Systems Thinking Ep. 8 - Networks

hello and welcome to my next episode of systems thinking where today we will cover networks so a quick preface of today's video we're going to start by talking about Network systems we're going to characterize them and talk about the components of network systems in the second part of the video we're going to talk about Network effects such as the behaviors and failures of networks and then finally we'll wrap up with a few examples of networks so that you can really get your feet wet and and be grounded in some some real world examples and and

start to apply your new systems Thinking Tools around networks okay so first and foremost Network systems what is a network the definition of a network is simply internet interconnected components and their interactions so there's a few a few things that go into this definition first is interconnected components so nodes or elements in a system there are complex relationships between these different nodes there might be different kinds of connections and feedback loops and other things that create aggregate or emergent behaviors in the networks so holistic view so one thing that you need to keep in mind

is that a network is a type of entity that emerges from the collection of independent components it is a kind of system that takes on a life of its own uh there's some adaptability and other emergent properties that come from networks so adaptability refers to a Network's ability to change some networks are very very structured and rigid While others are very plastic and changeable that's what adaptable means moving on nodes so networks are fundamentally constructed of nodes these are individual components or elements within a network that have their own individual functions or identities so let's

talk about identities the identity of a node is the the attributes that uh that basically Define that node within a network so that might be a person it might be a computer it might be a city and a network it might be an institution in a government each node has its own function so in the case of this graphic here you might have an engine assembler so as a node in the network that builds cars and engines you might have someone who's responsible for bolting on carburetors you might have another person who's responsible for painting

the outside of the of the car body that sort of thing connections so one thing that you need to understand is that nodes are connected to each other there's all kinds of connections that can go into this it can be informational connections it can be physical or material connections so for instance in within the system of a car there are many physical linkages such as belts and chains that that interconnect the different parts of the network the system of the vehicle in information that works however such as on the internet those those connections those inner

links are purely digital they're only information in financial systems it is money that is that is uh the connection you know money moves from one account to another as part of financial transactions uh local rules so basically the local rules are the guidelines and behaviors that dictate or characterize uh the constraints and protocols that an individual nodes follows so for instance uh as a lawyer you have certain rules that you have to follow as a participant in the legal system if that makes sense and then finally state so nodes can change State and what I

mean by that is say for instance you're working in a large Factory and a machine breaks down so a state goes from operational to broken down or defunct um in a longer term system you might have things that are in different phases of their life so life cycle state is another thing so something might be brand new it might be old it might be time for retirement and so I'm thinking in terms of software and Hardware in those kinds of systems and networks there okay so moving on uh from nodes we have interfaces so another

thing that you need to understand is the interface so the interface is the linkage between nodes in a network so there's different kinds of connections uh this might be information energy material those are those are the generally the three kinds of Connections in any network so in a computer network the interface is all about information so you might be sending data you might be sending commands API calls or that sort of thing in a power grid you're sending electrical energy and other kinds of uh and other kinds of Networks you're sending mechanical energy so for

instance a car that is a network that has a mechanical energetic connections and then of course across factories and delivering of of goods so like if you've got an Amazon van um there's material being moved through the system so then the second part is directionality so many interfaces are bi-directional uh meaning like say for instance I give you money you give me a bar of soap right there is a bi-directional interface there in some cases it's not sometimes it's one way uh with extractive Networks you're basically you're going into a coal pit you're extracting coal

and you're not putting anything back in so this is another thing to keep in mind when looking at networks and network systems uh Transformations so a transformation is the product or that that information or material that's moving through a network um sometimes the Transformations happen within nodes but sometimes it happens between nodes and so what I mean by that is that uh once something is transmitted from one location to another it might change form or meaning and so in this case what what that means is is let's take example of the Amazon delivery guy uh

when the when your package is in the Amazon delivery truck it is Cargo once it arrives to you it you know the the cargo status changes and now it is transformed into a product or a delivered good um so that's what I mean by transformations in other cases uh the way that information is handled changes as it moves through a network so for instance your email might be encrypted while it's in transit and so in that case it's just it's an encrypted payload but then once it gets to you the content of the email is

transformed and it's decrypted um and then it it takes on an entirely new meaning so constraints interfaces often have constraints such as time distance or throughput which is so in this case like it takes time to deliver the package from the Amazon warehouse to you or it takes time to deliver the email from one place to another depending on the size of the attachments so that's what I mean by constraints so these are usually time energy material throughput those sorts of things that constrain the individual linkages between nodes and then finally protocols so protocols as

