Co-design with Raspberry Pi: Developing and Hosting Sustainable Community Application

The recent sub-Saharan financial crisis has drawn attention to the role of affordability in Information and Communication Technology (ICT); unfortunately, affordability has a devastating effect on growth and diffusion of ICT in the developing context (Udo et al. 2008). However, with affordability dilemmas, the sub-Saharan region embraced financial and social sustainability as an agenda to be considered throughout the conception and deployment of ICT initiatives.

Since 2015, Mozambican developers hosted most of the non-profit applications, content and media produced in Mozambique in open access servers hosted elsewhere. Needless to say, that developers known the limitations, and technical implications provided by open access servers. The impact of this action was soon noticed by the users adopting international applications to access content and media produced in Mozambique and being exposed to an overwhelming amount of ads and media produced elsewhere.

According to David et al. (2013), co-design refers to the conception or creation of artefacts drawing on a shared vision, social learning and mutual understanding among all key stakeholders, taking into account that all those involved in the design process have somehow different perspectives and expectations which should be adequately considered.

The recent popularity of Internet of Things (IoT) and the growth of bandwidth-hungry applications, it does turn down the cost of ICT (Guo et al. 2012). IoT is a new communication paradigm that envisions a near future, in which the object of everyday life will be equipped with microcontrollers, transceivers for digital communication, and suitable protocol stacks that will allow IoT to communicate with users, becoming an integral part of the Internet (Krishna 2017).

Bearing in mind the advancement of IoT and design approaches that nurture users perception and context; a group of Mozambican developers, active users and promotors of Mozambican content and media, set on a journey to adopt co-design as an approach to address affordability and development of non-profit applications.

In this study, co-design was used to infuse communities knowledge and developers practices to produce an application for the local commuter while addressing affordability in the early rounds of design. Furthermore, affordability and sustainability constraints where addressed by adopting Raspberry Pi and by imbuing developers practices with communities knowledge.

The study presents on the first instance thoughtful interaction design actions conducted by a well-structured and selected group of individuals to co-design an application, and on a secondary instance, the human-computer interactions are evaluated to provide understanding about the impact of contextual interaction design.

In Mozambique, software development has been gaining space and acceptance since the introduction of the subject in the higher education curriculum. The field has also gained acceptance in industries such as entertainment, education, health and banking, as result of the urban diffusion of the Internet.

The Internet commercial growth and diffusion created favourable conditions for the emergence of third-party domain services providers and hosting companies with prices ranging from free to a couple of US dollars (Rabinovich and Aggarwal 1999).

The free hosting is a limited service that provides limited bandwidth, storage and memory, in most cases space and bandwidth provided is not enough for hosting web applications. With this type of service, ads from the hosting provider or other third-party vendors are part of the free hosting package.

The paid hosting services are provided in package bundles from which the developer can select the required bandwidth, storage and memory. For developers in the sub-Saharan region, the main limitation is the price and method of payment. Hence free hosting, which only required a mail registration with all limitations becomes the first selected hosting.

The high cost of hosting it does not only affects developers, but it also affects the users and people who produce content and media. The Internet provided in the sub-Saharan region in most cases covers urban areas, there are few exceptions where the service reaches the rural areas. There exists in the region a digital divide which creates a significant disparity between the has and has not. With developers aiming at hosting services in free hosting sites, there will be a social group within communities which will have difficulties to access content and media created in the country mainly because internet services providers who are not limited to urban areas charge high sums. The problem here addressed are not faced only by users but also to producers of content who are geographically located in the rural areas, they will face the unpleased reality of not being able to upload local content.

The practice of collective creativity in design has been around for nearly 40 years, going under the name of participatory design (Northern European) or user-centred design (United States) (Sanders and Stappers 2008). In 1995, Edworthy and Stanton (1995) presented user-centred as the approach that identifies and capitalises upon the knowledge of the relevant user-population (e.g. medics, pilots, control room operators), moreover Nesset and Large (2004) emphasised that the user-centred approach “had been typical to position users as a testing or evaluation services for designers to ensure those users needs are met”. Furthermore, Nesset and Large (2004) mentioned that the methods employed in user-centred design involve participant observation, system logs, qualitative written surveys or interviews to determine likes, unlike and difficulties.

