An invisible technological world
So many things our lifestyles depend on that many of us overlook and take for granted when we really shouldn't
Some time ago, walking to the Bay Bridge, I took a photo of work being done on the equipment at the base of a cellular antenna tower; here it is with some annotations:
Many people take for granted that when they reach for their cell phones they'll be connected to a network, but few people grasp the systems that support it. Despite how physically big some of them are, these systems are invisible.
Those invisible systems are everywhere, they are important, and some recent disturbances (pandemic, war in the Ukraine, Suez Canal closure, and the responses to them) have shown that our lifestyles are not as secure as many of us think they are, particularly because these invisible but essential lifestyle support systems, like Rodney Dangerfield, don’t get no respect.
From the photo we get an idea of some of these invisible systems: a cell phone connection to the network requires communications ground equipment, electrical infrastructure to power it, skilled labor (the workers doing maintenance), support services, like transportation, and infrastructure for those, like roadways.
Let's take a look at those invisible systems:
Expectations and use
How can big and important things become invisible?!
Let us begin with something that was the topic of a previous post, the evolution of preferences: how changing expectations and the difference between expected and delivered value create the miracle to product to commodity path:
Note how the quality of the delivered service keeps improving, yet our perceptions of it go from miracle to commodity (taken for granted). Some authors hypothesize that the increasing speed of technological advance and the shorter intervals between major innovations (relative to the past) have made people ever more oblivious to the increasing complexity of the underlying systems.
Why bother with this basic marketing factoid? Precisely because it's this process of something that appeared miraculous becoming a commodity and getting taken for granted that has led to a number of bad choices at individual levels and bad policies at institutional and sometimes government levels.1
Commerce
Most people only see the visible side of commerce (retail) and have, at best, a vague notion of the functions of a distribution channel. Few people have any idea how much is involved into getting something out of a factory and making it available at convenient locations in convenient quantities for a reasonable price.2
Let’s just consider two of the many functions of a distribution channel: matching assortment and the logistics of getting a product to a store.
Matching assortment is so prevalent in our commerce that most people don’t realize it exists. (It has been commoditized into invisibility, per our illustration above.) To illustrate, let’s do a simple exercise: take our last supermarket visit and check how many different manufacturers made the products in our cart.
Mine had 14 manufacturers. Some of these might be part of the same large conglomerate, but they operate as independent companies, so we’ll use 14.
Let’s imagine there are 1000 customers making the same exact purchases. (Warning: overused MBA Marketing Channels visualization coming up — because it makes the point well and fast.) So, how many transactions happen with and without a distribution channel acting as an intermediary?
Absent a distribution channel, each manufacturer has to make 1000 separate transactions and each customer has to make 14 separate transactions; with a channel, each manufacturer and each customer only makes one transaction: the other party in all of these is the channel.
More important than the total number of transactions, though, is the size of those transactions: for many of the manufacturers, directly serving a customer buying, say, one $1.99 bag of walnuts makes no economic sense; using a distribution channel as intermediary to aggregate all the individual costs of serving a product to a customer lowers the total cost in the system.3
(In fancy words, the channel creates economies of scope and uses them to create economies of scale for the manufacturers.)
Logistics (moving and storing stuff) are another important function of channels. Many people believe that courier companies “solved” that problem — and they have partially done so, for the last leg of online retail — but there’s a lot more involved in getting, say, an iPad from Foxconn in Taiwan to the San Francisco Apple Store.
Foxconn doesn’t ship individual iPad boxes to the San Francisco Apple Store. Those boxes are palletized (put on pallets); the pallets are containerized (put in shipping containers); the containers are placed on rail cars, then on ships to the US, then rail cars and/or container-frame trucks to a breakbulk facility; there, the pallets of iPad boxes are removed from the container and matched with pallets of other Apple products going to a regional distribution center, either in large tractor-trailer trucks or in containers; at the regional distribution center, the pallets are broken up to make the required assortment for the local stores, and one of those assortments is sent to the San Francisco Apple Store in a new pallet or in big shipping boxes, using a small truck or van; the store then organizes the products for retail.4
Now do this process anew for each of the 75 million SKUs (stock-keeping units, i.e. different items) that WalMart added in 2022 (online). That’s logistics.
Logistics is such an important part of value creation that many large corporations bring it in-house (if we want to impress our friends we can say these corporations vertically integrate their logistics).
And some people think it’s just UPS and FedEx, and that’s the ones who even notice that products don’t just materialize in stores.
Industry
Sure, most people understand industry exists; some even have a basic, toy understanding of how some things are manufactured.
The difference between toy understanding of industry and real understanding of industry can be observed when we ask scientists about production engineering: someone who understands, for example, the chemistry of making poly-[ethylene terephthalate], i.e. “react ethylene glycol with terephthalic acid, then polymerize,” will be of no use to anyone who wants to make PET plastic at scale.
Production engineering takes the science as a given and focuses on making things at scale, efficiently, and on a schedule. All the elements that matter in production engineering appear as unknown unknowns to scientists who’ve never left a lab: things that are of no importance in the lab are key success factors in production.
