The Digital Aqueduct: What Rome’s Hydraulic Failures Teach Us About Data

As the grandson of a University of Texas at Austin mathematics professor, I grew up constantly being grilled and quizzed with quantitative analogies — the relative size of sunspots to earth, the number of stacked pennies needed to reach the moon, or the classic of lily pads doubling on a pond to illustrate exponential growth.

Pondering how my grandfather, the late Professor James Mann Hurt — who ran a statewide math tournament for students in the 1950s — might illustrate the risks of poor data management, I hit upon the Roman Empire’s flawed understanding of fluid dynamics. His memory inspired me to an apt cautionary tale.

When Engineering Outpaces Understanding

The Romans were famous for (among other things) their vast networks of aqueducts. These were engineering marvels that carried water for miles to sustain cities, baths, and agriculture. But as masterful as their infrastructure was, the Romans lacked a scientific understanding of fluid dynamics. In brief, Romans’ inadequate math assumed a linear relationship between aqueduct size and flow: double the size of the pipe, double the flow of water. Not quite. As we were taught in high school physics with the Hagen-Poiseuille equation, flow increases with the fourth power of the radius — so doubling the radius increases the flow rate by a factor of 16.

Without access to such knowledge when things failed — when water stagnated, overflowed, or never arrived — physical collapse was often wrongly blamed. This was because no Roman understood how the system worked as a whole. They lacked a model, and they paid for it over time.

Today, data flows through the cloud, our digital aqueduct. It flows invisibly through every enterprise, feeding systems, guiding decisions, and sustaining operations. But just like the Romans, too many organizations build vast data architectures without understanding the principles that govern them. Definitions are unclear. Relationships are ambiguous. Governance is an afterthought.

We live in an age when data should be our most powerful asset — fueling innovation, agility, and competitive edge. But when that data is poorly defined, incompatible, or unmanaged, it becomes a silent saboteur.

At data.world, we help clients deal with this in ways small and large every day. But to illustrate, let me confine this analysis to a few high profile, public examples:

The Silent Saboteurs: Modern Data Failures

In 2018, a single data entry error at Samsung Securities triggered the issuance of 2.8 billion shares instead of a modest dividend. The result? A $300 million loss in market cap, regulatory bans, and the CEO’s resignation. Why? No validation logic between dividend disbursement and share issuance systems.

You’ve probably heard the tale of Uber, which over years miscalculated driver commissions by deducting its fee from gross rather than net fares. This cost the company up to $50 million in back payments and corroded trust with its most vital workforce. The root problem? No systematized checks, no clear governance.

These aren’t mere anomalies — they are symptoms of systemic data blindness. When definitions are murky and systems don’t “speak” the same language, organizations make decisions in the dark.

Other examples abound. Equifax ignored a known data vulnerability in 2017, the short-term savings backfired and revealed personal information on almost half the U.S. population. Going back a little further, many will remember Knight Capital’s rogue code that launched a $440 million meltdown in under an hour. