Open standards: open opportunities

Response submitted on behalf of SALTIS to the Cabinet Office’s consultation, Open Standards: Open Opportunities, the consultation period for which closed on 4 June 2012

The Cabinet Office proposes:

  • to compile a list of approved open standards;
  • to mandate the use of these standards by central government departments and their agencies;
  • to encourage the wider public sector to follow the lead of central government.

A summary of the SALTIS position is that:

  • we strongly support the wider use of appropriate open standards;
  • we believe that the Cabinet Office’s approach may improve the transparency of central government data;
  • we believe that the current proposals will have little positive impact on the wider public services, where the imposition of bureaucratically selected standards is likely to hinder innovation.

We urge the Cabinet Office, before proceeding with this policy, to consider carefully:

  • the different requirements for:
    • standards for central government data,
    • standards for the peripheral public services (such as health and education);
  • the dependencies between these different types of standard;
  • the dynamics by which consensus over standards emerges in commercial markets;
  • the reasons why these dynamics do not currently work in public service markets;
  • the best way to support the creation of standards for public-service markets which are fit for purpose and up-to-date in a world of increasingly rapid technological development;
  • the risks inherent in a policy which centralises control of standards in Whitehall.

A

Criteria for open standards

A1

How does this definition of open standard compare to your view of what makes a standard ‘open’?

A single definition of an open standard is only necessary for a policy whereby standards are going to be mandated. As it is neither possible nor desirable to mandate standards on the wider public service, this definition is of limited importance from the public service perspective.

In certain respects, the definition is too loose:

  • in the context of education, any “pay to view” cost has been found to present a major obstacle to adoption;
  • it does not include the right to produce derivative standards.

The significance of the second point rests on three further considerations.

  • There is a need for constant innovation in the standard itself—this requirement becoming increasingly important as the pace of technical innovation increases.
  • Organisations which have the sole rights to develop existing standards (even if freely available in their current version) enjoy a pseudo-monopolistic position, being able to sell influence over the future development of those standards to interested parties, which may be hostile to the cause of open standards and open markets.
  • Standards do not exist in isolation, but rather in a continuous and inter-dependent ecosystem. Organisations which hold property rights in a standard may not allow that standard to be modified for use in combination with the specifications of rival organisations, creating a series of competing “walled garden” standards environments.

In other respects the policy may be too tight, outlawing de facto standards which may in some circumstances represent the most effective solution currently available. The requirement for standards that are open needs to be balanced against the requirement for standards which offer appropriate functionality—and central government is not in a good position to compare the relative importance of these different criteria.

A single definition of an open standard is unnecessary because the policy of routine mandation (on which the requirement for a single definition is predicated) will be harmful. Where mandation is judged to be necessary, the best standard should be chosen with regard to the circumstances of the particular case, and not with regard to a general rule of thumb.

A2

What will the Government be inhibited from doing if this definition of open standards is adopted for software interoperability, data and document formats across central government?

  1. It may be inhibited from selecting the most capable standard in any particular circumstance (in the case that the current most capable standard does not meet the chosen criteria for openness).
  2. By adopting a relatively undemanding definition of openness, it may inhibit the emergence of standards for public sector data which meet a more demanding definition.

A3

For businesses attempting to break into the government IT market, would this policy make things easier or more difficult – does it help to level the playing field?

This depends on the definition of “government IT” and whether this includes markets for peripheral public services such as health and education.

With respect to these markets, the policy will help SMEs by giving them access to data which is currently controlled by dominant suppliers. But dominant suppliers have means of protecting their position other than by restricting access to data. The most effective (and socially useful) way in which SMEs can challenge market dominators is through disruptive innovation. By mandating consensual standards which reflect current orthodoxies, the policy will make disruptive innovation more difficult to achieve.

