37OV ERV IEW OF THE FL AGSHIP UNI V ERSITIES
Chapter 6 Case study B: Evolution to a nascent tech hub: Positioning the Joburg Centre for Software Engineering
6.1 Evolution of software development capacity building through the Joburg Centre for Software Engineering
6.1.1 From thinking to concept to contract to centre: 1990-‐‑2004
This section of the narrative was written based on data from interviews in 2008 and 2014, from participant observation incidents between 2008 and 2013, and from a review of institutional documents (secondary sources) for the period 2010-‐‑ 2013.
The origins of the JCSE date back to 1990 with early discussions on where ICT should be located in the post-‐‑modern university. In the 1990-‐‑91 Wits model, ICT was located across three faculties with little collaboration and with different cultures and ideas of computing: telecommunications, data processing, signal processing and electronics in the engineering faculty, computer science and information systems in the science faculty, and information systems management in the commerce faculty (key informant JCSE-‐‑S) (coded as institutional design). A single engineering academic was engaged in cross-‐‑disciplinary research on software process and quality in an era where cross-‐‑disciplinary work and industry engagement was frowned upon. Despite a merger proposal for
computer science and information engineering presented by the Dean of Science in 1989, no coherent entity was established to populate this emerging field of knowledge, even as technological and services convergence advanced. This split approach (coded as institutional fracture) created some degree of confusion in the marketplace about how industry and students should relate to Wits offerings (JCSE-‐‑S, 2014):
This fracture continues to this day…. By 2002, there was some interest in starting a software research centre, because of strong and continuing linkages with and research sponsors from industry…The idea of creating a “front end” to ICT led to proposal to establish the JCSE as a strong industry-‐‑facing centre that could present a coherent face to industry.
The initial proposal to Senate was rejected, as were several further proposals, partly because of the desire for a “no change” environment (coded as actor contestation).
The actors from engineering were regarded as too close to industry and there appears to have been a lack of understanding of the very different type of knowledge being pursued in software engineering as compared to computer science. Indeed, it can be argued that the discipline of software engineering is entirely connected with industry and has limited reason to exist outside of industry applications, just as mining research has limited value outside a mining applications context and health sciences research has limited value outside a health applications context. These knowledges cannot grow in any significant depth outside an applications context, establishing demand for a different form or medium of knowledge than research publications (coded as values: contestation of knowledges).
Despite rejection of the proposals by Senate, a business plan was written for establishment of the “gateway centre”, because the City of Joburg metropolitan municipality sought to profile Johannesburg as a software hub as part of its city development programme. However, stakeholders consulted on the business plan did not want such a centre as previous initiatives (iLab) had not been successful and had created no value in the market. A few software suppliers and users argued for creation of a centre that promoted better quality software
development, cementing the formative idea of the JCSE. Further university-‐‑ industry discussions and pre-‐‑funding from the municipality led to Senate approval of the Centre in 2004 and a formal contract with the City of Joburg (coded as coincidence of interests of actors). A review of best practice in software engineering led the main actors to a key report produced for the Department of Trade and Industry (the dti), namely the SAITIS14 report (the dti, 2000), which exposed the weaknesses in the local software sector as related to skills and performance. The report provided the foundational thinking for what came later at the JCSE, setting out a detailed ICT sector development framework (coded as coincidence of interests of institutions).
While the conversation with Wits and the City of Joburg was maturing, the Gartner group advised Wits that it was important to work with a CMMI level 4 or 5 rated company to develop the Oracle student software information system. Tata Consulting Services (TCS), rated at level 5, became engaged in this work, resulting in a visit to TCS India operations in 2004, the flow of events influencing the early agenda of the JCSE towards building a South African CMMI training programme, still offered in 2014 (coded as coincidence of the interests of institutions).
6.1 2 Early years of the JCSE: Laying foundations and developing internal capacity 2005–2011
This section of the narrative was pieced together from foundation interviews in 2008, a review of documents published between 2010 and 2013, as well as a review of the JCSE website. Publications where this researcher is the first co-‐‑ author are cited.
The JCSE was launched in May 2005. A relationship was established with TCS India and separately with Carnegie Mellon University for the purposes of CMMI (Capability Maturity Model Integration) training to advance software innovation in South Africa (coded as global institutional linkages). The Advisory Board was constituted in 2005 and early relationships were established with the first funding partners, including the City of Joburg, Vodacom, Microsoft and other software firms (coded as local institutional linkages). It was essential to unlock industry funding since no university funding was available, only space and use of Wits facilities. Funding was very limited at the inception of the JCSE – Microsoft and IBM each funded a software lab, offering only equipment, not money. Under these relatively open conditions, both proprietary and open source software interests were established at the JCSE and opened the door to other partners coming on board. The only partner who put in significant funding in the early days was First National Bank (FNB), because of their need for skills
(coded as resource availability). Initial financial inputs were relatively small, thus development of the centre was relatively slow.
A Director was appointed in 2006 and international relationships with Carnegie Mellon and TCS began to mature with the establishment of local CMMI training capacity. However, money was needed to train trainers and JCSE attracted funding of ZAR1.6million from the dti enabling it to create the CMMI
programme, teaching and certifying local trainers, consultants and appraisers, which increased income flows. The focus advanced to growing income through Master classes, a CMMI conference, forums and industry events (coded as resource generation).
The JCSE had five founding principles, some traditional university values and some not (coded as values: non-‐‑traditional), namely to support the software industry through promoting best practice in the African context; to grow skills and capacity; to support transformation of the sector race and gender profiles; to participate in trade missions with the Export Council; to make available applied research including the annual skills survey; and to promote innovation and incubation. Industry promotion and applied research are not highly regarded at Wits, though the reasons are vague. At the early stage of grounded theory data collection, the following insights emerged (JCSE-‐‑S; Abrahams & FitzGerald, 2012, p.7):
If we bought all our software from (Sakai), ... we become a client to someone else'ʹs way of doing things. If your cell phone talks to you in Sotho, you are more engaged in that technology’…However, there is resistance to the existence of such a centre at a university, ‘...if you look at the (institution'ʹs) values you find things about engaging in the broader society... if I talk to Microsoft that'ʹs perceived as something you shouldn'ʹt do because they'ʹre the commercial demon. That stuff devalues what I'ʹm doing against a perceived purity that it'ʹs too
commercial, too engaged.
Despite university resistance to the values of heightened industry engagement, by 2013, the JCSE had programmes against each of its five espoused principles (coded as values: explicating new values).
From 2007, the emphasis was on getting the JCSE footprint into industry. Public relations activities, press releases, dissemination of an annual report and
marketing messages, and attending trade shows started to push the identity of the centre, which started to have some visibility, but was still operating on a shoestring budget. The research focus developed from 2008, initially through the production of the annual skills survey, which was influential in industry because it identified the top skills priorities on an annual basis, and was the only
available research to identify the size of the sector, the skills profile and the skills gap (JCSE, 2010). The centre developed high visibility in the banking sector, the software sector and the government IT sector, leading to increased income earnings (coded as academic value). The JCSE had to match expenditure with income, as the two main founding partners, the City of Joburg and Wits were insistent that the centre operated on a zero deficit business model. The upside of this agreement was that if the centre earned its own income, in theory it would not be at risk of withdrawals of grants. The downside was that (JCSE-‐‑RS, 2014):
…if we were a business, we would have access to other means such as selling equity or getting loans, but we don’t have access to mechanisms that businesses use, so we can only operate through once-‐‑off grants from business and
government (and income from training), so a lot of time is spent on looking at ways to get injections of money. Writing proposals is very time consuming, managing agreements, and managing non-‐‑payment…no deficit but hardly any profit.
But finances did come, including ZAR5.4 million from the dti to do Team Software methodologies, the next step in the CMMI framework. The JCSE grew its annual income from under a million rand in 2004 to ZAR24.3 million in 2012 (JCSE, 2013a).
6.2 Advancing through software engineering and prospects for data science