I alluded to briefly earlier Protocols are the the rules governing the way that those transitions happen so this might be a little bit confusing and Abstract you but think about handoffs within your company so if you have a product or a project that is that is getting fired up you have certain protocols about like okay who's responsible for what at what time so a racy Matrix if you're familiar with that that is an example of protocols about interfaces who is responsible for what when and at what time Gantt charts and burn down charts are also

examples of protocols where you say hey there is a series of operations or a series of events that must happen as this work product is developed such as a new software application or service that your company is offering boundaries so a boundary in this case is the edge of a network system so basically that is the sphere of influence that that network has domain over uh so uh there's a few other things to think about is the permeability or plasticity of those boundaries so is it easy to add and remove nodes from a network if

if yes then it is high permeability which means that it can accept stuff into the network so social media networks are high permeability where pretty much anyone can come and go as they please whereas a factory floor is low permeability because you actually put systems in place physical systems to preserve the structure and integrity of that Network likewise Logistics Supply chains they are permeable but you don't want them to be permeable and so for instance maintaining supply chain Integrity is actually an entire discipline unto itself influence adaptability and constraints we've already kind of talked about

those so I don't think that I need to go back over them but the point here is that when you're thinking about networks you need to think about the bound injuries of those networks and most importantly the permeability or or vulnerability of those Network systems okay so that was a really fast crash course in networks for a quick recap there are networks which is just an interconnected uh set of components with linkages and relationships and you take a holistic view looking at it as a whole uh then there's nodes within the network which is the

individual elements or components with their own identities functions roles rules and States within the network there's interfaces which is the the actual connections between nodes in the network and so taking a step back a network is just the nodes and interfaces so that's what I mean by interface and interfaces are governed by the type of of information material or energy being transmitted the method by which it's being transmitted and the characteristics of the links so say for instance a chain on a motor is very different from you know an API call is very different from

word of mouth and so on and so forth but these are all different ways of looking at the interfaces within within a network and then finally boundaries which is the edge of the network which is is defined by the sphere of influence if a network does not have the ability to influence a node then that node is outside of the network so for instance a a closed social network like a private group on Facebook can only affect the people that are on that Facebook group but people that are not on that that are not in

that domain that are not in that Network they might not see that information or they might be peripheral nodes because you know some of the information on that Network might spread by word of mouth but then you can have entirely adjacent networks where you've got Network a and network B and there's very little communication between those networks so that's what I mean by boundaries all right now moving forward to network effects now that you have a mental model of what a network is let's talk about the aggregate behaviors of networks and in terms of behaviors

and failure modes so one failure mode that everyone is familiar with is Network congestion so Network congestion is the most familiar example is going to be traffic jams so traffic jams are networks of roads and the the nodes in the network can be seen as the intersections and destinations of the highway system or Road system and then of course the the linkages or connections are the actual roads and interchanges that allow cars to to be transmitted across that Network Now cars you might say those those are nodes but I would say that actually in from

a network systems thinking cars are actually the material being transmitted across the network and so there's resource limitations such as bandwidth you can think of the throughput of a highway system as a function of its width right so an eight Lane highway is going to have a higher throughput than a two-lane highway or a one-way dirt road uh so when you have Network congestion uh basically you have decreased throughput because the load is greater than the capacity this also increases latency meaning transmit time or Transit time is longer than it otherwise should be you another

effect that you get is queuing so uh queuing is basically when you're waiting at the stop light or you're waiting to get on the highway because the network is currently overloaded and then finally one of the most important things to think about here in terms of network congestion is feedback loops so there there are some feedback loops that are self-correcting meaning that uh you know a sufficiently sophisticated highway system will have alternative routes which means that you know hey here's a here's a sign that says you know this this road is congested use a detour

uh so being able to reroute things is a way to alleviate Network congestion but in many cases that's not available and so what you have is knock-on or Ripple effects that actually make the make the situation get worse over time so an example obviously everyone's familiar with uh with traffic jams but from Storage networking there's this this phenomenon called slow drain which basically means that uh when when your storage network is trying to transmit too much data if you have a node that is causing a backup you end up with this ripple effect that is