A factory is a far cry from a lab: labs order reagents in grams, with impurities measured in parts-per-million or -billion, and trust the label, factories order inputs in tonnes, with impurities measured in parts-per-thousand, and must test batches to be sure; labs typically don’t care for efficiency because most of what they make is for disposal, factories live or die with minor changes in efficiency and costs; and labs tend to be staffed by well-trained, motivated, conscientious researchers, while some factory workers aren’t unknown to be none of those things.5
And, at best, most people have a scientist-like understanding of how things are made. That in itself isn’t bad — and it places these people in the top stratum of knowledge about how the world works — but it’s important that we understand the difference between the core idea of how something is made and the practical execution of that core idea when it becomes an industry: many, many layers are added to that core to make it practical.
That goes double for understanding how an idea eventually becomes an industry over time and the number of failures involved in getting one success. (20,000 words excoriating media reports of the newest revolutionary new new thing removed.)
As an example of general misunderstanding of how industries evolve and the process of going from an idea (or a scientific discovery) to a product deployed at scale (illustrated in a previous post), let us consider Tesla’s recent presentation of its robot:
When Elon Musk presented the TeslaBot a year before, he used two presentation props: one was a mockup of the robot, which many people dismissed as a mannequin, unaware of the widespread use of physical mockups in the early phases of product development; the other was a dancer in TeslaBot livery, clearly intended as a joke.
That’s a presentation of the idea, and some entertainment for the audience.
One year later, Musk presented a basic technology demonstration, and it’s interesting to see the contrast between the reactions of experts in robotics, most of whom congratulated Tesla on the quick evolution from idea to tech demo, and the reactions of the engagement-driven social media influencers (pro and con), mostly showing they don’t understand this is a crucial step and Tesla took it very fast (con side) and there are a lot of steps left before having a product deployable at scale (pro side).
In another example, entrepreneurial MIT professor Donald Sadoway describes the development of his process of decarbonizing the production of steel, at the time of recording the video still at the tech demo phase:
Supply chain and logistics
Now, let’s take the complexity of one industry and its distribution channels and multiply it across a chain of industries (or more precisely a network) that use one’s output as the others’ input: that’s the supply chain. And it’s even more invisible to the average person, until it fails.
The main difference between a supply chain and a distribution channel is that in a channel the product itself doesn’t change,6 but each industry in a typical supply chain includes a conversion element (turning inputs into outputs: for example, taking electronics components and making controller boards that are sold to other companies downstream in the chain, for example to be put in home appliances) and possibly a distribution channel of its own. Each of these conversion systems links to other supply chains, those of its inputs, and to supply chains that use its outputs.7
(Here’s an illustration of possible unforeseen effects of a supply chain problem, in this case a potential shortage of Diesel exhaust fluid threatening to shut down trucking, and the logistics sector, from Doomberg; it’s from December 2021 and the looming crisis was averted, mostly due to hard work of people no one will ever hear about.)
A complex system made of many interlinked parts is likely to display hard-to-predict responses to perturbations. (What mathematics calls chaotic behavior.) This is a result from systems theory and would suggest two general rules here: that each element in the chain should be as resilient as possible to perturbation, so that incoming perturbations don’t propagate downstream in the chain; and that external perturbations should be minimized whenever possible.
So, basically the opposite of what we have built in the past four decades or so.
Resilience to perturbations is usually accomplished by having reserve capacity that can be made available on demand (“dispatched”) and input and product reserve stocks to deal with input or demand mismatch to production. So, have we done this? (Hint: no!)
Since the 1980s, reserve capacity has been treated as a cost that doesn’t produce value, and therefore something to minimize (as opposed to an option to be exercised in case of emergency, which has a value of its own just by existing); and, with the rise of just-in-time production, advocating for safety reserve stocks in production systems gets treated as outdated thinking and lack of faith in supply chains.
As for keeping avoidable perturbations to a minimum, while occasional accidents like having a box carrier stuck in the middle of a main commercial sea lane like the Suez Canal are unavoidable, there are many self-inflicted perturbations, such as mandating or outlawing technical matters for political reasons, sometimes with barely any lead time for the system to prepare (ex: actions in the pandemic).
An example of the effects of perturbations in supply chains is the Bullwhip Effect, illustrated here with retail sales, by the WSJ:
Basic infrastructure
There’s a number of what we could call foundational supply chains, but more commonly called infrastructure: these are essential services that all supply chains (and consumers) use regularly.
As per the miracle-to-commodity diagram, these are so regularly used that we take them for granted. And sadly those whose responsibility is to make sure they are kept operational sometimes take them for granted as well.
The photo at the top shows some visible infrastructure: the telecommunications equipment, the electrical grid, and the road network used by the truck. These are, today and in the US, essential infrastructure. So, how well are they understood and how carefully are they treated?
Telcos are highly-regulated private companies and are in the growth phase of their category life cycle, so despite providing what are now essential services to many different supply chains and consumers, they are still in the “product” level of visibility. And, while most people have no idea how telecommunications infrastructure works, they understand that their personal iDevices rely on a network that needs careful handling by people who know what they’re doing.
