What I learned from telecoms

By Martin Geddes January 1, 2022

Battles over what is rational and righteous are not new to me

How many bottles of wine does it take to get 100 wedding guests moderately drunk? How many pints of milk do you need to run a hospital kitchen for a day? How many gallons of kerosene does it take to fly a 747 for 6,000 miles?

In the “real world” we are accustomed to reasoning about supply and demand, and having a unit of measure to relate the two. Bottles, pints, and gallons are abstract measures with which we can relate the concrete actions, such as wine (supply) disappearing down the gullet of a freshly made father-in-law (demand).

The discipline of engineering extends this matching of supply to demand beyond liquid commodities to complex structures. For instance, a bridge has a static load (the weight of the steel and concrete) and a dynamic load (cars moving over it, wind, earthquakes). Proven trustworthy engineering models give us confidence in the safety margin built into the construction. Novelty is only introduced gradually, since architects and engineers are criminally liable for misrepresenting the safety of what they erect.

From 2001 to 2017 I participated in an industry that has bypassed basic science and engineering, and embraced unbounded novelty: broadband telecoms (via packet networking). How much DSL broadband do you need for three Skype calls (one to Australia) plus some Dropbox and general web surfing? “Lots!” …Will this be enough? “Buy more!” …What’s the safety margin? “Can I interest you in some 5G too?”.

It turns out that this industry lacks a “unit” of supply and demand, and that is because it (embarrassingly) is missing a basic branch of mathematics to express it. If we didn’t have “complex numbers” (with a “real” plus “imaginary” component) we couldn’t model electromagnetism. The analogous statistical equivalent of “complex numbers” for packet networks — which enables a “unit” for broadband — is the ∆Q calculus (with continuous and discrete components).

I helped to document and promote this breakthrough, which is unavoidably needed to have a “probability unit” that “adds up”. That is to say… an algebra that lets you compare supply with demand to see which is “bigger” and by how much. Yes, you have read this right. There is a whole industry that is missing a unit of supply and demand, and doesn’t care about its absence. Indeed, it is commonly seen as a feature, not a bug!

This means there is currently a large-scale industry that is fundamentally insane, both in its engineering and product development methods, as well as how it accounts for resources and prices its services. The answer to “how much?” is always “lots — and preferably lots and lots”. The very act of inquiry why there is no unit of supply and demand, nor accountability for user experience outcomes, is taboo — and anyone who goes there is a troublemaker and an outcast.

This is an industry that is not only mad when it comes to rational engineering, but is also bonkers when it comes to righteous regulation. Industry campaigners (funded by the corrupt usual suspects in Big Tech) came up with the idea of “network neutrality”. This is a theory that the “setting knobs” of a telecoms network (packet scheduling, buffer sizing, fragmentation, control loops, etc.) have a “neutral” position that is “fair” to the different users of the network.

This theory crashes and burns the moment you try to implement it in practise. How do you detect “unfair” traffic management? You cannot, since there is no such thing as a “neutral” network, only a choice of (unintentional) biases. Any attempt to implement a regulatory framework is 100% guaranteed to fail since the phenomenon it purports to regulate is not observable in reality. If you cannot measure it, you cannot manage it: “neutrality” is fantasy invented by lawyers who don’t understand the science!

There is no (general) intentional causal relationship between the “settings dials” and the actual network performance (which is emergent, not engineered). So just because a particular application gets a good or bad (relative) performance outcome cannot (in general) be traced back to specific changes in configuration by the network operator. The “current” packet flow performance is constantly evolving and shifting like the tides on a sandy bay, with occasional storms that reshape the landscape.

The moment any application provider tries to go to court to demand “more neutral” treatment from an ISP they face an insurmountable hurdle: proving that their transient outcome is the intentional result of the ISP’s selection of performance parameters for the network. It cannot be done, ever. Yet there are whole books out there on the topic of “network neutrality” discussing this fantasy world that has no relationship to actual network performance science.