It might be argued that central authorities will be able to approve innovative standards as they emerge—but in practice, public authorities are unlikely to authorise innovative standards because:

  • in vertical markets serving public services, there is currently no satisfactory means of creating such standards;
  • innovation, which is easy to recognise in retrospect, is surprisingly difficult to identify at the time, when innovators who challenge current intellectual orthodoxies are commonly regarded as eccentric mavericks;
  • the great majority of stakeholders will always have interests in the preservation of the existing standard;
  • the current proposals commit the public authority to minimise duplication between standards, forcing it to choose between a safe, established, widely implemented standard on the one hand and a new, untested, experimental standard on the other.

A4

How would mandating open standards for use in government IT for software interoperability, data and document formats affect your organisation?

SALTIS has been formed to support the development of new data standards for learning, education and training (LET). To the extent that existing (in nearly all cases ineffective) LET standards were imposed by public authority, and particularly in the context of a regime which was intolerant of competing standards, the proposed policy would make it very difficult to develop any new or improved standards.

A5

What effect would this policy have on improving value for money in the provision of government services?

In respect of peripheral public services (such as education), the policy would seriously restrict innovation (see answer to A3). As education technology is still an immature market, in acute need of innovative solutions, the overall effect on efficiency would be damaging.

A6

Would this policy support innovation, competition and choice in delivery of government services?

Not in the medium-to-long term. By hampering innovation (which in education, is already at a low level) it will eventually also reduce competition (see answer to A3 and A5).

A7

In what way do software copyright licences and standards patent licences interact to support or prevent interoperability?

Software copyright licenses in themselves do not necessarily effect interoperability. Proprietary software commonly interoperates, as proprietary iPods plug into proprietary sound systems.

Efficient commercial markets will find their own virtuous equilibrium between what may sometimes be antagonistic requirements to protect proprietary business models and to deliver interoperability. Over-enthusiastic interventions which undermine intellectual property rights will simply undermine the case for business to invest in the development of innovative products for public service markets.

Standards patent licenses are quite another matter. The specifications which underlie interoperability standards should as far as possible be published under an open license, without being encumbered by any restrictive property rights.

This does not mean that the government should ban standards which do not meet these standards of openness (and which in certain contexts may offer superior functionality to open equivalents). Rather, in vertical public service markets, the government should encourage the development of competing standards which meet the highest standards of openness, allowing the market to act as the final arbiter between the two.

The suggestion that the government should drive out duplication from the standards ecosystem would prevent the adoption of such a pragmatic approach to established de facto standards.

A8

How could adopting (Fair) Reasonable and Non Discriminatory ((F)RAND) standards deliver a level playing field for open source and proprietary software solution providers?

Many products are not viable unless they can interact with other systems. The adoption of FRAND standards could help deliver a level playing field by allowing new, niche suppliers to sell their products freely, allowing them to integrate automatically with other systems, without having to pay “corkage fees” to their larger competitors (or being denied access completely). This holds true so long as:

  • appropriate, fit-for-purpose standards are specified;
  • the standards are correctly and consistently implemented, which is unlikely to be true without some kind of conformance and certification regime.

Providers of dependent, satellite services are commonly required to integrate with central infrastructures through proprietary (and often chargeable) APIs. This not only increases the cost of entry to the market for the SME suppliers of such dependent services, but also makes it difficult for anyone to challenge the position of the supplier of the central infrastructure, whose dominance is re-enforced by existence of dependent satellite services. These are effectively recruited into proprietary domain of the dominant supplier and become part of their competitive offer.

In education, examples of dependent suppliers are:

  • in the administrative domain: suppliers of cashless catering, registration, library management, assessment management and VLE systems, all of which depend on the MIS;
  • in the teaching and learning domain: suppliers of instructional software, which are dependent for their effectiveness in a classroom context on integration with a learning platform.

Appropriate open standards will level the playing field for suppliers of both dependent services and central infrastructures.

A9

Does selecting open standards which are compatible with a free or open source software licence exclude certain suppliers or products?

Possibly but not necessarily.

This is something of a confusing question, suggesting a lack of clarity regarding the difference between open standards and open source. A standard is not a piece of software and software itself should certainly not be standardised. All standards should therefore be compatible with both open source and proprietary software. A standard which insisted on an open source solution would obviously be compatible with open source but not with a proprietary solution.