felt everywhere else in the network because everyone is having to wait their turn and you end up with these really weird exponential decays of transmit time which again this is why sometimes you're on the highway and even though there's there's a lot of cars on the highway everyone is still going 60 or 70 miles an hour but with just five percent more cars everyone slows down to 30 miles an hour so that's what I mean by an exponential decay and performance and that is a knock-on or ripple effect and that act that honestly is is

very very similar to a slow drain phenomenon and you can see this this uh these sets of network effects in all kinds of networks Supply chains this is why we had a really big supply chain meltdowns during the pandemic as an example viral effects so viral effects are particular generally particular to High Velocity networks where basically you have an exponential spread of information uh behaviors or other contagions within a network so information can spread very quickly and then of course we all just survived a pandemic and so you can see exponential growth in in these

but how does that happen you have a high transmission rate so that's the number one thing is whatever the method is you have a high transmission rate but the way that this happens is what's called Network amplification where each node can create more new nodes and so say for instance the rate of infection is each node can infect five new nodes per day but then you know so you start with one node and then the following day you have five new nodes and then the following day you have 25 new nodes so on and so

forth so that's what I mean by viral effects and of course this was uh noted at the onset of YouTube and Twitter and other social media things where there's this new information viral phenomenon another thing is that it happens in a short period of time but also what you can have is you can have what's called a saturation point where basically the information has spread as far as it's going to or the disease or whatever and then you start to get diminishing returns or this tapering effect and this is why you see kind of a

bell curve uh of viral effects on you know YouTube and Twitter and other places even the spread of viruses within uh within computer networks so in my past life as an I.T guy we would see if you had an infestation it would start very small but then once it got into the you know right or the wrong places rather it would spread very quickly but then it would it would fill up that container whatever those Network boundaries were those security boundaries and then it would stop spreading so that's what I mean by a saturation point

and then finally there are also feedback loops that can that can play into this such as if you have knock-on effects that cause systemic failures and actually magnify vulnerabilities or if you have the opposite thing where you can reduce vulnerabilities over time to insulate against viral effects induce demand so this is one of the less intuitive Network effects which is uh this is really common in cities actually where what happens is uh there's a there's a city street that's congested and so what they do is they say aha we obviously need to make this one

city street wider to accommodate more people well what happens is more people are trying to get onto that City street but the rest of the system is not actually accommodating the network traffic and so what happens is you end up with these weird bottleneck effects that is called induced demand where rather than looking at the the grid as a whole whether it's a city street or a power grid or a computer network or whatever if you look at just one choke point and don't look at the whole system you might actually create induced demand where

everyone's now trying to go through the one gate that you keep making wider but really what you need to do is have alternative paths or alternative routes and so again this this some of it comes down to perceived efficiency because it's like oh hey I'm just going to take this road because it's convenient to get to but everyone thinks the same way and so it's like well if everyone's making the decision to take one road because there's no good Alternatives that road is always going to be the most congested so that's an example of induced

demand Network equal equilibrium so equilibrium is basically a network either will reach stability or instability and it will have some intrinsic structures and rules that either lead it towards stability or instability so self-regulation uh these are ways that networks can be self-stabilizing so this uh so self-regulation might be cut offs or breaks or other kinds of friction that prevent things from spiraling out of control so in the past in the 70s and 80s and even into the 90s there have been cases of massive massive power grid failures because they were they were intrinsically unstable meaning

that if uh you know one major station failed all the rest tried to take the loads and there was no self-correction mechanisms that made it uh made it self-stabilizing so some of these instability triggers can be things like node failures or linkage failures so for instance if if any network doesn't have the ability to reroute then you're going to try and fit everything through the only remaining routes which means that you're going to have really horrible congestion there's other external interventions that can be done such as you know shutting off the power you know rather

than have more power stations melt down maybe you just turn off the power until you fix the problem you can also add in fail safes and other things that react to those failure conditions in order to make your network more prone towards stability rather than instability and then another one is forcing functions so forcing function is usually uh some kind of constraint such as the limited availability of resources or other internal behaviors for individual nodes and so for instance hunger is a really good forcing function if you look at yourself as either a node in

a network or as a network system yourself you're a system of organs and you know life and all that sort of stuff hunger will force you to go solve that problem so that is that is a forcing function that will compel you to either go hunting or go to the grocery store or go to a restaurant or ask someone for food so that's an example of a forcing function likewise if you go to the grocery store and the grocery store is out of food that is a forcing function of the grocery store to buy go

