Thoughts on networks, communications, blockchains, and a space-age internet – Info News

Networks are an important way to visualize and quantify objects and their relationships with other objects. Particles exchanging photons with each other constitute a network. Companies with warehouses sending and storing freight carried by trucks, ships, and aircraft also constitute a network. Many interesting examples exist including television broadcasting, the food chain, consumer retail, and the power grid.

Businesses can be understood by considering networks that describe their operation. For example, oil wells are located in certain places that naturally occur, but oil or its derivatives are everywhere (gasoline and plastics, to name the major ones). It would be vastly inefficient to distribute oil from wells directly to people’s cars, for example. It’s easy to see this if you draw a geographical network where nodes are locations on a map and edges are the distances between locations. Then you can draw another graph on top, where new edges indicate oil pipelines and their capacities or distance. By adding up the “weights” of the edges, you can get a good idea of how expensive it would be to build your network of pipelines.

But the one-to-many network structure describes television pretty well. It is a broadcast medium that effectively takes a single source and distributes it directly to its audience at the same moment in time. Same goes for radio broadcasts and any live video on social media today.

Contrasting these two, we see that sometimes a very simple graphical perspective can tell you a lot about the nature of a business. We can tell that oil companies face a large distribution problem due to geological constraints, and why renewable power could be more efficient if it can be done in a larger geographical area on Earth. We can see how traditional TV faces an obvious crisis of personalization, or alternatively we can see how personalization is the driving factor in the explosion of content creation.

What becomes immediately visible from picturing networks are the scaling problems. A simple question to ask is how the number of edges grow in relation to the number of nodes. For example, in a telephone network people need to talk to each other, but as the number of nodes (people) grows, the number of edges increases as the square. For 1 million telephone service customers, that means 1 trillion phone lines. For 30 million customers, it’s 900 trillion! Another way to look at it is to see how well you can organize the graph in two dimensions (or three, if you’d like, the insight is about edge density between nodes). If you can draw the graph in an organized and reasonable way without lots of edges crossing each other, it means you are dealing with a decomposable problem. The lesson in all of this is hierarchy: if you have it, you know you can scale whatever it is you are looking at to a first approximation.

Note in our examples how we have used network graphs in two ways which may have been a little confusing at first. In the case of oil extraction and distribution, we considered how physical goods moved through space. In the TV example, we certainly could have also thought about the physical transmission of television signal and shown how it was similar to oil distribution. We instead focused on insights about the perception of the service by customers, which unveiled its clear limitations. Considering how the graph in mind is constrained by other graphs, such as geographical or infrastructure graphs to name some common ones, can add a new dimension to the picture. Or imagine the more subtle example of how personalization creates a graph with more limited connectivity between social groups. This leads to the emergence of echo chambers and fake news being used for large scale political influence.

So the network graph is a powerful mental tool that helps reveal the structure of systems. For businesses, it helps clarify how the business works, how it would continue to work, and when it might stop working.

The networks of the communications industry

Mail delivery in the United States used to go through one office in New York City and distributed to local post offices (conceptually a basic star graph, or a hub-and-spoke model). But the physical movement of mail is constrained to transportation graphs for roads and rail lines. This is not a particularly surprising fact, it’s just the most fundamental one. People want to communicate where they are. The telegraph was the first device to enable the transmission of electrical pulses to communicate information. They began to appear first in railway stations and quickly moved to post offices.

Telegraphs were primarily used by railway stations, post offices, other important government offices, and large companies. For the average person, it took too much work to maintain the hardware and record messages between themselves. The invention of the telephone changed this entirely since people could hear sounds and transmit their own almost instantaneously.

Consider how the network changes when we go from a few important organizations communicating with each other versus everyone in a middle-class household communicating with any other. The graph becomes far more dense, and the natural way to solve a potential explosion of telephone lines is to set up routing hubs. The switchboard was exactly the invention necessary for building such a network for telephones.

