Digital Skills and Data Science Technicians

Our wonderful government yesterday published a consultation paper on its industrial strategy. Along with typical pronouncements of aspiration and money already spent, plans previously published, there were a few new things, and then some much more open questions. As usual the debate in the House of Commons was parochial and underinformed, while the Lords developed a few interesting points. An industrial strategy makes tacit assumptions about the nature of industry, that ist is well defined and nationally contained. But supply chains extend into other countries, particularly with advanced manufacturing, and the delineation of industry from the services sector is limiting. Obviously some services are essentially industrial aftercare, as in the case of jet engines, but the neglect of services when they are a large part of the economy seems to lack strategy, especially financial services. Some interesting points are made about skills, and the need for adults to continue gaining new skills. Mostly this rolls up perennial concerns about basic skills like literacy, without identifying that oral language is more important now. Another continuing promotion is apprenticeships, although it is recognised they have many at a low level and of low quality, and of course the unquestioning focus on STEM. Indeed it looks like the aspect on skills was not quite ready for the publication of the strategy, like Adrian Smith's review of post-16 maths. Specific Skills Shortages One aspect of the consultation is very interesting and very open:

13. What skills shortages do we have or expect to have, in particular sectors or local areas, and how can we link the skills needs of industry to skills provision by educational institutions in local areas? (p49)

This combines a query about sectors, localities, now/future with providing to meet these needs locally, inviting specific responses but vulnerable to special pleading. This is specifically framed at the 'technician' level, as with the apprenticeships strategy, recognising gaps in technical skills, which are not at degree level. Another point is made about poor 'digital skills' being a problem for 20% of the adult population, who need access to basic training. Although it does not explain exactly what is meant by digital skills, they are listed alongside literacy and numeracy, whcih seems a missed opportunity. Embedding digital skills into existing capability is what would make a strategic investment in training - teaching everyone to send emails does not. Data Managers For some time there has been a recruitment and career development problem for data managers - people document, code and verify data. These are not mindless data entry roles, although sometimes they are treated as such, but require knowledge of meta data standards and an array of legal frameworks on ethics and personal data. Data managers assist researchers particularly in accessing data and understanding prospects for data linkage in a contribution which is not appreciated until you have tried to work with undocumented data. Data managers exare essential for larger clinical research projects and social surveys but there are not clear routes in or development in role. Previously, our wonderful government has published a science strategy focusing on 8 great technologies. Most of these represent evolution of existing research, so that the shortage of technicians and technical skills described is about local recruitment of people with higher level technical skills in specific sectors. However, 'big data', for all its hype, does bring some novelty - MSc programmes in Data Science are popping up all over, even if they are just rebranded statistics courses. But undergraduate programmes, let alone anything more technical have not been identified as having scope for development.   Skills for Data Science Data handling is already a core part of the maths curriculum in schools, throughout primary and secondary. Coding in schools has found no difficulty in showing people how to do basic programming, and the Royal Society report found no difficulty proposing an end to windows application in favour of computer science. None of this links the two - statistics in schools is still generally based on data that can be manipulted and tabulated by hand; computing is about self contained exercises not wrangling external data sources. We need to develop resources and then curriculum which support data science activities by school children, both in and out of the classroom. The gap is greater in FE, post-16, and I hope the Smith review will comment on this, as the only route is to go and take a degree, which we know many people (around 40%) do not. These are people who have capability but different ambitions whose circumstances tie them to home or earning, without the readiness to take on the higher tax ensuing from student loans. There needs to be a vision for statistical, mathematical and scientific computing as a substantial apprenticeship route, into data science, data management, and whatever else is coming around the corner. This fits exactly the local skills gaps and training post 16 that the government is consulting on, but it is not easy to solve. Data Science Futures Data science is a new area of work, and still has maverick tendencies, as seen in some unethical use of machine learning algorithms, and widely trumpeted findings which turn out to be a oneoff. At present, routes in are from statistics or from computer science, with one group picking up the other set of skills (or not). But programming a computer to work with data is not hard, indeed it is the basis of creating APIs, so there is no reason it should not be statistically rigourous. Conversely statisticians have created R - there is no reason to stand sniping on the sidlines about the cavalier approach of the machine learning community. The Government Statistical Service (GSS) has established a data science campus at its office at Newport, Gwent. They have recruited a small number of apprentices, and they are now recruiting many more people, university leavers and above, as data scientists, in Newport and across the GSS. These are the large employers who have the capacity to take on apprentices and 'overtrain' creating more than they need so that the sector is well supported. But this has not found its place in a wider strategy - it is not clear that the maverick view of data science is fading yet, and shortages of mavericks are not the same as shortages of skills. We need to anticipate skills needs to support our great aspirations in big data as a great technology, and start attending to how this may be dveeloped locally and in schools, not just in cash cow MSc courses.