As a “wild child” of telecoms and noteworthy blogger I quickly got invited to speak at conferences and participate in invite-only private retreats in nice places. This lasted for exactly as long as I subscribed to the “fat dumb pipes” orthodoxy. The moment I started to ask awkward questions and challenging the legitimacy of those with prizes for “Internet architecture” I became persona non grata. You didn’t even need to become a conspiracy theorist or Trump supporter to get disinvited from the “polite company” in-crowd.

I was not the only dissident in town, and my key contribution was to link two renegade groups: those working on the ∆Q calculus with those working on the complementary RINA architecture. You could demonstrate beyond any doubt that those pioneering teams were “correct” since they could build technologies (like quality assured broadband, or secure relocatable information “containers”) that proponents of the “fat dumb TCP/IP pipe” could not. No matter how much the orthodoxy denounced and shunned you, they were provably wrong!

The adoption of a false logic and false morality is the basis of authoritarian thought. Telecoms taught me that groupthink could take over an entire industry, and obstruct meaningful progress. There was a hidden shame to having endorsed a pessimal performance architecture, and being unable to answer simple engineering questions. Rival systems of thought had to be denied, excluded, and denounced — lest they expose the incumbent “winners” as fools. My experience was that there were many “scientists” with no interest in science, only in personal status and perceived success.

Telecoms networks are “ludic” systems where we play the “local deity” by setting the “rules of the universe”. It is a bit like how we can design different kinds of card games: poker and bridge take the same input components, but configure them into completely different output “game systems”. My own “hardcore” background in computer science gave me confidence that I wasn’t missing something basic, and I could safely ignore the folly of those who subscribed to irrational and unrighteous schools of thought.

Other kinds of systems are “cosmic” ones — encompassing disciplines like physics and theology, and “ecological” ones — which would include law, sport, and culture. In these cases it is harder to determine when you have properly ascertained the relevant “universe of discourse” and understood the necessary rules of logical inference. If I was asked to comment on matters of metaphysics, for example, I would presume near total ignorance and adopt great humility.

Yet the same essential dynamic plays out each time. Is your “knowledge” socially constructed from intermediate sources of “approved orthodoxy”? Or are you willing and able to go back to the fundamental and foundational elements of the domain, even if that results in disapproved conclusions and social exclusion? For the “feeling of knowing” that comes from a consensus of approving peers is a heady drug, and its addicts make heroin junkies look like health freaks in comparison.

Searching for reason and righteousness can be a lonely path to take, and your travel companions may not have the qualities or endurance to “go all the way”. If you find yourself in complete solitude as you seek truth, then that is not a failure or a problem. The loneliness rarely lasts long. Yet completing the journey takes a determined individualism that few exhibit: a self-compassion that allows for ruthless self-examination without destructively brutal self-criticism.

The discipline needed to maintain intellectual honesty and integrity is akin to taking a monkish vow of “chastity and sobriety” — one that doesn’t sit well with the orgiastic self-congratulatory norms of “clever people” and their egos. Any desire for validation, reward, recognition, approval, or status is an obstacle to the end goal of virtue. A determination to “win” in the passing world of the social is the ultimate barrier to determining what is invariant in the eternal. Telecoms taught me that a brilliant mind or passionate heart are helpful, but ultimately the unflinching pursuit of truth is a spiritual matter.

~~Find the original blogpost here. Reprinted with permission.

By Martin Geddes January 1, 2022 Battles over what is rational and righteous are not new to me How many bottles of wine does it take to get 100 wedding guests moderately drunk? How many pints of milk do you need to run a hospital kitchen for a day? How many gallons of kerosene does…

By Martin Geddes January 1, 2022 Battles over what is rational and righteous are not new to me How many bottles of wine does it take to get 100 wedding guests moderately drunk? How many pints of milk do you need to run a hospital kitchen for a day? How many gallons of kerosene does…