But recall the numbers from our dense graph before for telephone networks. The amount of demanded communication bandwidth hasn’t decreased, we have simply moved that burden to the fewer “fatter” pipes that constitute the backbone of a more sparse and efficient network we have today. Note that fatter pipes means a larger investment in “vertical” scaling, where “horizontal” scaling would be to simply add more lines as we considered before. There was a brief moment in time when telephone networks were built in rural areas by farmers using cheap insulated copper wire on poles, or even co-opting the barbed wire already strung across the fences that contained their livestock! For them, horizontal scaling actually worked, but vertical scaling brings with it the technological advancements only a large, focused company could attain. In the abstract graph view, certain edges might visually appear thicker and be weighted more to depict the capacity and volume of traffic they service.

The transition to optical fibers from electrical telephone lines built heavily upon the prior infrastructure. It too shows further vertical scaling, and it is the mechanism by which telecom companies have maintained their control. They are the benefactors of technology that imposes hard physical and geographical constraints that make it prohibitively expensive to build completely new networks.

It would be unfair not to mention how this power is being challenged today. Telecoms have enjoyed being front-and-center in the lives of their customers. While that is technically still true, communication has evolved from talking on the phone to video chatting, text messaging, blogging, tweeting, liking, and commenting in public. People’s attention as consumers are now distributed across many conceptual mediums. Telecoms are now the annoying utility company that seem to just get in the way. It is the communication facilitators like Facebook and Twitter that people care to give attention to.

For the new technology companies, telecoms present a sizable threat. As long as the internet runs on physical cables, the telecoms will rule. Nothing makes this more evident than the battle for internet neutrality in the past few years.

But there is technology that could change the landscape completely.

Wireless and space

Suddenly the efforts of SpaceX and Facebook to bring about internet infrastructure in radically new ways make sense. The only way to unseat telecoms is to make their current infrastructure irrelevant. Telecoms do own some wireless technology today, but it is not a mature enough technology to replace physical cables in a significant way. Indeed, they are not even mature enough for internet service providers (ISPs) to use between them and their customers. And although mobile service providers are getting pretty good at this, it won’t change the landscape as long as physical cables are still the backbone providers.

For Facebook, investing in moonshot communication projects is critical to their continued prosperity. If one of them succeeded, they would be the most powerful communications company ever to exist (just as the Bell System was before). For SpaceX the rewards are even more staggering: they would own space transportation as well as intra and probably inter-planetary communications. That would make them one of the most powerful companies to ever exist.

Imagine for a moment how the landscape of intraplanetary communications changes through a truly mature wireless technology. If wireless signals can be broadcasted to the entire planet via satellites, the burden of prior infrastructure is gone. Telecom companies, whose ground-level infrastructure is integral to their business, would find their leverage severely hampered. Whether they fall or simply continue in a reduced state is up for debate, but it’s clear there will be significant change in the power landscape.

The very interesting implication of removing the need for heavy ground infrastructure is that it removes a huge capital barrier to entry for new ISP companies. Launching a network of satellites, reorganizing them, or decommissioning them will be far easier and more cost-efficient than laying down millions of miles of new cable in hard-to-reach places (like underneath the ocean). A scrappy upstart with new technology beyond present-day radio or optics could package that into a few satellites and deploy it to the entire planet. The barrier for the consumer to switch is negligible, in theory.

Assuming wireless technology outpaces space transportation technology, it is clear that the power landscape for communications looks much flatter. Eventually we will have colonized enough planets for interplanetary communications to become important. Then the infrastructure for deploying those networks would again become insurmountable for new entrants.

And so the cycle repeats.

The other side of the coin

A discussion about the networks of systems is never complete without talking about centralization versus decentralization. Whether something is centralized or decentralized depends on what view you are taking. But if you pick a view and look at the graph that represents it, centralization is easy to spot: you’ll see few elements (either nodes, edges, or both) that have high value and many that are low value. For instance, from a computer networking hardware perspective the computers that people own (laptops, phones, etc.) are low value. High value computers cost hundreds of thousands of dollars and constitute the backbone of these communications networks.

