When Openness Meets a Boundary

Sharing more is not always the more open thing to do.

Sometimes a useful public account depends on something remaining private. A name is withheld. A dataset is restricted. A resource is shared, but only with the author’s permission. A report is made public while the individual cases behind it stay confidential.

Those can look like failures of openness if openness is treated as a simple instruction: make everything available.

Clyde Street contains a more difficult problem.

What happens when the decision to share meets a boundary that should not be removed?

Privacy was there early

In November 2009, Keith was preparing a presentation for the National Elite Sports Council Forum in Canberra.

He was already working in a strongly public mode. He had posted a draft and presentation online, was using Creative Commons images and hoped to blog during the event.

But the post did not present openness as an uncomplicated good.

Keith had been reading Graham Attwell, who argued that open education, open discussion and a culture of sharing had to coexist with people’s ability to manage their digital identities and privacy.

That warning belonged to Attwell. Keith’s move was to bring it directly into a post about his own plans to share publicly.

The timing matters.

The surviving trail does not support a story in which privacy appeared only later, after an initially unqualified openness. It was visible inside the openness conversation almost from the start.

That makes the archive harder to summarise with a slogan.

Keith could value open knowledge and still encounter a limit on what publicness should mean.

Open does not mean ownerless

A few months later, another boundary became concrete.

Keith had been working on the proceedings of the 2009 International Association of Computer Science in Sport symposium. In April 2010 he created a public repository for additional material from the event: abstracts and twenty-three PowerPoint presentations that had not been included in the main SlideShare collection.

He wanted them to become open-access resources for students and colleagues.

But he included an important condition.

The materials had been added with the permission of their authors.

Making the repository public did not make the authors’ work Keith’s or Clyde Street’s.

Two years later, the same issue returned from another direction.

In “Sharing Openly and Open About Sharing”, Keith followed a trail from Richard Byrne to Sue Waters about copied blog content, reuse and plagiarism. The argument he encountered was not against sharing. It was about what sharing still owed the person who had made the original work.

Keith connected that discussion to Creative Commons licensing and to what he was encouraging students to do with their own e-portfolios.

The point was not that openness required tighter possession of everything.

It was that access did not cancel labour, attribution or authorship.

A resource could be generously shared and still remain somebody’s work.

That distinction matters later in this trail, where presentations, data and conversations remain other people’s contributions even when Keith makes them reachable.

A public event can still contain restricted data

By 2017, the boundaries were becoming much more operational.

In February, Keith wrote about a data hackathon hosted by the Western Bulldogs and the City of Ballarat.

Keith framed the Ballarat event as an opportunity to be innovative, after recalling an earlier data event he had praised as an adventure in sharing and exploration. The Ballarat post also stated plainly that some of the data being used would not be in the public domain and that attendees would sign a non-disclosure agreement.

Nothing in the post suggests Keith treated that as a contradiction requiring repair.

The event was being organised for participants to work with data while some of those data remained outside the public domain.

The boundary in this case was formal: a non-disclosure agreement.

Other boundaries in the same year were much more relational.

A name can stay out of the record

In April 2017 Keith wrote about one of the coach-education Gatherings he had attended in England.

He described the people, the ideas and the way the conversation sent him off on intellectual tangents. But he also told the reader something unusual for a blog built around noticing and remembering.

Then, when he came to an idea shared by a friend working in an Olympic sport, Keith withheld the person’s name. He explained why: he had not asked explicit permission to identify people other than the organisers.

That missing name is easy to misunderstand.

From the perspective of an archive, more detail would be useful. A future researcher might want to know who spoke, what their role was, whether the idea appeared elsewhere and how the network around the Gathering was formed.

But the post gives us no licence to reconstruct the identity simply because the research would become richer if we could.

The missing name records a boundary Keith placed on his own publication because permission had not been asked for or secured.

This is one of the sharpest distinctions in the whole trail.

An incomplete public record is not always a defective public record.

Sometimes it is incomplete because permission to identify a person has not been asked for or secured.

Transparency can still stop

A few months later Keith was thinking explicitly about reproducible computational research.

He followed work shared by Mara Averick to Joris Muller and, through Muller, to the reproducibility rules proposed by Geir Sandve and colleagues.

The rules emphasised traceability, version control, raw data, scripts, intermediate results and public access to the materials behind an analysis.

Keith admired the approach. He saw reproducibility as important for sport analytics and wanted to make his own processes more explicit.

But the source he was following also contained a small qualification: the workflow met most of the reproducibility criteria except, among other things, complete public access — because everything could not be shared.

The value of that detail is not that it settles what reproducibility should require.

It prevents a false binary.

A workflow can become more inspectable without every component becoming public.

Openness can increase while disclosure remains bounded.

Confidence is not a gap to fill

Later in 2017, Keith spent time with two coaches whose work he had followed closely.

He described the access as a privilege. He asked one question about how they had changed the energy in a team and wrote that what they shared with him, in confidence, fascinated him.

The post tells us almost nothing about the substance of that confidential answer.

Instead, Keith wrote about what it was like to be near the coaches’ practice and to watch the team perform.

Again, an archive-minded reader might feel the pull of the missing material.

What exactly did the coaches say? What had changed? What did Keith hear that altered his interpretation of the performance?

Clyde Street does not answer those questions.

And in this case, the silence should remain a silence.

The post marks the exchange as confidential. That gives us no basis to treat the missing substance as a gap the archive is entitled to fill.

That is a different kind of boundary from the hackathon NDA. It is not a legal instrument recorded in the post. It is relational confidence.

The two should not be collapsed into one general rule.

But both show why an instruction to make the process public can be inadequate.

Sometimes the route into understanding passes through material that is not ours to expose.

Consent inside possibility

In September 2018, the problem became more explicit again in a post about a new student portal at the University of Canberra.

Keith had attended a presentation by his colleague Scott Nichols about a system intended to support students from course choice through enrolment, study, graduation and later connection.

He was excited by its potential for personal learning journeys.

But the presentation had been shared in confidence, so Keith said he could not provide the platform detail before launch.

He then moved to a second boundary.

A system like this would create a data-rich environment. Keith imagined that data being used to support personal learning — but he placed informed consent inside the possibility. He also wrote about the need for appropriate checks and balances.

The sequence matters.

The existence of useful data did not itself establish permission to use them.

The potential benefit did not erase the question of who had agreed to what.

For this trail, the distinction is clear.

The technical possibility of using the data was not the same as permission to use them.

Public learning without public cases

Two days later, Keith published one of the clearest late examples in this trail.

For five years he had been a critical friend to coaches, teachers and educational technologists. His conversations with coaches addressed the public nature of their work and the implications for families, friends and themselves. His conversations with educational technologists took place during organisational uncertainty and discussed physiological and psychological impacts.

Keith wrote that the conversations were held in confidence.

He had received funding for some of the work, but he had been clear that there would be no reporting of individual cases. Reporting would be generic.

On #RUOK Day in 2018, he shared parts of one of those generic reports.

The distinction is important.

The educational-technology conversations could inform a public account without becoming public cases.

The individual cases did not have to be exposed for a generic public account to be made.

This is not a perfect solution to every problem of representation. Generic reporting can still distort, select or oversimplify. Confidentiality does not guarantee that the resulting account is correct. Nor does the post tell us enough to evaluate every consent or governance arrangement behind the work.

But it does make the boundary visible.

Keith had drawn a line between what could inform the report and what the report was entitled to reveal.

The boundary belongs in the account

Across these posts, the forms of boundary differ.

There is privacy in digital identity. Permission from authors. Attribution and respect for labour. A non-disclosure agreement. A deliberately unnamed contributor. A reproducible workflow that cannot expose everything. A coaching conversation held in confidence. A confidential presentation. Informed consent. Generic reporting instead of individual cases.

They should not be flattened into one policy.

Clyde Street does not provide a universal disclosure framework, and the sources do not justify turning Keith into a personality portrait of the discreet or ethical sharer.

There are also other openness problems elsewhere in the archive that this article is not trying to absorb. A resource can be technically open but inaccessible because of bandwidth, disability, language, format or platform failure. Open data can still be invalid. Participation can remain unequal. Public material can disappear.

Those are real Clyde Street trails, but they are not the same question.

The narrower trail here is about entitlement to disclose.

Again and again, Clyde Street contains cases in which the useful public object is made by sharing something — while leaving something else protected.

That does not make the work less open in any simple sense.

It suggests that a boundary can sometimes be part of the conditions under which sharing happens.

The reader can see what was shared, what remained another person’s, what required permission, what was held in confidence and where the public record deliberately stops.

A boundary, in other words, does not always mark the point where openness failed.

Sometimes it marks the point where sharing remained bounded by another person’s work, permission, privacy or confidence.