An Inquiry into Liberating Structures (part I)
Liberating Structures claims to boost innovation, accelerate and improve the quality of implementation, and enable rapid adjustments to change. Does it deliver on its promises?
Do we Need Another Framework?
A business associate emailed me asking, “Are you excited about Liberating Structures yet?”
I wasn’t, even though a lot of people are talking about it on my LinkedIn feed.
Straight away, the name Liberating Structures (LS) strikes me as presumptuous. When Toyota needed a name for their framework, they called it the “Toyota Production System”. They didn’t claim it to be generally applicable. The simplicity and modesty appeals to me.
My other reservation is that LS aims to improve collaboration in command-and-control-led organisations, which have a history of distorting new frameworks beyond recognition. These are the people who brought us “Product Owners” that are merely rebranded project managers, and abominations like the Scaled Agile Framework that are “agile” in nothing but name.
Still, a framework shouldn’t be faulted for what people do with it. And there are reasons to be curious about LS.
For one, LS is a pattern language, which means its creators, like me, must be fans of Christopher Alexander’s work. I guess that puts us in the same crowd. There might be some neat ideas in it.
So, let’s keep an open mind, take a closer look at LS, and see what it has to offer.
What is Liberating Structures?
Liberating Structures is a collaboration framework developed by Keith McCandless, founding partner of the Social Invention Group, and Henri Lipmanowicz, a former Merck executive and co-founder of the Plexus Institute.
According to the FAQ page on the web site, Liberating Structures are:
Thirty-three adaptable microstructures that make it quick and simple for groups of people of any size to radically change how they interact, coordinate tasks, and work together.
LS can replace or complement the big five conventional approaches that people use all the time: presentations, managed discussions, status reports, brainstorms, and open discussions.
In contrast to the big five, LS are designed to include and unleash everyone in shaping their future.
So far, so good. LS is a way to exchange ideas and work together while involving everyone’s voices and talents.
But why do we need “radical change”? Let’s take a closer look at the problem.
What is the Problem that LS tries to solve?
Quoting from the LS home page:
So why is it that so many organizations of all stripes are filled with disengaged workers, dysfunctional groups and wasted ideas?
Unwittingly, the conventional structures used to organize how people routinely work together stifle inclusion and engagement.
Conventional structures are either too inhibiting (presentations, status reports and managed discussions) or too loose and disorganized (open discussions and brainstorms) to creatively engage people in shaping their own future.
They frequently generate feelings of frustration and/or exclusion and fail to provide space for good ideas to emerge and germinate. This means that huge amounts of time and money are spent working the wrong way. More time and money are then spent trying to fix the unintended consequences.
The problem statement is clear. Traditional ways of collaboration fail to involve many people, leading to their ideas being overlooked and them feeling left out. This is a waste of talent, time and money.
We can all relate to this. We’ve all been in team meetings that leave us drained and achieve little. We’ve all seen good ideas being ignored for the wrong reasons. If there’s a way to fix this, we’re listening.
So, what does LS promise to do about the problem?
What does Liberating Structures Promise?
The FAQ page says:
LS stimulate innovation and productivity at all levels.
For a large class of management challenges, too few people are included in planning and coordinating a response.
Engaging more people at multiple levels, earlier and more strategically, can dramatically boost capacity for solutions that generate spectacular and unexpected results.
LS answer the question “how can we engage the ‘vast majority’ practically and cost-effectively?” Much more is possible as trust and shared ownership increase.
The promise is becoming a bit more clear: LS taps into an organisation’s collective intelligence in order to improve business outcomes.
Further, the FAQ page states that Liberating Structures:
- improves decisions;
- boosts innovation;
- accelerates and improves the quality of implementation;
- enables rapid adjustments to change.
While these are compelling claims, they are also quite broad. A more detailed explanation would be useful.
What Makes Liberating Structures Work?
Let’s again start with quoting from the FAQ page:
LS employ micro-structural design elements that distribute participation, engaging everyone in shaping their future. These novel structures guide new behavior.
LS are more unit-based and local, with solutions worked out by front-line groups in partnership with leaders instead of imported “best practices.” LS grow through informal social networks and decentralized communities-of-practice rather than the organizational chart via buy in initiatives.
The structure side of LS make it easy—and safe—for all participants to express their views freely and fully.
There is no control on the content of group conversations. Instead results emerge bottom-up from the whole set of interactions liberated by LS.
Rather than serial processing in one large group, LS utilize parallel processing among individuals, pairs and small groups.
By design LS distribute control so that participants can shape direction together as the action unfolds.
Though a bit verbose, this passage does provide some clues on what’s going on. Let’s condense it a bit for easier understanding:
- Participants work in small groups on solving different parts of a problem at the same time.
