1 AGIMO BIG DATA STRATEGY ISSUES PAPER AIIA Response 5 APRIL 2013
2 2 INTRODUCTION The Australian Information Industry Association (AIIA) is the peak national body representing multinational and domestic suppliers and providers of a wide range of information technology and communications (ICT) products and services. We represent over 400 member organisations nationally, including global brands such as Apple, EMC, Google, HP, IBM, Intel, Microsoft and Oracle; international companies including Telstra; national companies including Data#3, SMS Management and Technology, Technology One and Oakton Limited; and a large number of ICT SME s. All of our members, large and small are committed to developing Australia s digital capability and presence nationally and on the global stage. We are pleased to have this opportunity to respond to AGIMO s Big Data Strategy Issues Paper. OVERVIEW AIIA welcomes AGIMO s development of a Government Big Data Strategy and is particularly pleased that the paper was announced and released to coincide with our recent Big Data Summit in March. The Summit explored the benefits of Big Data and its potential for delivering better decision-making and more effective analysis to support enhanced government policy making and improved service delivery for consumers, businesses, government and citizens, and we note the AGIMO Big Data Strategy Issues Paper identifies similar opportunities. A copy of the Communiqué released following the Summit is attached for your information. We ask you to specifically note the key outcomes and call to action section of the communiqué. COMMENTS 1. While the paper sets out a rationale for Big Data in Government it does not propose a vision for the Big Data opportunity. It will be critical that this is clearly articulated in the Strategy and that supporting objectives and activities are closely tied to that vision.
3 3 2. As is the usual practice with Strategy papers AGIMO s document must also explicitly cover the why, who, how and what in key areas such as: Leadership Skills Infrastructure Interoperability Privacy and security Data governance Data curation Risk management 3. In terms of definition AIIA would make the point that a key differentiator of Big Data is not that it is simply big but that it can legitimately include vast amounts of unstructured and incomplete data. A unique attribute of the Big Data concept is that deep analysis can be achieved with less than 100 per cent data accuracy. The focus on accuracy in the realm of Big Data is less an issue than the provisioning of the data which is the converse in this paper. 4. The Paper provides a useful overview of the application of Big Data capability, including in relation to Government policy making and service implementation, but to achieve buy in from relevant Government agencies will need to articulate a clearer and more compelling value proposition. AIIA suggests that this include the perspective of Government, customers of Government services and citizens generally. Given the vast scope and scale of data assets held by Government, Big Data capability facilitated and used by Government (through its Departments and Agencies), has potential to drive a range of economic, social and personal benefits. We suggest therefore, that Big Data wins are clearly articulated and re-enforced in areas such as: personalised medicine better medical treatments evidence based social policy programs real time evaluation of the effectiveness of service delivery Early successes need to be leverage to create a positive culture/message around Big Data.
4 4 5. We note that the paper is somewhat at odds with other major government approaches (eg US CTO) that are adopting the view that vast holdings of government data should be released for public access. The Paper seems to assume the Australian Government owns the data it stores and will release it only if it thinks it should be in the public domain - an approach quite different to that in the US. Whilst mindful of potential privacy sensitivities the US has addressed privacy concerns by masking or obfuscating sensitive personal information and markers, etc. In this context AIIA holds the view that data is owned by all Australians and a resource that we all own and should be able to access and utilise. The balance of power should be viewed to sit with the Australian public with Government as the custodian of this valuable public resource. 6. AIIA would point out that the summary of challenges introducing Part 3 of the paper are exactly what Big Data is about - this is the opportunity not the challenge for Big Data and is what makes Big Data valuable and valued. The tool/s to analyse or treat data holdings specific to a problem are available in one form or another. 7. We note the concerns related to privacy, security and trust but believe a sensible and pragmatic approach is required. As was discussed at our Big Data Summit nothing comes without risk and like most issues knowing, mitigating and managing risk and enforcing sanctions and reparation as appropriate, are the foundations for consumer and citizen trust and confidence. This equally applies in the world of Big Data. Also, as noted in Paragraph 5 the issue of sensitive personal privacy issues while real, can be managed and should not be an excuse for not using this very powerful and beneficial technology capability. 8. We would also make the point that in order to get broader public buy-in to the concept of Big Data there needs to be a transparent process for the Australia public to opt-in or out of data sharing and clarity around what data is used for. Additionally, processes for reviewing the validity of citizen centric data need to be put in place and policies around management of data when it is used for a purpose for which it was not originally collected. 9. Another issue raised at the Summit and relevant to the AGIMO Strategy development is the need to contextualise and optimise the use of Big Data by ensuring the right questions to inform analysis are identified and asked in framing Big Data analysis. While the value of data holdings may be unclear or even unknown until the analysis is done, it is nonetheless critical that Big Data analytics is framed by a focussed investigatory and discovery process.
5 5 Capturing, manipulating and preserving data simply because we can is not reason enough to do Big Data and arguably in time will diminish its value and undermine the bigger issue of trust. 10. Also to the extent that Big Data is a collection of data sets so large and complex that it becomes difficult to process using on-hand database management tools or traditional data processing applications, the role of real people in setting analytic parameters and framing research/investigatory hypotheses remains critical. 11. The aggregation and combination of data, including from public information sources and data already available in both the broader public and private domain needs to be included in the Strategy. With appropriate (and necessary) privacy controls, there is enormous opportunity to stimulate the unlocking of analysis previously neither contemplated nor possible. One of the key value propositions of Big Data is its release to see what happens when integrated with other information in conjunction with analytical tools for new insights. There is a wealth of data that is held by Government agencies that is not related to the Australian public but has enormous potential benefits for example research data, weather data, and environmental data. The US Government s Big Data Research Initiative is a positive model for tapping into this valuable resource. The issue therefore is about how Government will ensure data is (responsibly) released into the public domain. This also needs to be addressed. 12. As an extension of Paragraph 6, as the custodian of huge national data assets, Government is well placed to facilitate innovation - in knowledge, product and service development. To this end the specific economic opportunity of Big Data is not mentioned in the Paper. This includes: commercialisation of the R&D that is coming out of Australian research institutions; and early adopter advantages that Australian businesses may derive from innovating with Big Data. AIIA strongly recommends that the themes of innovation and potential commercial opportunities require attention and deeper analysis in the Strategy paper. 13. Interaction between the public, private and research sectors in this space also needs to be called out, and specifically in the following areas:
6 6 to set privacy and security standards to drive Big Data best practices to encourage commercialization of Big Data research/ip developed in Australia to disseminate knowledge, successes and failures across agencies, across industry and across the business community. 14. Data management and curation needs to be further developed, particularly how this is operationalised. While AGIMO s paper will have a strategic purpose, it should foreshadow the need for and development of, these supporting frameworks. This is an area in which industry is developing a high level of expertise and we would recommend that Government draw on industry s experience moving forward. 15. AIIA supports the need for skills to be addressed in the Strategy. This is not just an issue for Government but for governments and business nationally and internationally. If Australia is serious about Big Data we need to support the Strategy with an appropriately robust skills development program. If we fail to invest in building our capability there is a real risk that analytics and data management functions will themselves move off-shore. 16. We note AGIMO s intention to establish a cross agency Big Data Working Group. We recommend that the outcomes of that Group as they relate to development of the Strategy are clearly articulated. There is a risk that the opportunity of Big Data literally gets bogged down in internal, agency specific details and concerns and that the broader intent of the Government to progress the Big Data agenda is lost. 17. By way of observation the Paper discusses the Whole-of-Government Data Analytics Centre, and the remit of the ATO Data Centre of Excellence, neither of which is explained. The Paper assumes industry is familiar with these initiatives. Clear explanations are needed as to why these have been established, what they are intended to do, who is represented on them and the outcomes that will be delivered. We would also make the following subject specific comments: 18. The Strategy needs to have regard to visualisations/simulations/environmental models of Big Data to support initiatives such as Australian development of the Digital Earth, Virtual
7 7 Nations, DataCi ties movement (as outlined at dcitynetwork.net/manifesto and virtualanz.net). The Strategy needs to have regard to our strategic involvement in these overarching initiatives. 19. In the area of medical research Australia needs to invest in Big Data analysis and IT infrastructure opportunities. New IT platforms are required to enable new medical research discoveries to happen faster and secondly to enable our research teams to collaborate more effectively internationally. Infrastructure investment, including analytics solutions for secondary use of data, is critical and necessary to fully capitalise on previous investments in the NBN and PCEHR. Importantly, at the individual patient level the aggregation and consolidation of genomic and clinical data will further enhance efficient healthcare delivery, cost savings and improved patient safety - all objectives espoused by DOHA and State Health Departments. In support of this we would make the point: Data used in medical research are often structured and codified to facilitate analysis and aggregation of data. Well established international coding standards have been enforced by regulators like the FDA & TGA for reporting of Clinical Terms, Adverse Events and Drug Names. This simplifies downstream epidemiological analytics at the population levels. The PCEHR already provides a facility to collect and consolidate data for individuals who register for this service. Moreover, the PCEHR enforces data and document standards across primary and acute care facilities and across states to allow access to patient shared summaries and event data. As well as providing support for improved personalised healthcare and continuity of care for individuals, deidentified data from the PCEHR will ultimately provide a valuable resource of longitudinal data to supplement other data sources and assist in medical research and healthcare policy. The rate of healthcare data accumulated is exponential, especially with the rising tide of more cost-effective genomic testing and the pressure from social media and consumer groups to offer an increasing array of genomic testing for specific diseases. Putting in place the necessary IT infrastructure at a Federal and State level to support these data will be essential (similar to the impact of digital radiography, which pushed the need for infrastructure to support that revolution). Countries which are early adopters of these technologies will be well positioned to develop solutions and associated IP to drive
8 8 commercial growth and industries (e.g. South Korea with digital radiology and Singapore with genomic research and informatics). Patient safety will be improved. Ready access to centralised and coordinated drug safety data will be an asset to complement the initiatives of the TGA, aimed at improving compliance for reporting of adverse events and establishing systems for early warning of epidemics or other unforseen health risks. 20. We have specifically called out these examples related to health and medical science because this is an area that offers immediate potential to exploit Big Data capability and in developing the Strategy should be identified as a potential priority area of focus.