This highlights a general principle, that interoperability standards should as far as possible not interfere with the internal implementation of compliant software systems, but should limit their scope to a particular exchange of data between systems.

Even file formats define the limits to which the functionality of a software system can be persisted between working sessions. For this reason, standard file formats will often restrict the functionality of competing products, making innovation more difficult.

The current emphasis on file formats is also being overtaken by the market trend towards cloud-based “software as a service”, where state is stored on the domain of the service provider and not in data stored on local hard-drives.

A standards ecosystem which focuses on plug-and-play, runtime interoperability between service providers (instead of file formats) will better reflect current technology trends, as well as being more conducive to innovative development of new software services.

A10

Does a promise of non-assertion of a patent when used in open source software alleviate concerns relating to patents and royalty charging?

It depends on the wording of the non-assertion promise. The rights to create derivative works should be included (see answer to A1).

This question only becomes important when it is assumed that the government is going to pick specific standards to mandate. In many circumstances, there will be a strong case for the use of de facto standards which do not meet the highest standards of openness.

A11

Should a different rationale be applied when purchasing off-the-shelf software solutions than is applied when purchasing bespoke solutions?

Yes—this difference is critical. For off-the-shelf software:

  • given that the software is unencumbered by support costs or advertising, the protection of property rights will be more important than for the developer of bespoke software, who will normally be paid for time spent and who will not therefore need to make a return on self-funded R&D;
  • conformance needs to be more strictly applied to ensure plug-and-play integration—while the developer of bespoke software can charge for ad hoc integration work;
  • standards compliance will not be specified in the course of formal tendering processes, so assurances about conformance to standards need to be given through other means, such as certification and badging regimes;
  • not being constrained by customer-written specifications and not being subject to the barriers that such costly procurement processes impose, off-the-shelf systems are more likely to be the vehicle for disruptive technology marketed by innovative SMEs—for the same reasons, off-the-shelf products are more likely to be disadvantaged by conservative, consensual, publicly-approved standards—and their success is therefore more likely to be dependent on opportunities for non-consensual, industry-led standards development.

A12

In terms of standards for software interoperability, data and document formats, is there a need for the Government to engage with or provide funding for specific committees/bodies?

Yes, there is a strong case for carefully targeted funding.

There is no need for government to support the development of generic, horizontal standards, which are adequately catered for by commercial markets and existing SDOs.
In the interests of competition and innovation there is an important public interest in the development of appropriate interoperability solutions for vertical, public sector markets such as education and health, where appropriate specifications do not at the moment generally exist and current processes are ineffective.

Because government is not suited to lead technical innovation, it must be prepared to pay the piper:

  • without calling the tune (in respect of directly developing and selecting specifications);
  • without relieving those who do call the tune of moral hazard;
  • without setting up prescriptive standardisation processes which prevent the market from being the final arbiter of which specifications attain the status of general standards;
  • without funding regulatory, “best practice” standards, which entrench existing practice instead of encouraging innovation.

Government should be wary of funding the development of particular specifications, but should rather establish an infrastructure to support industry-generated specifications, providing appropriate incentives for industry-led working groups to develop and implement open standards which:

  • meet the highest standards of openness;
  • are subject to the judgement of the market as to which will be successful.

The first of these reinforces the fact that a single definition of an open standard is unhelpful. While it will almost certainly be prudent to accept a lower standard of openness when listing a de facto standard that has already achieved general acceptance in the market; a much higher standard of openness can be required of standards developed with government support.

At the same time, if those standards are to be maintained on a sustainable basis, a supervising body should be allowed to develop an appropriate business model—and the Cabinet Office should consider what types of business model it wishes to encourage.
Government should also ensure that public service standards which gain traction in the market should be progressed actively by BSI through international standards organisations. Standards with a purely national scope are unlikely to survive in the long term—and the need to re-engineer software to meet an eventual global standard is likely to cause significant competitive damage to UK industry.

