigeria 'readable' to various segments of the public. With two million views and 275,000 unique visitors as at 2014, BudgiT also works with journalists so thereby making their contents accessible to the public (Onigbinde 2014).
While for now mainstream media organisations are yet to adopt algorithm journalism to write stories, the diffusionist tendencies of such technologies won't take long time to find their way into newsroom in the country. However, stories generated by algorithm in quasi-journalistic and non-journalistic entities find their way into news and hence the need for an effective framework for algorithmic governance (Ndlela 2020). Stated that in Africa algorithmsbots and trollsare deployed to influence political outcomes during elections which endangers democracy in the continent. One of the institutions which serve as a victim of negative is the mass media which are left with reduced credibility and relevance.
Some of the fake information generated by Cambridge Analytica, the infamous found their way into mainstream media in Nigeria during the 2015 general thus B a s h i r : A l g o r i t h m G o v e r n a n c e F r a m e w o r k f o r M e d i a R e g u l a t i o n i n N i g e r i a n M e d i a S y s t e m | 185 making it election to become one of the most divisive in the history of the country (Oparah 2015;Apuke 2018) also noted the use of computational algorithms and networks for propaganda especially during election campaign in many countries including Nigeria.
At a conference aimed at combating misinformation and disinformation, a Nigerian journalist in painting how challenging the situation is was quoted as saying: people tend to believe whatever they see on Facebook and Twitter more. And the tech guys such as Google, Twitter are not helping the matter. They created monsters they cannot regulate, you cannot control over 40 million people in Nigeria, you ask a robot to perform the same function. We have less than 300 radio stations in the country which are regulated but on Facebook we have over 40 million people dishing out whatever they like, whenever they like…. (Adenekan 2019).

In view of the use of algorithm by quasi-journalistic organisations in
Nigeria who also feed journalists in media organisations with information, there is the important need to identify regulatory gaps and proffer possible solutions. Such role, which may be minor for now, may likely grow as is seen in other developing countries. This is significant because algorithms can now be said to be potential future of journalism every including Nigeria. However, issues of regulations or governance are better dealt with normatively with each society having its own unique environment. The aim of this paper therefore is to examine the current status of algorithm governance in Nigerian media landscape and identify an appropriate framework for governing algorithms in journalistic domain.
Algorithms are man-made artefacts which originate or contribute to journalists' efforts. As such, the society must have measures with appropriate sanctions to control their use and misuse. Previous studies on media regulation in Nigeria focus on human dimensions of media regulation with no researches on algorithmic or human algorithmic journalism.
This study focuses on regulatory framework for algorithmic and humanalgorithmic (combined) regulation. Another, new dimension that have not been given focus is the separated organization of media regulation in the country with print media being separated with broadcast and online media. Technological convergence is making the media to become unified with multimedia contents. However, the performance of the council left much to be desired largely due to mistrust by stakeholders (Maho 2013). This has contributed to high level of unethical practices in Nigerian journalism especially corruption (Adeyemi 2013) and journalism being manned by many quacks (Talabi and Ogundeji 2012).
Overall scholars as stated below have argued that the media regulation in who commit "minor sins" rather than media owners who commit "mortal sins".
That is media proprietors are more likely to be major culprits but the laws and codes targets journalist. In addition, the laws and codes have their own internal inadequacies (procedures, remedies, etc). Therefore, with these regulatory challenges, how can and to what extent can algorithm journalism be governed in Nigeria?
It can be noted that the existing media regulatory framework in Nigeria privacy, unethical use, exploitation, access, etc (see Table 2). It can also be seen that in the age convergence media regulation is not connected with data protection laws with each operating without complementarity. Also, the two frameworks do not tackle the computer codes (algorithm) and how they are deployed to lead to bias, censorship, social discrimination, abuse of market power, effect on cognitive capabilities, violation of property rights, etc. Thus, if not governed effectively, algorithms can be programmed to work for the interest of the few and therefore in conflict with the interest of society. The fact that algorithms can be selfishly deployed with negative outcome simplies an appropriate form of regulation from within the industry and by authorities. Algorithms are computer formula designed to solve some problems following sequences of steps. They are the codes that make our computers, mobile phones, websites and other IT artefacts to work, thereby, enabling automation which has the advantage of speed, accuracy and saving human energy especially in routine, laborious and sometimes difficult tasks. The popularization of computing and internet has made algorithms to be widespread even though they operate behind the scene. journalism. Linden declared that algorithm journalism is capable of automating only certain routine tasks of news work, which the author called 'low level journalism' especially in domains like sport and finance. However, journalistic work that requires passion and creativity which for now is not amenable to automation. Using expert interview method, He found out that while computer experts are enthusiastic about computer journalism which see journalists leasing part of their work to computers, journalists are concerned about lack of relevant computational skills and it will affect their job satisfaction. Secondly, the result show that algorithm journalism can free journalists from some routine tasks (e.g writing structured stories, news selection and placement, scheduling, queuing, among others) and they can be smarter with good use of algorithm. Thus, hybrid form of human-machine journalism is recommended.
Linden therefore concluded that computational (algorithm) journalism is a reality that journalists have to deal with to add value to their work. However, there are ethical, moral, and operational considerations with this. For instance, algorithms can be manipulated by its creators or others. Hence the need for framework to regulate their use in journalism.

