Historically, IT projects in the public sector have tended to be overspent and underperforming. Not the sort of track record that sends out the right signals during a poor economic climate – but can things be done differently? One major problem is that central and local government have never been regarded as a joined up set of agendas or processes, and each part has gone its own way with its own systems and software. Therefore, when central government decided that a main aim was to centralise reporting, problems arose around what format the data needed to be in – and how to retro-fit existing systems to monitor and report events in the right manner.
Many different approaches have been put forward over the years, with minimal success due to lack of impetus, funds or just poor implementation. One that keeps being trotted out is the use of open source software to lower the cost of public sector IT.
On the face of it, this seems valid – using commercial, off the shelf software (COSS) involves paying a licence fee up front and then a maintenance fee per year after that, or an on-going yearly fee that nominally covers the licence and maintenance such that the customer (in this case the government) can use whatever version of software it wants whenever it wants. With free open source software (FOSS), the licence fee disappears. With millions of employees in the public sector, saving even a few pounds per person on licence fees must be good news.
However, software support is still required, and for FOSS this tends to come as a subscription, which in many cases is pretty similar to the maintenance component of COSS. For front end systems, such as operating systems and office software, familiarity is a big issue. The majority of employees understand Windows and Microsoft Office, and training can be kept to a minimum. A Linux/OpenOffice or equivalent system can present a major training problem, and on-going help desk issues which can easily wipe out any savings made from the licences. Also, there can be issues with “round-tripping” – the sending of documents to others in the process chain where actions are required and the document is then sent back. For standard government-to-citizen documents, this should not be a problem, as forms should either be web based or sent out in final format (e.g. Adobe PDF), but for supplier documents or others where e.g. macros may be involved, the loss of fidelity during the round-tripping could result in a less than legal contract being put in place.
But, for back-end systems, COSS can make a great deal of sense – the use of Linux as an operating system for large scale operations is proven, and can save significant amounts of money where the application is agnostic as to the underlying operating system. An area where this may be the case is in what was originally termed by Sir Peter Gershon in his report looking at streamlining government IT as “shared services” and is now being touted as the G-Cloud.
The idea behind the G-Cloud is to provide a set of services that are common amongst a broader group of users than a single department in central and local government. For example, HR tends to be similar across the public sector – employment contracts are by role, booking holiday is a common process and so on. Therefore, having many hundreds or thousands of bespoke systems dotted around the public sector is nonsensical, and expensive. By centralising a HR application in a manner that allows any department to use the same system, redundancy of function can be driven out of the process. Along with this, the scale of the master contract can drive down licence and/or support costs, and data becomes centralised allowing better monitoring and reporting.
A cloud platform requires flexibility at the basic level, however. The workloads involved may be unpredictable or, even where there is predictability, may be cyclical as with payroll, which runs weekly or monthly with little activity in between. Therefore, a highly scalable, flexible and sharable platform is required spread across a large number of servers. Using COSS operating systems can end up with large costs to cover the number of CPUs and cores involved, whereas FOSS gets rid of this cost.
Another aspect of G-Cloud is that it should lead to a lowering of costs for COSS through the centralisation of licensing. At the moment, a high degree of procurement is carried out at a local or departmental level, leading to licence and maintenance costs being relatively high. Centralising license usage through to G-Cloud means that greater economies of scale can be applied, and central contracts used to drive licence costs down to a minimum. Also, actual usage can be monitored, and over-licensing can be identified and removed, as well as negotiating with vendors to allow the use of concurrent licences, where the number of people using a licence at any one time is covered, rather than the number of people who could use a licence.
Another aspect of G-Cloud is a statement by central government that it agrees that there is a need to make dealing with government easier and cheaper so that smaller suppliers can get involved. Historically, public sector projects have involved suppliers jumping through so many hoops that only the large vendors and systems integrators could afford to get involved. G-Cloud is supposed to make it easier and less costly to become a government supplier – and once on the list, for government and local departments to easily find what is most cost-effective for them and take the service directly without the need for new negotiations. Encouraging smaller suppliers into dealing with the public sector should create more competition and so drive costs down.
Even with the financial climate remaining gloomy, the government has stated that G-Cloud remains a priority. It is to be hoped that this is the case, as it does offer the best promise for IT costs within the public sector to be driven down.
However, the biggest issue may well be around how data is managed and used across the public sector. It is all well and good being able to gain access to applications and services directly and to implement them via self-service, but if the end result is a silo of data that cannot easily be shared across departments as required, or cannot be reported against in context against other relevant information, then any cost savings will have been wasted.
Open data is required – both in the way that it is formatted, and in the way that it is made available for use. It is a given that the public sector has to be very careful in how data under its control is secured – there have been far too many headlines in the press about how information has been lost, leaked or stolen. However, this should not result in a knee-jerk reaction aimed at locking down the data – far from it; the end target has to be for all information to be made available in a manner that ensures that it is secure in the context in which it is being used.
For example, a citizen should be able to see all information that appertains to them – but should not necessarily be able to see the same information for their neighbours. A public sector worker may need to be able to see a sub-set of information about a defined group of people, whereas an external contractor may only be able to see such results with any personal identifiable information excised. This capability requires data to be held in an open format, such that the overlying systems that manage the granular security required can deal with each data source adequately.
On top of this is the need for other data types to be included. An increasing amount of information about citizens, departmental work and laws is being held in standard documents, rather than directly in formal databases. As time goes on, this is likely to grow to include voice and video information as well. Tools will be required to ensure that any type of information can be included in public sector processes, and that the process chains can be open enough for public and private employees to be able to fully participate in the processes as needed.
The future of public sector IT is still in the balance. The G-Cloud is undoubtedly the biggest change that can bring cost savings to public sector processes, but needs more weight and impetus behind it to make it more generally available with more functionality, faster.
The judicious use of FOSS can help in managing costs – but an ivory-tower attitude of “FOSS everywhere” could have process costs that wipe out the technical savings. Open data has to be at the core of any public sector technology strategy – getting this wrong could destroy any possible cost savings, and could make it difficult to recover the situation in the near term.
Although still in the balance, the government must take this opportunity to drive a more flexible and functional platform across the public sector. This will be a slow process overall – but quick wins can be made for little investment, and the savings made must then be ploughed back in to gain greater savings in the future.