Kensing and Blomber (1998) described the participatory design as an approach to the assessment, design, and development of technology and organisation systems. Following Nesset and Large (2004) participatory design concept, users change out roles from simple observers to the source of knowledge on their domain, their participation takes a role of peer co-designers, in which users design their expertise and context information.

Sanders et al. (2010) described the participatory design in early 2010 as an emerging design practice that involved different non-designers in various activities throughout the design process.

The terms co-creation and co-design have often been confused and treated synonymously with one another Sanders and Stappers (2008), and Sanders and Simons (2009) pointed out that co-design is directly related to co-creation, in other words, co-design is a specific instance of co-creation Sanders and Stappers (2008).

Russo‐Spena and Mele (2012) highlighted that many scholars (Ramaswamy and Gouillart, 2010; Chesbrough, 2011; Russo-Spena and Colurcio, 2010) use the term co-creation to address how social, cultural, economic and technological changes enable organisations, groups and individuals to interact, collaborate, and solve problems by jointly generate solutions and create sustainable artefacts.

Since the past decade researchers (Kleinsmann and Valkenburg 2008; Sanders and Stappers 2008; Steen et al. 2011; Russo‐Spena and Mele 2012; David et al. 2013) presented different definitions and approaches for co-design, each with its particularity vision and understanding but all the definitions and approaches pointed to the same fundamental approach.

Kleinsmann and Valkenburg (2008) defined co-design as the process in which actors from different disciplines share their knowledge about both the design process and the design content. They added mentioning that co-designing was done to create shared understanding on both aspects, to be able to integrate and explore their knowledge and to achieve the broader common objective: the new product to be co-designed.

Sanders and Stappers (2008) described co-design as broader sense to refer to the creativity of designers and people not trained in design working together in the design development process.

Steen et al. (2011) perceive co-design, as an approach in which diverse experts (researchers, practitioners, designers and developers) come together to create an artefact cooperatively.

Russo‐Spena and Mele (2012) presented their view of co-design as a wide range of practices based on the engagement of many actors linked by a shared context, interest, and it is aimed at a more specific purpose: to bridge the gap between identified ideas or needs and the possibility of finding a solution. Russo-Spena and Mele (2012) went forward showing that the distinguishing characteristic of co-design practices was the high level of interactive learning content, and that co-design was not only seen as an output from inspiration but also as a thought process that involved various activities, such as speaking, writing, drawing, showing, modelling, constructing and documenting.

The approach suggested by David et al. (2013), it circles the co-creation of artefacts drawing on a shared vision, social learning and mutual understanding among all key stakeholders. The approach adopted by David et al. (2013) takes into consideration that all those involved in the design process have somehow different perspectives and expectations which should be adequately addressed. This approach allows researchers to infuse different social and design perspectives to create a solution that would nurture both community and developer’s knowledge.

To bound the population of people interested in media and content for entertainment, a group of developers from the University of Saint Thomas of Mozambique benchmarked most of the websites who promote Mozambican events and media (music, cinema, arts). From the exercise were located sixteen (16) active promoters from Maputo, three from Beira and two from Nampula.

Survey questionnaires consisting of demographic items and 20 primary coded questions integrated with 10 secondary coded and open-ended questions were provided online through survey monkey. The survey resulted in 19 valid responses of which 17 were willing to participate physically in the design process, and the remaining two were willing to participate using Skype.

The second phase was to look for active users of content and media produced in Mozambique, to do so were searched online users who had more than 10 downloads and stream of music per week. This search was done on social medias of websites such as musicafresca.com and musicaboa.net.

Where identified three active users, from them it was possible to understand that most of these users are located in schools such as Josina Machel and Francisco Manyanga. Hence five surveyors scaled the schools where there were selected students from grade eight, nine and ten. The surveyors scaled approximately three classes per grade and were located approximately 30 active users. The 30 active users were divided into two groups with two team leaders that would be speaking for each group.