buy more produce or go buy more Goods so that they can sell it and so these aggregate behaviors of forcing functions that are um that are going to be in in individual nodes but also in the aggregate behavior of the network uh are going to create systemic pressures feedback mechanisms and other uh things that basically Force the network to behave in certain ways and so forcing functions could also be a prohibitive so what I mean by prohibitive is that if say for instance the network gets too hot you turn it off that forces it to

calm down um so an example of this is going to be in the stock market where you have uh where you have fail-safes that basically if if the the stock market detects a major sell-off it actually stops trading and can actually reverse those trades so that's an example of a prohibitive force of forcing function where it says hey we know that you're trying to do these things but we're not going to let you do it so you have inhibitory or prohibitory forcing functions as well as excitatory forcing functions which basically says at a certain point

you must do this thing an example of of a of a positive forcing function is your kitchen when your kitchen gets too dirty you're forced to clean it up because the kitchen is no longer useful so that basically forces a good behavior because your kitchen is dirty it's too small you can't use it and so now you have to engage in a positive behavior such as cleaning it up Cascade failures the Cascade failures are similar are alluded to Cascade failures with the example of the power grid so in in other examples of cascade failures you

have economic collapses like the Great Depression so the Great Depression was a Cascade failure that uh was a combination of all kinds of network effects such as destroyed trust so the nodes in the networks such as individuals Banks and corporations they no longer trusted each other they didn't trust the government so there's a lot of mutual suspicion and people withdrew from participating in the system as a whole which caused the system to fully collapse for more than 10 years and so this uh this kind of systemic collapse or Cascade failure is kind of what people

are afraid of um in terms of either economic failures such as great depressions or hyperinflation you're also going to be afraid of cascade failures in the case of like zombie apocalypse so zombie apocalypse is is a thought experiment of a Cascade failure of um you know Society comes unraveled disease becomes prevalent and so on and so forth and but all of this illustrates the interconnectedness of these massive systems with uh with the potential for breakdowns or disequilibrium or disharmony based on the behaviors of those individual nodes okay so I just threw a heck of a

lot at you and I'm going really fast so you might need to watch this video a couple times and look up some of these terms but now you are familiar with these terms so let's get into Network examples so that you're a little bit more familiar with what we mean by all these things so first natural ecosystems so natural ecosystems are composed of many nodes within an environment and a lot of connections between those nodes so for instance in this arboreal this this Forest environment you have trophic levels such as predators versus autotrophs versus heterotrophs

so you've got carnivores at the top then you've got herbivores in the middle and then you've got the the decomposers and and autotrophs or plants and fungus at the bottom and so these trophic levels uh describe and characterize the connections between the Flora and Fauna of the environment but then another set of nodes and interconnections are the hydrological cycles so the rain the rain rivers and runoff that are the basically underpin all of life because all life depends on water there's other environmental factors that have you know certain forcing functions so in this case a

forcing function in the ecological landscape might be rain or lack thereof so if you don't have rain in an environment that forces certain behaviors or characteristics to arise such as you know trees will start conserving water they'll breathe less animals might die off conversely if you have a lot of rain the environment will respond to that by becoming more Lush more green there will be more food available and so then the animals will start to reproduce more and that sort of thing and so then we talk about equilibrium versus this equilibrium the most healthy vibrant

ecosystems are going to be very resilient meaning they can tolerate the loss or damage to any particular node or system but in some cases ecosystems are actually very fragile meaning that so some famous examples are if you lose apex predators then you will often have cases where the herbivores take over and you can actually destroy entire forests by losing apex predators and this is a non-intuitive effect because what the apex predators did was they kept the population of the herbivores in check um and then the herbivores if they're no longer kept in check they're going

to eat all of the plants all of the saplings and eventually all the trees will die and the entire ecosystem will fall apart we also see this if you lose Birds uh if you lose trees you lose birds and if you lose Birds then you particularly in Island ecosystems the loss of birds can actually cause a complete breakdown of the entire ecosystem we saw this on Easter Island actually Easter Island used to be heavily forested but the locals harvested all the wood for various purposes and then the seabirds stopped coming and the seabirds stopped fertilizing

the ground the the land became infertile and then the population of the entire Island collapsed so that's what I mean by balance and equilibrium within ecosystems another example is capital economies so Capital economies you know the economic system is something that a lot of people are familiar with but it is also incredibly large so there's various nodes within the capital economic ecosystem such as governments consumers voters politicians businesses corporations other corporations other nations trading partners and those sorts of things and so this system is mediated excuse me this system is mediated by things like monetary