Other researchers like Wölker and Powell analyzed the perceived users'
credibility of automated and mixed (human-computer) written news stories among European audience. Using an experimental design of online readers in Europe, the study found out that there is no difference between automated journalism, human journalism and mixed one as far as message and source credibility is concerned in finance and sports news domain. The automated journalism stories actually score a little higher for message credibility in sports news which the researchers attributed to specific details provided which may not be easily recalled by human journalists. Despite the promising outcome for algorithmic journalism in the study, they are of the view that automated journalism cannot act as fourth estate of the realm (Wölker and Powell 2018). They therefore, argued along the line of (Stromback 2005) that journalism has a moral obligation to democracy which algorithms cannot fulfil. They therefore recommended mixed journalism.
Although algorithm journalism is in its infancy in Nigeria and many developing countries, the issue of credibility and moral obligation may likely be universal

| T h e J o u r n a l o f S o c i e t y and M e d i a 4 ( 1 )
with slight variations in details. Therefore, algorithm cannot replace human journalists and will still need to be regulated appropriately.
Looking at the reality of algorithm journalisms diffusion into many countries there is the need for appropriate regulatory framework to manage its However, the media convergence has made delimiting algorithm governance to only media domain as archaic and unrealistic. So, algorithm governance needs to be broad and cut across many domains that are similar to the mass media and need to be broad to include all actors (individual journalists, programmers, technicians, audience, etc. and organisations like service providers, marketers, host, public institutions, associations, and artefacts. it is the novelty of this research.

METHODS
As discussed above, algorithmic journalism has advantages but also comes with risks that can endanger society. According to Saurwein et al. 'from a public-interest point of view, governance should reinforce benefits and minimize risks. The benefits of algorithmic journalism include increased quantity of reporting, speed in content production and distribution, increased innovation, additional profits for media entrepreneurs, feeing journalists from drudgery of B a s h i r : A l g o r i t h m G o v e r n a n c e F r a m e w o r k f o r M e d i a R e g u l a t i o n i n N i g e r i a n M e d i a S y s t e m | 191 some work, more collaboration with computer experts, among many others (Verschuren and Hartog 2005). However, with the advantages also comes some risks: manipulation, censorship, fake news, low artistic creativity, inability to properly serve as fourth estate of the realm, tension in the relationship between journalists and machine, among many others. Therefore, this study adopts riskbased approach to govern algorithm in media system in Nigeria.
Risk-based regulation is a set of strategies or framework that entities wit regulatory powers adopt in order to focus on activities that pose risk their regulatory system to enable them achieve their objectives ( collective behavior and social order. For the mass media sector for instance, governance by algorithm, according to them, with the help of big data and machine learning, is more evidence-based and data driven. However, the dominance of private sector or market driven present risks that will need to be balanced with public policy considerations. As a result of the gap in media regulation of algorithmic journalism in Nigeria, this research fills this gap by proposing a framework for media regulation in a converged media environment using algorithm, human agents or mixed journalism. The study uses qualitative library research and generate the framework (both graphically and conceptually) for algorithm governance in Nigeria. Secondary sources of data used for the study were generated from existing literature and the framework was generated and presented conceptually and graphically.

