Tag: Enterprise Architecture

Canto X of Dante’s Inferno maps surprisingly well onto some of the recurring structural failures of Enterprise Architecture.

The central analogy is this: the heretics in Canto X can see the distant future, but they cannot properly perceive the present. That is an uncomfortably accurate description of a weak EA function.

Enterprise Architecture often becomes excellent at describing a future state—target architectures, capability maps, principles, strategic roadmaps, reference models—while having only an approximate understanding of how the organisation actually works today.

Farinata is therefore a useful patron saint of dysfunctional architecture.

He can see far away. He cannot see what is directly in front of him.

In EA terms, the organisation has a beautifully articulated Target Architecture for 2030 but cannot answer basic questions such as:

• Which applications actually support this business process?
• Which team owns that integration?
• Why does Finance still depend on this spreadsheet?
• Which Oracle database is really authoritative?
• Why are there three identity stores?
• What happens if this server is switched off?
• Which project introduced that interface?
• Who is actually paying for it?

This is the first structural problem: strategic farsightedness combined with operational blindness.

Canto X goes further. The inhabitants of the sixth circle are enclosed in burning tombs. Each tomb contains people whose intellectual position has become, quite literally, their permanent enclosure.

That is a good metaphor for architecture domains.

Business Architecture sits in one tomb.

Solution Architecture sits in another.

Security Architecture has a particularly well-governed tomb with an approval workflow.

Infrastructure Architecture has several tombs because nobody completed the consolidation programme.

Data Architecture has drawn a conceptual model of the cemetery.

Each discipline possesses a coherent internal worldview. The difficulty is that the enterprise exists in the spaces between them.

This produces what might be called the architecture-of-tombs problem: organisational structures encourage architects to optimise their own representational domain rather than understand the operating system of the enterprise as a whole.

Business architects produce capabilities.

Application architects produce portfolios.

Data architects produce information models.

Infrastructure architects produce platforms.

Security architects produce controls.

None of these things is necessarily wrong. But the organisation does not actually operate as a capability model, application portfolio, information model, technology stack or control framework.

It operates through transactions, decisions, dependencies, people, contracts, workarounds, incentives and exceptions.

That distinction is fundamental.

Cavalcante introduces another EA pathology.

He misinterprets Dante because his information is incomplete. Dante uses a tense that Cavalcante interprets as evidence that his son Guido is dead. The inference is rational given the available information, but wrong.

Enterprise Architecture routinely does exactly this.

An architecture repository says:

Application: ACTIVE

The architect concludes:

Application is operationally required.

But perhaps the application is technically active and functionally obsolete.

Or:

Capability: CUSTOMER MANAGEMENT

The architect assumes there is a coherent organisational capability behind the box.

There may instead be six departments, four SaaS products, two outsourced teams, thirteen spreadsheets and a woman called Janet who reconciles the whole thing every Thursday.

The architecture model is not necessarily false. The problem is that semantic compression destroys context.

Architecture repositories tend to replace complicated operational truths with categorical statements:

Strategic

Tolerate

Invest

Retire

Compliant

Non-compliant

Cloud-ready

Legacy

These categories become dangerous when decision-makers forget that they are abstractions.

Cavalcante makes precisely this mistake. He takes an incomplete representation for reality.

There is another, more uncomfortable parallel in Farinata.

Even in Hell, Farinata remains obsessed with Florentine factional politics.

His circumstances have changed absolutely.

His mental model has not.

This happens constantly in enterprise transformation.

The organisation announces:

Cloud first.

But procurement remains designed around buying servers.

It announces:

Product operating model.

But funding remains annual project funding.

It announces:

Agile delivery.

But governance still requires eighteen sequential approvals.

It announces:

Data-driven organisation.

But authority remains hierarchical and political.

It announces:

Zero Trust.

But network access is still implicitly trusted according to location.

This is the Farinata problem: organisations carry obsolete institutional identities into radically changed environments.

The architecture changes.

The institution does not.

A new target operating model can therefore reproduce the old organisation almost perfectly underneath different terminology.

Departments become “value streams.”

Projects become “products.”

Technical design authorities become “architecture enablement forums.”

The old bureaucratic relationships survive.

This connects to another feature of Canto X: the damned remain intensely interested in Florence despite being permanently excluded from it.

Enterprise Architecture can develop the same relationship with delivery.

Architects discuss delivery continuously while being structurally separated from it.

They review designs.

They issue principles.

They maintain standards.

They chair governance boards.

They create roadmaps.

But they may not build, operate, support or retire anything.

EA therefore risks becoming an exiled class describing the city from outside its walls.

That produces predictable consequences.

Delivery teams perceive architecture as theoretical.

Architects perceive delivery teams as undisciplined.

Programmes seek dispensations.

Architecture creates stronger governance.

Programmes create better techniques for avoiding governance.

Eventually the architecture repository records an enterprise that exists primarily inside the architecture repository.

There is also a temporal problem embedded in the canto.

Architecture normally describes three temporal states:

Current → Transition → Target

The model looks rational.

Real enterprises behave more like:

Past + current + abandoned future + emergency workaround + acquisition residue + regulatory exception + someone's strategic pilot

all operating simultaneously.

Enterprise systems are archaeological.

You rarely replace one architectural era completely. You accumulate them.

Mainframe assumptions survive inside APIs.

1990s organisational structures survive inside cloud tenancy models.

Old ERP data structures determine modern business processes.