policy fiscal policy trade policy and other Regulatory Agencies and bodies that require certain behaviors or protocols around transactions who can sell what to whom at what rate and how much those things cost the medium of exchange whether it's US dollars or Euros or you know Yuan or whatever there's all kinds of mechanisms that participate and try and modify the network system of the flow of of goods services and money around the world but that's fundamentally what's being transmitted in the capital economy is good services and money uh human bodies so I alluded earlier in this

video to the human body as a system or a network so you you are composed of a system of organs uh you know your circulatory system your digestive system your skin system your muscular musculoskeletal system and so on and so forth you are your boundaries are very clear your skin is your primary boundary and that boundary interacts with other systems including people the environment the physical environment as well as the microscopic bacterial environment that you're in there's inputs and outputs to your body so you take in nutrients and water and food and you you have

output such as the expulsion of waste carbon dioxide and excess heat you also output sound and other things so you are hearing me speak right now because my my body system is able to produce meaningful acoustic vibrations that are then picked up by my microphone here um and then is recorded and transmitted to you via other systems so I am interacting with several systems right now uh you know such as you watching this video as well as the electronics that are transmitting this information to you homeostasis is an example of a system that has built-in

equilibria so what I mean by this is that your body works to maintain its own internal state so your body maintains its internal temperature its internal consistency chemistry so on and so forth and so this is an example of a very sophisticated system that has invested a lot of energy and uh and structures to maintaining equilibrium rather than disequilibrium and then of course there's models of disease so a disease talks about the uh the the systemic failures that can happen within your system or the vulnerability and permeability of your system so internal disease such as

cancer and metabolic disease these are examples of network failures where certain nodes in your body like maybe your heart isn't working how it should or your liver isn't working how it should so that's an endogenous disease versus an exogenous disease or externally mediated disease which means that something has come into your body that shouldn't be there like it might be toxins or bacteria or viruses and so those are exogenous models of disease but if you look at it as as a system your system is being infiltrated by something external or it is being mated mediated

by an internal failure that is a way to look at the human body as a system or a network corporations so more specifically corporations can be viewed as systems or network systems so the inputs to a corporation are going to be things like labor Capital material energy and data and then the outputs of the uh of any Corporation there's three primary outputs which is goods or services and or the payment of taxes and dividends and so the input is uh is you know labor Capital material the output is going to be finished goods and services

and then the output is also usually going to be money of some sort to someone some kind of stakeholder there's there's a lot of internal components or organs within a com company so you have the HR department you have the you know the c-suite you have the IT department so these are all systems within systems so these are embedded systems or nested systems and then these uh they're all interconnected by you know circulatory systems or nervous systems within the organization and you can look at these either like the systems of a body or you can

look at them like assembly lines and we'll talk about assembly lines in just a moment and then finally like uh like anybody or any other system there is waste versus efficiency which basically businesses are always trying to maximize the output of goods and services uh by minimizing waste and finding new efficiencies assembly lines so assembly lines are an even more zoomed in view of a network system where basically you have inputs such as material and energy in order to produce a particular good you have stations within the assembly line so these are a kind of

node so a station on an assembly line could be a person that is assembling things or a person who's painting things or a robot that's automatically welding so stations are nodes that use tools labor and energy in order to modify a work product and then there's conveyances so conveyances are what are the the tools and equipment or processes that move work product across an assembly line or move it through the system transmit it as in as information energy or material through that and so the reason that I'm that I'm saying information or energy is because

you can also look at any intellectual behavior in your company as a kind of assembly line so for instance you have a software product that you're trying to build uh it's still an assembly line because you have you have uh you know product folks you have developers you have QA you have all kinds of people working on it at various parts along the life cycle of your software or anything else in the business you also have stockpiles so stockpiles are storage of either raw materials or finished goods that are waiting for conveyance or Transit and

then you have backlogs so backlogs are you know cues or schedules of orders of work that is waiting to be done so this is a way of looking at assembly lines as a network system as well so thanks for watching I know I threw a lot at you all at once but I also have a link in the description to my medium article on the same thing which will give you the same information but you can take a little bit more time to digest it I hope you got a lot out of this episode of

systems thinking specifically about networks like subscribe and consider supporting me on patreon thanks for watching to the very end have a good one