At this stage, three groups were formed to co-design the solution, one group of developers, the other of active users and a group of events promoters who understand about the online and physical promotion of events and media.

The active users were entrusted with the responsibility to select at least five things they like on websites and five things they do not like seeing when they are trying to access content and media. The developers were entrusted with the responsibility of coming up with low-cost technical aspects to develop and host a geo-contextualised solution. Promoters were responsible for specifying the categories to be promoted and provide information about how they promote each category. To evaluate the usage and adoption of the application, google analytics was the selected tool, as it is free for most analysis and to provide real-time data.

During the early process of designing the solution, developers requested that all the participants (promoters and active users) review additional websites for benchmarking purposes or provided a sample of websites or platforms used for downloads and promotion of music, cinema information and events, and provided at least three positive aspects and negative aspects of the selected websites. Moreover, developers had to evaluate technological infrastructure and explain to the participants the financial implications. At the end of both processes, each group had to present the outcomes of their task and later brainstorm on the implementation process.

The promoters focus on defining the name of the application “+Musica” (www.musica.co.mz) a web-based application, device independent which aims to provide information about entertainment, cinema for entire Mozambique. The platform is feed by the community, with individual accounts able to upload content to be displayed to people who are looking for entertainment information, the application should not have international, or Google Ads instead it should have contextualised and well-structured ads from local companies or business.

The active users addressed the application layout, emphasising the results of the benchmark about the likes, dislikes and needs of a user on a platform. The active users reported that there are five things that the whole group agreed that are unnecessary and unsatisfactory in applications:

• Google Ads, which according to them are annoying and inappropriate;

• Contents are displayed without any regard for users age. Hence material on the website should be censored before making public;

• Content should not contain ads on the website where it was downloaded;

• To access the content and media, there are several redirections;

• Most of the information in a website must be organised chronologically to avoid accessing old content.

The application was developed in three months (October-November 2016), it was taken into consideration the design perspectives of active users, developers and promoters. The application was hosted under the domain www.musica.co.mz, with no ads and where created conditions for any person within the community being able to insert content and media. The application created had (3) categories namely: (1) music, (2) cinema; and (3) events.

Developers turned attention to IoT devices; they are expected to substantially support the sustainable development of future smart cities (Vlacheas et al. 2013). The IoT has been widely adopted due to the urgent solutions sought for viable living conditions and sustainable development. The IoT is inbuilt with a dynamic global and self-configuring network infrastructure where physical, virtual things are identifiable and communicate with interoperable standard and protocols.

The developers were bearing in mind the hosting limitations and the financial benefits IoT devices have to the developers community. Raspberry Pi was adopted to host the application; this is a credit card-sized Single-Board Computer (SBC) developed in the United Kingdom (UK) by the Raspberry Pi Foundation with the intention of promoting the teaching of basic computer science in schools (Wachira 2015). It is interesting to note that Raspberry Pi has a wide range of utilities and can be easily converted to work in any environment.

The Raspberry Pi Foundation developed numerous versions (Raspberry Pi Zero, Zero W, 1 model A+, 1 model B+, 2 model B, 3 model B) of the device since the beginning of the foundation to the present.

In this study, the version used was the Raspberry Pi 1 model B+, which comes with 512 MB of RAM, HDMI port, four USB ports, micro-SD card socket, audio part, camera and DSI display connector, built into a single component. The processor of the model B+ is 32 bit, it has a system chip of 700 MHz, which is built on an ARM11 architecture the same Cambridge-based company that develops processors used in many smartphones (HTC Dream, iPhone 3G, Nokia N97, HTC Hero) (Forhad Symon et al. 2017).

Similar to any other computer the Raspberry Pi 1 model B+ use an OS in our case we adopted the Raspian, an open source OS that kept the price of the platform low (Maksimovic et al. 2015).