Temporary integrations become permanent.

Pilot platforms become strategic because somebody built something important on them.

The enterprise therefore resembles Dante’s Hell more than an architecture roadmap: layers of historical decisions continue to exist after the circumstances that produced them have disappeared.

There is a final and particularly useful lesson from Canto X.

The damned know the future only while there is a future to know. Eventually even that form of knowledge disappears.

Architecture faces an analogous problem.

The farther a target state extends into the future, the less it describes an actual destination and the more it becomes an argument about desired direction.

A five-year architecture cannot sensibly specify detailed systems.

Technology will change.

Suppliers will change.

Regulation will change.

Business strategy will change.

Acquisitions will happen.

Projects will fail.

Someone will buy Salesforce.

Consequently, mature EA should not primarily attempt to predict the future.

It should improve the enterprise’s capacity to change.

That means the real architectural objects of interest become things such as modularity, substitutability, interoperability, ownership, dependency transparency, data semantics, technical debt, lifecycle state, operational coupling and decision rights.

Instead of saying:

In 2031 the enterprise shall consist of these twenty-three strategic platforms.

EA should be able to say:

If strategy changes, these components can be replaced independently; these dependencies are understood; these interfaces are governed; these data responsibilities are explicit; and these systems can be retired without discovering that an unknown payroll process depends upon them.

That is a fundamentally different conception of architecture.

And it gives Canto X an unexpectedly good architectural maxim:

Do not become so good at seeing the distant future that you lose the ability to see the enterprise standing immediately in front of you.

The deepest structural failure of EA is therefore not bad modelling.

It is confusing knowledge about the enterprise with the enterprise itself.

Dante’s damned inhabit their intellectual constructions forever.

Enterprise architects should probably take the hint.

---

Canto 10 takes place in the Sixth Circle of Hell, inside the walls of the City of Dis. Here Dante encounters the heretics, specifically those associated with the Epicurean belief that the soul dies with the body. They are imprisoned in burning open tombs. (Digital Dante)

The canto is important because it is simultaneously about religion, Florentine politics, family, prophecy, and the strange nature of knowledge in Hell.

The first major figure is Farinata degli Uberti, a powerful 13th-century Florentine Ghibelline leader. He rises from his tomb almost majestically, standing upright “from the waist upwards,” behaving as though even Hell cannot humiliate him. (dantelab.dartmouth.edu) Dante and Farinata immediately recognise one another as political enemies: Dante’s family were associated with the opposing Guelf faction.

Their conversation is remarkably political. Farinata is dead, burning in Hell, yet remains obsessed with Florence and factional politics. He reminds Dante that Dante’s ancestors were expelled from Florence; Dante replies that they managed to return, unlike Farinata’s family. (Bicerin)

This gives the canto one of its central ironies: death has not freed Farinata from worldly identity. He is still a Florentine aristocrat, still proud, still fighting political battles that no longer matter.

Then Cavalcante de’ Cavalcanti suddenly rises from the same tomb. He is the father of Guido Cavalcanti, Dante’s friend and fellow poet. Cavalcante asks why Guido is not accompanying Dante.

Dante answers in a way that accidentally uses the past tense. Cavalcante interprets this to mean that Guido has died. Horrified, he collapses back into the tomb before Dante can explain that Guido is actually still alive. (La Divina Commedia)

That apparent misunderstanding leads to one of the most interesting ideas in the entire Inferno.

The dead can see distant events in the future, but they cannot properly perceive the present. Farinata explains that their knowledge resembles farsightedness: distant things are visible, but things immediately before them are obscure. (LitCharts)

So the damned possess a grotesque kind of prophetic knowledge.

They can see what is coming.

They cannot see what is happening now.

And after the Last Judgement, when there is no longer any future to see, their knowledge will effectively disappear.

That is part of the contrapasso—the poetic correspondence between sin and punishment. These people denied an eternal spiritual existence and concentrated excessively upon the temporal world. They now exist eternally but experience time imperfectly.

The canto also contains an important prophecy concerning Dante himself. Farinata predicts that Dante will soon understand how difficult exile is. This anticipates Dante’s real political exile from Florence in 1302. The fictional journey of the Inferno is conventionally set in 1300, so Dante is writing retrospectively while allowing characters in Hell to “predict” events that the author already experienced.

That date structure matters:

• 1260: Farinata and the Ghibellines defeat the Florentine Guelfs at the Battle of Montaperti.
• 1264: Farinata dies.
• 1300: fictional date of Dante’s journey through Hell.
• 1302: Dante is exiled from Florence.
• c. 1308–1321: Dante composes much of the Divine Comedy.

The commonly reconstructed chronology places Canto X around Holy Saturday, 9 April 1300, although the precise chronology of Dante’s journey has generated scholarly debate. (Wikipedia)

The deeper structure of Canto 10 is therefore rather elegant. Dante puts people who believed that death ends consciousness into tombs where consciousness never ends. He gives politically obsessed men knowledge of Florence’s future but denies them knowledge of its present. He gives Cavalcante prophetic sight but prevents him from knowing whether his own son is alive.

And Farinata is perhaps the most striking contradiction: physically damned but psychologically undefeated. His body rises from a grave while his personality remains completely intact—proud, aristocratic, partisan and contemptuous.

That is why Canto X is one of the first places in the Inferno where the damned stop feeling merely like examples of sins and become extraordinarily convincing human personalities.