RESULT AND DISCUSSION
In response to the regulatory challenges and the technical and environmental changes in the Nigerian media ecology, this paper recommends converged media regulatory framework to govern algorithm and mixed journalism. Converged framework is essential to take care of various journalistic

Converged Media Governance Framework for Algorithmic and Mixed Journalism
The conventional and online/multimedia organisations and freelancers.
The layer also accommodates bloggers, citizen journalists, social media influencers/content producers, bloggers, proprietors, management, algorithms, amongst others. The content producer in the layer is processed in the second layer using end user hardware and various software (including algorithms) and transmitted through infrastructure layer to public. The conventional print and broadcast technologies and telecommunication system (networks, transmitters, internet, servers, etc) are part of the second layer and as all the three layers are subject to regulations. Whether a content is produced by an individual, an algorithm or a mixture of human and algorithm, the author must be subject to regulation depending on the level of performance and liability. The user layer consists of individuals, groups, or communities receiving or consuming mediated messages. There is symmetric communication flow between the three layers sandwiched infrastructure/hardware/software layer.
On the right side of the framework is the rows and columns. The rows show the layers of regulatory mechanism which range from legal instruments, agencies, industry/market mechanisms, consumers (as groups and individuals) and the users of computer algorithms. In a converged media system, these layers of In addition, algorithmic technologies can be deployed by the regulatory agencies or industry players following industry sanctioned and legally laid down procedures which should not infringe freedom of expression and of the press. Government agencies in Nigeria are known to be weak in adopting innovation (Adomako and Danso 2014), so this measure will be more appropriate either as a market mechanism or industry self-regulation especially as it relate to online media. In the middle of the framework, vertical columns show that converged media organs and values of regulation which vertically cut across all the layers of regulation and media production consumption chain. The first three layers (all the tiers of government in the country, public and private organisations as well individuals and communities) can be involved in regulation in many ways. The fourth column depicts the values that makes regulations and strengthening it imperative. These values and norms include accountability, transparency, honesty, fairness, freedom, truth, independence, etc. Overall, the regulatory framework is encapsulated within the Nigerian social, economic, cultural, political and cultural environment and should be adapted to contribute towards the fulfilment societal goals and aspirations. The framework must go hand in hand with the needs and challenges of the environment to succeed.
The Converged Media Governance Framework for Nigeria therefore implies the need for adapting Nigeria's media regulation to contemporary media and societal changes. One of which is the recognition of human and non-human actors who are hitherto not well regulated in the current regulatory framework.
The prominence of convergence and algorithmic journalism is also necessary in B a s h i r : A l g o r i t h m G o v e r n a n c e F r a m e w o r k f o r M e d i a R e g u l a t i o n i n N i g e r i a n M e d i a S y s t e m | 195 the light of technology changes. In addition, algorithm is also recognized as an agency or tool for media regulation. However, it must be observed that the bottom line of media regulation is behavior of actors in the system. However good a framework is on paper, it will never succeed without understanding, support and cooperation of the actors in the system. This is part of the reason why the current regulatory framework is not having the desired effect.

CONCLUSION
Algorithm has come to stay in today's communication environment and is shaping a number of activities and processes in various domains one of which is journalism. It can also be observed that algorithms can be used beneficially to enhance journalistic production by extending the human capabilities of journalists and for better services and profits. They can also be used for negative purposes like fake news, violation of privacy, abuse of data, hate speech, and other ethical and legal breaches for which the present regulatory framework in Nigeria is not equipped to regulate. Because of their importance, each society need to critically evaluate how to govern algorithm to tap their potentials and minimize their risks which can be normative and based on case by case. For Nigeria, the study, therefore, recommend the adoption of the Converged Media Governance Framework to accomplish this. The framework recognized algorithmic and human journalism as legitimate and further looks at journalism in the country from three major perspective with different layers and vertical slants operating in the environment. The study therefore recommends that law makers in Nigeria review regulatory instruments to suit contemporary media challenges. Also, journalists in Nigeria, before the advents of massive use of algorithmic software need to be retrained to take advantage of algorithms in newsrooms and to avoid ethical pitfall in using them. Also, journalists in Nigeria need to expand the definition of journalist to incorporate media workers in new platforms like online platforms like blogs, citizen journalists, etc.