ISSS Variety Comparison Grid Symposium: Belated answers to the text chat

There were a lot of points raised in the chat that are worth coming back to, and that weren’t addressed in my ‘First Lessons’ essay. Let’s do this. (I have de-identified the questions when presenting them here as an expedient workaround for

How does this include solutions that attenuate variety in the problem? For example, classroom management where the teacher with their limited variety deals with the complicatedness/variety that 30 children bring in by limiting what happens.

This is a really instructive question, because there’s more than one way to answer it. We could say that “The job of the teacher is to ensure that student behaviour is appropriate to the classroom.” If that’s the case, then it’s a layer 2 job of supervising immediate behaviours.

Alternatively, the job of the teacher could be to prepare lesson plans (layer 2), in response to a layer 3 curriculum that specifies what students must be taught.

The teachers I know tend to describe the job in terms of introducing ideas to people who have never encountered those ideas before. That’s layer 4 work.

Then we have the problem that standardized testing is a layer 3 entity. Whereas layers 2 and 4 exist in the relationships and the space between entities and are difficult to observe, layer 3 is highly visible. The desire for convenient visibility gives us this push towards the creation of highly visible paperwork — the gap between the work a teacher actually does and the standardized tests is a driver of frustration.

Need some kind of search/feedback strategy

If the grid was intended to specify a system that addresses the needs of an organization, then this would be true. However, a system specification is a layer 3 entity and the OE tools are designed to exist at layer 4. Therefore, they facilitate the asking of a question, instead of providing answers to questions.

Feedback can exist at all of the cognitive layers:

Layer 1: The act of listening to someone

Layer 2: Monitoring someone’s work with a view to making interventions as appropriate

Layer 3: A formalized feedback system

Layer 4: Cross-domain conversations that include feedback

Layer 5: Forming perceptions about feedback

A bit like Snowden’s regions of complexity, creating a separate universe, with all the zones hypothetical. Complexity is NOT meant to refer to the properties of the observers machination.

I’ve had a few conversations with Dave Snowden about this. My understanding of his view (which is probably a misunderstanding) is that he believes ‘complexity vs complicatedness is inherent in the situation’. My view is that ‘complexity vs complicatedness is an assessment made by a person with a perception of a situation’.

Consider the statement “Database authentication was delayed because the operating system was unable to evaluate if the client PC was one of the trusted hosts based on local data and had to wait for a DNS timeout.” Depending on your systems administration experience, that may or may not be a ‘simple’ statement.

There seems no relation between the mental models and the systems that may or may not exist other than mental models.

The question “Do systems actually exist in the natural world, or are they figments of human imagination?” is excluded from the scope of Organization Engineering. To attempt an answer is to invite controversy.

This is jogging thoughts about its similarities to the IT OSI network layer model and the 7 layers of knowledge management.

Well spotted. I was working for a telecommunications company while developing the model, and the OSI model was a constant part of my background thinking.

I don’t think complexity comes out of explanation, but the weaving and folding of the material systems, that never gets discussed in this framework.

This is another effort to avoid controversy. The notion that complexity/variety ‘comes from’ somewhere is outside the scope of an Organization Engineering analysis. It is enough to know that complexity/variety exists and can be measured. I’m attempting to be Newton (and claim that gravity exists and has predictable influence). I’m happy to leave Einstein’s work to someone else.

Situational Complexity can be due to one person’s cognitive limitations, but in most systems methodologies, we define it as the complexity perceived by a particular group of stakeholders. Similar to the case of individuals, also in the case of groups, situational complexity is related to that particular group. Another group might exhibit a different situational complexity.

Agreed. The intent of these tools is to empower the person using them to make assessments about a situation (including the relevance of tools to solving a problem in that situation). The ‘human response and performance’ sections of the theory make some predictions about how individuals will perceive something based on their competence. Something for another session, perhaps.

If we could measure the degree of weaving and folding necessary to organise the system then we might be able to quantify the complexity. That’s what Tom Stonier attempted to explain in his books.

Sounds like a worthy endeavour! For me, identifying observable qualitative discontinuities was exhausting enough. Good luck to Tom Stonier and those who continue his work.

How and how decides in which box a problem/task goes?? Is there any objective method for doing that?

No. The tool is intended to support ‘structured subjective’ assessment.

He does keep asking “if it will work” but then leaves that out of the model.

Yes. I found that ‘best possible performance under ideal circumstances’ was a sufficiently useful measure, especially when combined with ‘order of magnitude cost to implement’. Don’t use the Grid to select a tool: use it to develop a shortlist for further examination.

I would like to see a real-world problem be solved using this grid.

Me too. I got within inches of using it to restructure the IT function of a large client, but the deal fell through due to a freeze on external contractors during the Covid19 crisis.

