-END- (?)

How to viably progress the development of “safety assurance systems” for bureaucracy?

I started writing my Telepocalypse blog in 2003, while still an employee at Sprint in Overland Park, Kansas. I was early enough that there were no corporate policies about social media or blogging. I didn’t mention my employer or discuss my work, so I could opine on the general madness of the industry without being fired.

Now, 23 years later, and through multiple iterations of identity and subject matter, I am still doing free-to-read public writing with voluntary donations. I am considering whether that model is sustainable.

What I have concluded through my work over the last few years is that the world is, indeed, ruled by symbols. It is far easier to manipulate representations of reality for personal or political gain than directly to seize the assets of the world or disclaim its liabilities. Legislatures and courts can be weapons in the hands of predators, not just stabilisers of society.

Some of this is the sin of positive law: invent another debt for exercising an inalienable right, then use the coercive power of the state to enforce it. Some is the personal subversion of law: invert victim and perpetrator, then make the institutional record conform to the inversion. Some is ordinary human failure and variability.

My core observation is that we are missing a crucial civilisational engineering discipline:

the corrigible attachment of representations to reality.

It doesn’t even have an acknowledged name yet.

But this is one of those “once you see it” moments. How did we imagine that bureaucracy was somehow immune to failure modes that are taken for granted in every other safety-critical domain? What is so special about symbolic systems that we cannot model and measure the “representation load” put upon them, and determine whether it remains within their capacity for faithful representation?

We engineer aircraft so that erroneous representations of altitude, attitude, or heading do not silently become reality for the crew. Yet administrative systems can transform allegation into finding, estimate into debt, or classification into entitlement, often without any comparably rigorous discipline for preserving the distinction.

The result can be ruined lives, gross injustice, and cascading harms.

The transformational enabler is AI. It gives us, for the first time, machinery with which ordinary people can express required invariants, interrogate transformations, identify pathologies, and do so across volumes of information that previously made meaningful audit prohibitively expensive.

The challenge I now face is that safety engineering for large-scale symbolic systems is not a mass-participation sport.

I have been developing this work under the umbrella of general-purpose writing on Substack. But it is not something that most of the audience I built between 2017 and 2022 can readily relate to. It is intellectually demanding, often abstract, and difficult to apply immediately to daily life.

The tools themselves are nascent, because the underlying proposition remains unfamiliar: that we should use explicit frameworks to bound officially endorsed unreality, just as engineering disciplines use formal constraints to bound unsafe physical states.

That creates a practical problem. If this is merely commentary, then publishing essays about it is enough. If it is the early form of an engineering discipline, it needs a different development model.

Calling something an emerging engineering discipline creates an obligation that calling it commentary does not. The concepts eventually have to cash out into specifications: explicit invariants, defined failure modes, reproducible audits, measurable tolerances, and tools that work on bounded real-world problems.

I don’t claim to have all of that today. Part of what I am trying to work out is how to get from recognising the common structure to engineering it.

The Substack model is simply no longer a viable living for me, and I have to stop pretending that it is. It’s not just about whether there is a paywall; the issue is deeper. Something has to change: either some massive dislocation in the world redistributes resources, or I find an income that isn’t a kind of alms to the intellectual adventurer.

I don’t belong inside an institution; salaried employment is not an option. I won’t become a “brand ambassador” for somebody else’s agenda. But I do have a mission here, and an unusual combination of skills with which to pursue it.

The question, then, is how to make the work economically viable without making it intellectually captive.

That may mean a paid community of fellow early explorers on a gated platform such as Circle. It may mean developing books, downloads, guides and eventually tools. It may mean applying these methods to specific problems for people and organisations who are willing to pay for the result. One such experiment is already under way.

Probably it means some combination of all three.

So I am at an inflection point. I enjoy writing, and I know that some of you find what I have to say valuable. I don’t intend to stop thinking in public. But I can no longer pretend that the economic structure underneath the work is suitable simply because it has survived this long.

Twenty-three years is a respectable experiment.

Now I need to decide what comes after the question mark.

I am not ending this Substack in its present form. I remain committed to public writing, and right now I am completely dependent upon it for my livelihood.

But I am explicitly exploring what comes next.

Maybe it takes an institutional collapse to trigger overwhelming demand for a rebuild on stronger foundations.

LSWH.

-BEGIN- (!)

—

Comments are open to all.