A13

Are there any are other policy options which would meet the described outcomes more effectively?

Yes. See answer to A12 above. The policy of mandating established standards on central government and its agencies needs to be balanced with a policy of encouraging the development of non-mandatory, industry-led standards in the peripheral public services.
Central government should support one or more “public service standards incubators”. SALTIS and its partners are currently discussing the details of such a proposal with the DfE.

B

Open standards mandation

B1

What criteria should the Government consider when deciding whether it is appropriate to mandate particular standards?

With regards to software for peripheral public services, mandation constitutes a significant bureaucratic intervention in the market. Given the very poor record of such interventions in the past, the presumption should be against it.

  • A mandate should only be required where anti-competitive practice is identified. In such cases, the Competition Commission should impose on named market dominators a standard which had already been proven by their competitors.
  • Government should be sure that it has the means and the will to make any mandate effective—nothing is more damaging in this respect (or more common) than for government to talk the talk but not to walk the walk.
  • The market sector concerned should be mature and unlikely to produce rapid technical improvements requiring the development of new standards, which would be inhibited by the bureaucratic imposition of what is bound to be a conservative standard.
  • The standard being mandated should itself be mature, sufficiently tightly profiled to ensure that it offers effective plug-and-play interoperability, and widely implemented.

B2

What effect would mandating particular open standards have on improving value for money in the provision of government services?

See answer to A5.

B3

Are there any legal or procurement barriers to mandating specific open standards in the UK Government’s IT?

Not to our knowledge.

B4

Could mandation of competing open standards for the same function deliver interoperable software and information at reduced cost?

Mandating several competing standards will obviously involve implementers in more cost than a single standard.

Putting aside the issue of mandation itself (addressed in answer to B1), a degree of competition between standards should nevertheless be tolerated because:

  • it is only by encouraging competition that standards can themselves be improved, particularly as it is always difficult to supersede an established standard which will tend to be favoured merely because it is widely implemented;
  • competing standards rarely address precisely the same functionality, but more generally represent overlapping functionalities, so that different systems may have good reason to choose different standards for slightly different purposes (such as GIF and JPEG graphics formats);
  • while uncontrolled proliferation of standards is undesirable, the cost of implementing two or three competing standards for interoperability is slight, as the main costs to software developers are in developing back-end functionality, not in creating different import/export filters.

A well-balanced standards ecosystem will therefore tend to have a handful—say, two or three—overlapping standards in any one domain. Like other cases of equilibrium, it is best if the correct balance is allowed to emerge by the operation of a dynamic system and not to be imposed by regulatory diktat.

It is true that public services are likely to achieve better value for money if software is procured in a market which is based on a healthy standards ecosystem—but “better value for money” will not necessarily translate into “lower cost”. If innovative software is developed which transforms the efficiency of public service delivery, then public services may decide to spend significantly more money on that software, increasing the total cost of software while reducing costs elsewhere. A regime which focuses narrowly on reducing the cost of software will tend to prevent the delivery of such efficiency savings in the wider public services.

B5

Could mandation of open standards promote anti-competitive behaviour in public procurement?

Yes. This will occur when a conservative standard supports the limited functionality of the products offered by market dominators, acting as a barrier to the introduction of new functionality by market challengers (as noted in answer A3, the introduction of revolutionary new functionality is the most effective means of increasing competition in an uncompetitive market).

B6

How would mandation of specific open standards for government IT software interoperability, data and document formats affect your organisation/business?

See answer to A4.

B7

How should the Government best deal with the issue of change relating to legacy systems or incompatible updates to existing open standards?

If it gets itself into the position of picking and mandating specific standards, it will have great difficulty in dealing with this set of problems. The decision to update legacy systems may often be painful and the cost/benefit equation will need to be assessed on a case-by-case basis. The decision to deprecate a whole generation of software all at once will generally be too painful ever to be taken by central government.

B8

What should trigger the review of an open standard that has already been mandated?