Getting the feel for the right level of communication to have with someone can benefit a little research but I think must be tested in some way naturally in conversation.

Agreed. I apply it to my own conversations so often that it has become automatic, and I’m convinced it helps. If only that were compelling! I’d love to work with a proper scientist to tesst this.

The framing totally omits the normal kinds of complexity in environments, the natural centers of organization that generate the emergent properties with connections, that seem might all be in all the categories of the boxes Nick shows.

Yes it does! I’m powerfully influenced here by the five element theory of Chinese medicine. It’s a framework that facilitates extremely nuanced diagnosis — paradoxically, it achieves this by radically simplifying organic processes down to the interactions between five interrelated systems. Had we a lot more time, we could have introduced some of the other models and reflected more of the complexity.

Q: But, all things seem to be maps with no territory, what is one to do with that?

A: It could be presented as focused on maps

Yes, it’s focused on maps. The business of ‘checking the map against the territory’ is extremely important, but to say that more than that is to disappear down the rabbit-hole of actually checking maps. There’s a tension here: in order to have a conversation that is as ‘layer 4’ as possible, we have to acknowledge the potential validity of all maps. Then there’s the problem that layer 4 is fundamentally either ‘a space where the maps conflict with each other’ or ‘a space that isn’t mapped at all’. And so a toolset that aims to help people operate in layer 4 territory can’t talk too much about checking the map against the territory because there is no map that can fully prepare someone for operating in uncharted territory.

The fundamental assumption is that there is in fact or indeed an agreement that there is ‘A problem’…

We could say: the layer 4 goal “Demonstrate how to compare candidate solutions against a problem” permitted an assumption at layer 3 that there would be a problem, and that there would be at least one candidate solution.

If the presentation had a different goal, the assumptions permitted for tool selection would also have been different.

And the very use of the term ‘problem’ suggests it is something that can be ‘fixed’… which implies that systematic assumptions are infusing the design of the model

Yes. We didn’t really cover it in the presentation, but a core motivation behind the layering model is to observe the manner in which systematic (and systemic) assumptions create organizational constraints. By observing where the assumptions are brought into being, we gain some power with respect to the assumptions.

the challenge is hybrid balancing of concrete and abstract with some ‘meta-methodology’

Yes. As a ‘stacked discipline’ (a discipline that rests on other disciplines), systems science is particularly vulnerable to becoming a ‘floating discipline’ (a discipline that has no meaningful connection to reality).

No one is actually “in power”. They may be “in control.” Power is inversely correlated with violence. So brutality → little power. Cooperation and creativity → grows power.

I agree. One of my hopes is that Organization Engineering will help people to achieve cooperation and creativity, and that the organization who do this will out-compete organizations that rely on brutality/coercion.

My functional definition of “reality” is “what works”.

Mine too. For all my flaws, I think I’m pretty good at getting things working. I suspect that my definition of reality/truth has something to do with that.

Well, first we need to come up with a definition of “opinion.” It seems to me that “opinion” is a belief or proposition that someone doesn’t give much weight to.

My opinion, given in a court of law, once led to a criminal conviction. There are people who treat their opinions as disposable. I polish mine and treat them with care.

I see a lot of overlap with Elliott Jacques work but it had its beginnings in deciding a rationale for paying some people more than others and settled on time span of control. That works very well in a factory setting but becomes difficult in an information economy.BTW: In the VSM, Beer uses recursions to deal with level appropriate conversations. The easy example, like a bus, is government at the municipal, county, national levels.

Yes indeed. In my submission describing what I would present, I laid out the scope of Organization Engineering, and also mentioned the items from Requisite Organization that are excluded from OE.
The following Organization Engineering concepts are required for use of the Grid, and will be discussed:
Variety (as inherited from Stafford Beer)
Cognitive Layering, layer numbers, layer names, artefact category names
OE definitions of ‘Science’ and ‘Engineering’

The following Organization Engineering concepts are not required for use of the Grid, and are deferred until the workshop in South Africa:
Implications for dispute resolution
Performance appraisal concepts (Ladder of Trust, Penetration, Reach)
Naturally-occurring glass ceilings
Properties of Domains (Stacked, Floating, Bridging, Hybrid)
“The task of management is to translate, supervise, integrate, and protect”
Meaningful layer counts beyond 5 due to variations in the unit of conversation
Descriptions of Language (Ordinary, Jargon, Occult)
Concepts inherited from Jaques (Judgement, Meaningfulness, Task Initiating Role Relationships, Manager Once Removed)

The following concepts from Jaques are excluded from Organization Engineering, and are not part of either planned presentation:
Felt Fair Compensation, Strata, Maturation Curves, Mathematical Formula for Work Output, Definition of ‘Manager’, Timespan of discretion