Developers took into consideration other gadgets taking into analyses affordability, power consumption and size. In all three categories the Raspberry Pi 1 model B+ came on top with four main advantages:

• It is portable, it has a credit card size, can be carried in the pocket, handbag, and it is lightweight;

• Consume less power and energy as compared to traditional computers as showed in Table 1 Anwaar and Shah (2015), Raspberry Pi 1 model B+ consumes 11664 (eleven thousand and six hundred and sixty-four) Megajoule (MJ) of energy per year, a very lower amount compared (desktop computer, ProLiant Server);

  Full size table

• The most interesting feature is affordability; the cost vary between the 25 to 35 USD, very cheaper compared with standard computers; and

• It is suitable for new application development, testing application, debugging, hardware development, play games and hacking.

The data evaluated in this study was from February to October 2017; during this period the application had 2801 (two thousand and eight hundred and one) visitors from which 48% were new visitors, and the remaining 52% were returning visitors. The quasi-half and half percentage of returning and new visitors it is an indicator that the application has a significant probability to double the number of visitors, as data demonstrates that those who visited the application at least once, tend to become regular visitors and act as application refers.

The 2801 sessions resulted in 12698 (twelve thousand and six hundred ninety-eight) page views with an average duration of 7 minutes per session, from which most of the pages viewed were about music, followed by videos. It means that visitors are more interested in music than any other category.

From the music category where streamed approximately 53197 (fifty-three thousand and one hundred ninety-seven) songs, and were downloaded 6700 (six thousand and seven hundred) songs.

The video category had approximately 3400 (three thousand and four hundred) video views.

The above-mentioned high discrepancy between the streamed times and the number of downloads indicate that most of the visitors prefer to stream than to download, and this can be caused by the type of device, operating system or storage space. The devices used to access the platform where most mobile phones (55.9%) and desktop computers (35.8%) and tablets (8.3%). The operating system (OS) most used by visitors is Android OS, Apple iOS and Windows OS.

The data demonstrate that the majority of application visitors in Mozambique use devices with low storage capacity rounding between 2–4 Gigabytes (GB). Hence, the researchers assumes that the space required to run the OS and mobile content limits the space to store media and content from the application. The iOS users differently from those with low storage capacity, they cannot download music from the application due to Apple restrictions.

An interesting point to notice was that the more extended sessions were done via computers, which leads the researcher to believe that those visitors have been streaming. Further, the researchers could notice that visitors using personal computers access the website via a national network, free access points of internet and home network, differently from those who used mobile phones which in most of the cases used mobile operators network.

The number of megabytes spent by each visitor during the streaming is directly connected with the application server configuration. The streaming bitrate varies from 20 Kilobits (kbps) to 320 kbps; applications with high-quality such as Google Player, Beats Music and Spotify use up to 320 kbps for streaming and consequently increase the amount of data spent by the visitors when streaming compared with those using 64 kbps such as Pandora.

In one hour the amount of data spent by a visitor in an application with a configuration of 320 kbps is 115.2 Megabytes (MB), and with a lower configuration such as 64 kbps, visitors spend in an hour 28.8 MB.

Considering the context of the application users, and taking into consideration the financial constraints, developers configured the server with a lower bitrate (64 kbps). Furthermore, it was adopted a deductive method to calculate the amount of data spent by the visitors in +Musica (www.musica.co.mz). Therefore, it was made an estimative of hours used for streaming, using the average session duration (7 min) times the number of sessions (2801) which resulted in approximately 327 h.

The estimates mentioned above-allowed developers to calculate the quantity of data and traffic generated by visitors, where:

The xMB was converted to GB and Gigabits (Gbps) as shown in the calculation below and the results for the traffic generated with audio streaming was of 75.3 Gbps and visitors spent approximately 9.2 GB.

The average size of each music downloaded was 6 MB, and the total of downloads was of 6700, with that the researchers were able to calculate the approximated quantity of data (39.3 GB) spent with downloads which resulted in traffic of approximately 314.1 Gbps.

In the first months, the application only received traffic from direct access, and by the end of the evaluated period, the scenario changed, with 37% of the visitors being from sources such as organic search 21%, 15% from social media and 1% from referral searches. Direct access amounts up to date to 63% of traffic generated.