No one would listen to a teenage high-school dropout telling them how to manage the telecommunications network providing iPhones with TikTok and Instagram access.
The electric grid is only the most important system in existence, in the following sense: if the US grid stopped working for one year, between 80% and 95% of the US population would die.8 Because energy has become a politicized matter, we’ll leave the discussion of the topic to other blogs, but note that a paragraph like the preceding one cannot be written here.
(Apropos of energy infrastructure and its management, The Polymerist has an interesting analysis of the EU energy crisis here, from early September.)
As for the road network, one of the oldest types of infrastructure in existence, here’s a playlist by civil engineer Grady Hillhouse that shows there’s a lot more in the rock-and-gravel-based infrastructure of CivEng than even other engineers are willing to give them credit for:
FYI, Grady Hillhouse has a new book on infrastructure, Engineering in Plain Sight.
Skilled labor
Underneath all of this is a skilled workforce, and so few people realize that. Really.
For a technology-based industrialized society, we need skilled labor. Duh! Almost everyone gets that. What most people don’t get is that the distribution of skills in the population determines how fast its society can react to technological change (that is, how competitive it can be in an evolving global market).
Before we go on, it’s important to understand a distinction: skills versus credentials. A skill is the ability to do something; a credential is a document that says someone can do something. The overlap between them depends on the credentialing mechanism. We’re talking skills in this post, not credentials.
Typically the distribution of skills for a certain industry or product category (say telecommunications) in a population follows an inverted pyramid:
(Each level of the white rectangles requires the next one down to be operational, at least in the long term; that’s the meaning of their order.)
Typically there's a larger number of people who can use a product than maintain or repair it; more people who have the skills to maintain or repair a product than to manage/run the production system for that product; those who turn technologies into products (do product engineering) and design the production systems and processes (do production engineering) are fewer still; and usually there are very few people who have the skills needed for creation, research, and development of technologies.9
If the number of people with the level of skills needed for, say, maintenance become too low, people running production systems (for example, line engineers) start to move into maintenance jobs, and this effect can propagate towards the bottom of that diagram. In essence, if a society doesn't cultivate a technical-skilled class, it starts to starve the base of that inverted pyramid and to lose competitiveness.
This isn't just a temporary reallocation of talent; over medium to long terms, the incentives to learn complicated technical skills disappear if the jobs that require those skills are few and far between. More dangerous still is the widespread notion that those jobs can be left to “someone else” and are less valued by society than non-technical jobs.
That last part may appear strange to someone reading a Substack called Technology, Business, and Numbers, so here’s a clarifying example: how many of our acquaintances would prefer that their child get a college degree in the Humanities and work as a tutor for rich kids than become an industrial electrician and make five to ten times as much in income but without a post-secondary degree?
Exactly.
In conclusion
Almost everything in our lifestyles depends on the many invisible systems mentioned above; it’s unrealistic for any of us to know all of them in detail, but it’s important that we don’t discount their importance (and don’t accept others discounting their importance) simply because we don’t know much about them.
The pyramid of invisible systems doesn’t capture this one phenomenon: that most of the use itself becomes invisible, a combination of (a) users following the miracle-to-commodity path for things they personally use regularly and (b) of product engineers and interface designers ingenuity at hiding the complexity of how product functions are executed.
This was most apparent during the heyday of the first “dot-coms” in which many people predicted the end of all distribution intermediaries other than those providing physical transportation of goods. Jeff Bezos, a contrarian thinker at the time, gambled that the functions of the channel would still be needed. We know who was right now, and he's making space rockets as a retirement hobby…
This logic applies within the channel itself, so retailers don’t buy directly from the manufacturers, but go through wholesalers, possibly wholesalers specialized by category, etc.
This is a stylized and simplified version of a logistics process, for example missing all warehousing, for illustration purposes only, not a description of Apple’s actual logistics.
It’s hard to overestimate the importance of that last difference. It’s similar to, but orders of magnitude more important than, the difference between the selected operators who use a prototype in a tech demo and the final consumers who will use the product. But as many post-mortem reports of industrial accidents reveal, “human error” in industry has the multiplier effect of the industrial installation.
Significantly, that is. Some distribution channels offer value-added services at various stages of the process (beyond the usual functions of a channel, which as we saw above are also value-adding); these can be considered hybrid elements of a channel or a supply chain.
The use of “chain” instead of network is somewhat wrong, in that these industries have many linkages among themselves. Even at its more one-product-centric (the idea behind the original “chain” nomenclature, focused on one product line), it would be a supply tree rather than a chain. Some people who analyze these relationships skip all visual metaphors and just represent entire sectors as input-output matrices of transactions between companies, possible divided by product line within each company.
There are different estimates in technical reports depending on the preparedness of the population, whether the start of the blackout is in the Winter or Summer, and on different assumptions about the social cohesion of the population facing the challenges. In many of these reports, most deaths would occur between two and five months into the blackout.
The relative numbers of people aren't fixed, as some industries like software bypass many of the elements of traditional manufacturing and others have productized most of the advanced skills into software or actual machinery, again bypassing some elements of the more traditional divisions of labor in those industries. But it's a reasonable approximation.