- A safe environment encourages everyone to share their contributions in full.
- The process is not influenced by authority or organisational structure.
- The complete solution develops organically, without central control.
This approach reminds of Christopher Alexander’s Notes on the Synthesis of Form, which lays out how good architecture unfolds organically from underlying needs, guided by local rules. Notably, McCandless and Lipmanowicz use the same term, “unfold.”
So LS seems to be a decision synthesis framework, perhaps inspired by Christopher Alexander’s form synthesis theory (even though the web site does not mention his name).
The home and FAQ pages also provides some analogies to clarify the principles behind LS.
The Wikipedia analogy was helpful to me:
Wikipedia’s must-dos and must-not dos specify how anyone can write articles, edit content, reach consensus about the facts, and share with attribution. This structure makes it possible for a diverse community to generate and sustain accurate content that compares favorably with professionally edited encyclopedias.
Then there is a link to a podcast about emergent behavior and collective intelligence in ants. The question is of course if the same principles work for humans - after all, we have evolved to develop strong individual intelligence and communication skills.
According to the FoldIt study mentioned on the site, they do: collective intelligence was able to beat experts in understanding how a protein causing AIDS in rhesus monkeys folds. (Here is a link to the research paper.)
These examples, while thought-provoking, are not about LS itself. We still need more details. Let’s press on and see what else we can find.
Again quoting from the FAQ page:
Building on a few methods introduced in EdgeWare in 1998, LS draw from emerging insights from complexity science, organizational development, improvisational arts, and user experience.
That doesn’t give us much to go on, but a linked document lists ideas from books that served as early inspiration for LS (I added links for easy access):
- Joanna Macy, Coming Back To Life (1998)
- Kevin Kelly, New Rules for the New Economy (1998)
- Gareth Morgan, Images of Organization (1998)
- Ralph Stacey, Complex Responsive Processes (2001)
- Reuben McDaniel, Ruth Anderson, Healthcare Review (2000)
- Glenda Eoyang, Coping with Chaos (2001)
- Brenda Zimmerman, Plsek, Lindberg EdgeWare (2001)
The document, however, doesn’t explain how the ideas from these books were used in LS.
In a table, we expect cells in the same row to belong to the same entity. In this case, that logic doesn’t appear to work out.
Consider, for example, row 7. It might suggest that the LS concept of “Learn by Failing Forward; Build Trust As You Go” was inspired by the concepts of:
- “Feed the Web: maximize the network” from Kelly,
- “Improvising: intuition guides via min specs” from McDaniel, and
- “Order without central control” from Zimmerman.
Which doesn’t seem to add up. This table doesn’t help us understand how these foundational ideas were grouped into LS principles.
Google’s “Project Aristotle”
Going back to the history page, it mentions:
Many LS users have suggested that the Aristotle Project conducted at Google helps explain the science behind LS.
This sounds oddly half-hearted — why not just say whether Google’s research applies or not?
The findings from Project Aristotle shed some light: support for LS’s philosophy is mixed. While evidence about psychological safety is strong, there is none to support consensus-driven decision making, a concept close to LS’s principle to “include and unleash everyone”.
In other words, one of LS’s core assumptions is potentially flawed.
Superbug Case Study
The web site has twelve case studies from healthcare, law, business and academia.
An interesting case is Front-Line Ownership: Generating a Cure Mindset for Patient Safety. It details an intervention study in Canada, where LS techniques were used to involve healthcare workers in preventing hospital infections with MRSA and other “superbugs”.
Although the final publication is behind a paywall, a pre-release of the paper is available online.
The research team found that infection rates were cut in half compared to hospital units not involved in the study. Two units however did not report an improvement.
Unfortunately, the researchers did not disclose their study design or data, so we can’t verify these outcomes. Which is important, as most published research findings are false.
Other scientific evidence on the web
Searching for evidence on the web yields a handful of paywalled publications discussing LS, but no controlled trials.
In addition, there is one limited pilot study in Norway. But it was in a specific academic setting, it also wasn’t a controlled trial, it relied on self-reporting, and didn’t use proxy questions to avoid bias.
None of these papers convincingly demonstrates LS’s claims.
At this stage, I’m undecided about Liberating Structures (LS). The concept of a pattern language for decision making is intriguing, and it’s clear that LS has an enthusiastic following.
However the framework seems low on documented evidence, which is concerning. If LS is as effective as its proponents claim, you would expect more studies and more use cases with hard data. Not to mention a major new invention, or someone becoming a billionaire.
Yet, I’m giving LS the benefit of the doubt for now. In the upcoming posts, I’m going to look into some of the 33 patterns, to see if I can gain a deeper understanding.
In the next post, we’ll explore the popular 1-2-4-All micropattern.