The application was designed to allow users to be informed about all kind of event, visitors of this category were not tracked, because there is a missing link between actual visitors and people who did attend the event.

The internet frequently has been characterised as male-dominated service (Padilla-Walker et al. 2010; Tsai and Tsai 2010), in this study, the results are not different and indicate that the majority of visitors are male with 68% of visitors but with a lower average time spent in the site of approximately 7 min. The female users who are 32% of visitor access the application from developing countries such as Mozambique, South Africa and India with an average double time online than male users.

The visitors age varies between under 15 to over 45; there has been a high promotion of the application in social media (Facebook, WhatsApp and Instagram), most of the promotion is performed by people who visit the application. The highest age group which visited the application range 18 to 24 (45%) followed by the range 25 to 34 (40%) and the remaining 15% were from visitors with age over 35 years.

For the surprise of co-designers (developers, active users and promoters) social media made the application known around the globe in countries where the primary language is not Portuguese such as United States of America, India, Russia, South Korea and China.

Most of the website sessions are from Mozambique with 2398 (85.61%) followed by the United States of America with 193 sessions (6.89%), South Africa 40 (1.43%) and India 37 (1.32%) sessions.

During the analysis period the application was inactive during 3 h for six (6) scheduled maintenance sessions. The maintenance consisted of updating some functionalities and perform upgrades. The functionalities upgraded were related to album upload, and album streaming, the release of each upgrade allowed the co-designers to add new ideas and conceptions.

Co-design sustainable ICT solutions, became an emerging form of societal-scale computer supported cooperative activity which challenges communities and developers knowledge of information system. Not only due to the peer-to-peer interaction a phenomenon of sociological and technological design but also to the practices of skilled professionals that have to be infused into community ideas, perceptions and expectations.

The peer-to-peer communication activity anticipated contextual implications on the cultural and political point, and it concluded by becoming the driving force to solve complex critical pieces to deploy and register the application domain. During the development stage public participation had to be assured by allowing the application to work with mobile devices to allow communities the opportunity to interact further with the artefact.

The segregation of tasks between the promoters, active users and developers was the critical success of the co-design process and the application adoption. There were a clear perception and imaginary boundary in which each organisational structure was open to explain the twists and approaches that could be used to produce an application capable of solving the existing problems.

Active users stressed that ads in a foreign language make browsing experience complicated, the application should be easy to browse and capable of delivering content and media. This concern was taken in consideration by designers, by not allowing ads to be hosted in the platform under any circumstance.

The use of low cost and processing devices in this case study allowed developer to perceive that quality can be provided by using Raspberry Pi without reducing or impacting quality. Nevertheless, being a community application, the capability to have more than 50 users simultaneously connected, is a technology limitation due to unavailability of resources required by the hosting device to provide the same quality of service. Moreover, developers believe that such limitation will be solved with the next version of the application.

In this study interactions became imbued with participant activity, discussion from the social to cultural and technical perspective, reflecting on the practices of the social organisation and possible application users.

The release of this application changed the entertainment market in Mozambique directly and indirectly. The introduction of a new application changed the way things were done, after the release of the application, the co-designers continued to benchmark the application, and it was possible to notice competing platforms have been adopting the same design approach. Competing applications such as musicaFresca have reduced the number of ads and implemented direct download links while new platforms began to advertise events and local ads.

It is not known if this behaviour changes is related with the appearance of competition in the market or not, what is certain is that the market is becoming competitive which paves the way for more interaction design and collaborative approach to application development.

In this study, it was noticed that language is not a barrier to access a specific type of content, as in the application the portuguese language was used to display content. The information and the structure of the application are written in Portuguese which was not an obstacle for visitors that are non-Portuguese speakers, approximately 60% of the visitors were English speakers, the double percentages over the Portugues speakers and other language speakers.

Furthermore, it is not only about sustainability, or designing an application or putting something in the market, it is about the desire that the market has to use an application that will save bandwidth and result in the better use of technology. While the actual market is hanged on ads, users get frustrated because due to the time required to load the pages.