The government should avoid getting into the situation in which the continued use of a standard is dependent on the outcome of a bureaucratic review. Such reviews will almost always underestimate the importance of future innovation and over-value the interests of existing stakeholders. It will tend to shy away from the costs and disruption associated with wholesale deprecation of an established standard.

The only way around the problem is to avoid mandation wherever possible, allowing competing standards and creating an environment in which obsolete standards can fade away gracefully as they lose market share.

B9

How should the Government strike a balance between nurturing innovation and conforming to standards?

By creating the infrastructure and encouraging the commercial drivers to encourage the development and adoption of open standards, but refraining from mandating bureaucratically selected standards unless absolutely necessary (see answer to A12).
While it is true that standards provide a platform for innovation in product design, the standards themselves also require innovation, albeit at a slightly slower pace than do products. The rate at which both products and supporting standards need to be updated will continue to increase in line with the accelerating pace of technical developments in general.

B10

How should the Government confirm that a solution claiming conformity to a standard is interoperable in practice?

This is an important question. The experience in education is that bureaucratically-imposed standards invite avoidance and that, where this happens, the bureaucracy lacks the means to enforce conformance, leading to general confusion, the discrediting of the standard itself, and a growth in consumer cynicism.

The development of bespoke test harnesses and conformance regimes is expensive, acting as a barrier to the development of innovative and experimental standards.
The government should therefore support (where necessary through seed funding):

  • the creation of appropriate schema standards and supporting tools for defining data standards and testing their validity, providing generic test harnesses with a low marginal cost;
  • a badging / certification regime which has a sufficient marketing budget behind it to ensure that possession of a badge has commercial value to implementers;
  • a standards incubator which ensures that before any new badge is created, candidate specifications are backed by multi-lateral implementations and interoperability is demonstrated by informal plug-fests, with the cost of participation being borne by the standards developers / implementers;
  • before being formally recognised, candidate specifications and associated badges are required to progress through adequate beta phases, during which the success of real-world implementations is monitored;
  • the badging regime is backed by a robust legal framework, ensuring that any failure of interoperability is resolved rapidly or else the badge is withdrawn, with transparent reporting of all incidents and the processes by which they are resolved.

B11

Are there any are other policy options which would meet the objective more effectively?

Yes. See answers to A12 and B10.

C

International alignment

C1

Is the proposed UK policy compatible with European policies, directives and regulations (existing or planned) such as the European Interoperability Framework version 2.0 and the reform proposal for European Standardisation?

The proposed UK policy will only be compatible with European directives if the UK authority always chooses to mandate European standards over other alternatives (which may very well be preferable in terms of either functionality or openness).

In the field of learning, education and training (LET), European standardisation processes frequently create technical standards which have never been implemented by anyone at all.

The UK’s engagement in international LET standardisation processes is currently somewhat haphazard, being conducted by volunteers with much dedication but virtually no funding.

The fact that any decision to mandate will always be subject to European overview provides another reason why the UK government should avoid mandating standards. Instead, it should:

  • support the creation of industry-driven specifications for interoperability in vertical public services;
  • let the market choose which specifications should prevail;
  • ensure that specifications which have achieved traction in the UK market are promoted energetically within European standardisation processes.

C2

Will the open standards policy be beneficial or detrimental for innovation and competition in the UK and Europe?

In its current regulatory form, the policy will be captured by a frequently detrimental European standards regime that favours bureaucratic processes over market-led innovation.

The solution proposed by this response would reverse this polarity, protecting the UK industry from the imposition of European mandates, creating instead European and international standards which originate in specifications developed in a competitive UK market.

So long as they were pragmatic, effective and open, such standards would be beneficial to all those who used them and in the field of education technology, beneficial most particularly:

  • to the UK industry that developed the standards, helping to reinforce its leading position in what promises to be an important emerging technology market;
  • to the UK’s public education system, that would be first to benefit from the very significant, untapped potential of appropriate, technical solutions to many of the endemic inefficiencies in our education system.

C3

Are there any are other policy options which would meet the objectives described in this consultation paper more effectively?

Yes. See answers to A12, B10, C1 and C2.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s