A decentralized system has a graph where value is much more evenly distributed. For example, the logical graph of the internet as a network that connects simple computers to other computers has nodes that are fairly even in terms of value. In other words, people’s daily-use computers aren’t too different from each other and they mostly do similar things.

A more interesting example is that of western Europe prior to the European Union. They are a collection of countries, some more powerful than others, that interact with each other in a decentralized way. The benefits are clear: full autonomy and freedom from an overseeing organization that may make suboptimal decisions for these countries. The downsides are also fairly clear: there’s less coordination, alignment, and speed along many dimensions. As the formation of the European Union showed, there is strong interest in centralization for many reasons.

Probably the most interesting example of decentralization is Bitcoin, which is built on something called a blockchain. The graph of the Bitcoin network is very evenly distributed too, just like the internet. Decentralized systems pay the cost of higher overhead in return for getting rid of the negatives of a centralized system. For money, decentralization overheads may be an acceptable cost if it takes away power from incompetent or malicious governments.

But the costs in the case of Bitcoin have yet to be fully understood. As of now, blockchains in general face a crisis of scaling. This is not difficult to see from looking at the network: the amount of resources required scales as the number of nodes that require the service. One central limitation of Bitcoin is the number of allowed transactions per second, which is about three or four. There are many knobs here to turn, but the fundamental limit is that a consensus among all or at least most of the nodes is required before adding a transaction to the blockchain.

A useful analogy is democracy versus dictatorship: consensus amongst citizens is required in democracy, which severely hampers speed of action. It is the reason why the Romans used to appoint dictators temporarily in times of crisis. It is also why many democratic governments give their leaders certain powers under certain circumstances to short-circuit democratic rule. This highlights the fundamental trade-off between centralized and decentralized systems and how they must scale.

Constraint violation for disruption

By looking at the history of the telecom industry in the last century in terms of graphs and constraints, we can talk about how new entrants might make an impact. The constraint of their physical infrastructure network by a geographical network reveals a possible area of disruption. Satellite telecommunications is just one technology, but the fundamental principle behind its power to disrupt is to change or violate the constraints on the status quo. It’s fun to think that perhaps a few autonomous aircraft could beam signals using self-tracking lasers across the ocean. In that case, the entire telecommunications backbone that connects the world becomes supplanted by a small aerospace company with a bit of advanced technology.

Of course, it is entirely possible that such technology is not discovered. Then we leave it up to our governments to curtail an unchecked industry behemoth (as they have done in the past). But it is extremely unlikely that technological development stays in the hands of a few, for it is an extremely intensive process that doesn’t always flourish under the strict control of powerful organizations. It is also very unlikely that truly paradigm-shifting technology benefits only the powers that be. Think about how many other possible ways a technology could affect society and it becomes obvious that this is true. Blockchains are a prime example of this, as is the internet. One may argue that the benefactors change over time, but the race to control between the masses and the few is a fundamental fact. Consider how Bitcoin adoption has outpaced third-world governments who don’t have the resources to police it. For the citizens of those countries, they are or soon will find themselves with greater control.

And this too can be interpreted as a network. It is clear that current technology consolidated in the hands of the ruthless can be a poor deal for everyone else. But endless consolidation of power reveals a structure that simply does not scale. It shows that there is still significant force in numbers, even if individuals themselves are ineffective or even disorganized. It is one of the primary reasons for the fall of the Roman Empire (amongst others).

Technology may improve the ability for a small group to control a large one, but that simply cannot scale indefinitely. We know that physical space, time, and derived concepts like speed are limiting and always will be. Technology can make exponential improvements to our circumstances and environments, but each and every time we simply expand our environments exponentially and restore the familiar balance of power.

For such a simple tool, a network perspective provides a surprising amount of insight into our world.