Getting horky borky

Need an aide for conceptualising complex problems? Here's something that may help courtesy of a city building games and a tea loving man known as Biffa.

Getting a handle on complexity is simultaneously important and difficult. Being able to capture why something is complex is important because it can draw your eye to potential problems before they occur. However, analysing complexity is itself complex. There can be so many integrations, touch points or objects in motion that it becomes difficult to know where to start in analysing a complex system.

By 'system' I mean any environment where there are multiple components interacting with each other on an ongoing basis. So, I can see a DAM as a system. However, I also view an omnichannel environment as a system, the entire connected digital landscape, the users using it, a department, the company, and so on, as systems. All that changes is the scale or meta levels and, as a result, the complexity.

But this makes capturing how complexity scales difficult. We can innately understand that a system becomes more complex as its level of meta increases, but how much more complex? My belief is that it scales expotentially not linearly, but is that the case? I think in order to be able to understand that we need to:

1. Have a friendly frame of reference to aid understanding

2. Classify the different types of object in a system

3. Understand the different integrations in a system

4. Understand the different users

5. Create a scoring system to score the complexity of each object, integration and touchpoint in a system complete with an understanding of whether each additional object adds to complexity or multiplies it

6. Run the calculation to arrive at a total system complexity score

Sounds thrilling! So let's get started.

Biffa is a well known player of the game Cities: Skylines which enables you to build a city. Not just the housing, but the commerce, entertainment, tourism, transport, and industry. As he builds up a city, Biffa has frequent sips of tea while alternating between saying everything is “looking nice” or getting “horky borky”. This is a Biffa-ism for things are starting to break and go wrong. This can often come about as a result of traffic getting out of control. This can happen because, as in real life, industries have supply chains. This means that when you build one of the unique factories you also have to develop the materials it needs to make its products and the transport infrastructure to move them all.

Screenshot from Paradox

So, this is a clear illustration of a simple supply chain: making furniture. The complete furniture supply chain consists of:

• Growing trees in a tree plantations or sapling fields

  • Taking trees to make paper in a pellet plant or pulp mill

  • Taking trees to make planed timber in a saw mill or wood plant

• Taking paper and planed timber to a furniture factory to make furniture

• Taking furniture to shops or export for sale

So, at a bare minimum, we need five steps of transport to move everything. If we keep things simple and just use trucks, then we need roads between all the different locations. So the roads and all the junctions need to be capable of supporting all the additional traffic. This means that having a city with just one unique factory requiring one type of industry makes for a complex system.

In a MarTech system we could view it as:

• Making products

  • Capturing operational product data in SAP

  • Capturing technical product data in PLM

• Using an integration to push SAP and PLM into PIM

• Distributing PIM data to e-commerce to sell products

We can analogise roads to system integrations because the data traffic needs to flow through the entire system.

Let's take a breath to recap the original listof needs. We have:

1. A friendly frame of reference to aid understanding

2. Classified the different types of object in a system

3. Understood the different integrations in a system

Now, let's get into 4 - 6.

In MarTech, we could also expect each system in the chain to have at least one user group, possibly two. For example:

• Product: design and manufacture teams

• SAP: technical and marketing teams

• PLM: technical and marketing teams

• PIM: marketing team

• Stores: e-commerce team

If we’re building a digital supply chain from the ground up and don't have to worry about legacy, then we might see SAP and PLM in parallel. Each is integrated with the PIM. The PIM is then integrated with the e-commerce platform.

Now, we can see that our meta system encompasses:

• Platforms

• Users

• Integrations

Now, we need a unit of complexity. With gratitude to Biffa, I’m going to call it the ‘horky borky’ or HB. To calculate our total HB for a system, we need to allocate a number of HBs for platforms, users and integrations.

• Let each product feature = 2 HBs

• Let each additional product feature add 1 HB to the product score

• Let each platform = multiply by 2 HBs

• Let each additional platform feature add 1 HB to the platform score

• Let each additional platform at the same level be added to each other

• Let the first user group for each platform multiply by 2 HBs

• Let each additional user group add 1 HB to the user group score

• Let each integration multiply the HB score by 2

What does this mean for our supply chain? Let's suppose that we have a company making a product that has three features, that our SAP, PLM and PIM systems each have two features, that the SAP and PLM are at the same tier, and that the user groups are as laid out above. For us, with the product to SAP and PLM to PIM to e-commerce chain, our calculation would be:

• Product HB score = 4

• Product user group HB score = 3

• Total product score = 12 HBs

  • Because we have a product with three features and two user groups creating it

• 12 HBs integrated with SAP and PLM = (12 x 3 x 3 ) + (12 x 3 x 3 ) = 216 HBs

  • Because we have a total product score of 12 HBs being sent to 2 systems at the same tier which each have a platform score of 3 and a user group score of 3

• 216 HBs integrated with PIM = 216 x 2 x 2 = 864 HBs

  • Because there’s one more integration with one user group

• 864 HB’s integrated with e-commerce = 864 x 2 x 2 = 3456 HBs

  • Because there’s one more integration with one user group

Now we have a total score of 3456 HBs. Wow, that’s nuts! (Biffa might say).

My original assumption was that complexity would scale exponentially. Let's test this. A second assumption above was that each of the SAP, PLM, PIM and e-commerce systems would have 2 features giving each platform 3 HBs for a total of 12 platform HBs. What if the PIM had an additional feature? Then, we’d have a total of 13 platform HBs for an increase of 8.33%. However, if we run the calculation again, then the total HB score becomes 5184 HBs for an increase of 50%.

What if the product had 1 more feature for a total of 15 HBs? Then, we’d have a total score of 4320 HBs for an increase of 25%.

This is effectively the smallest change that we can make to the total supply chain yet it produces a disproportionate increase in complexity. The potential impact of adding an additional system would be substantial and we’d need to because the meta system I’ve used as an example is simple to say the least.

And, in case it needs to be said, each integration also needs to be viewed as a potential point of failure. If it does fail, whatever the reason, then everything downstream of it also fails. Just as Biffa wouldn't be able to get timber to the saw mill if a junction was horky borky, you wouldn't be able to get SAP data to the PIM.

You might quibble with my formula or my logic. You might think the whole thing is horky borky. But I never said this was a scientific metric. Instead, I’ve presented it to help us conceptualise how quickly complexity scales.

Biffa takes his industries frm horky borky to nice with tea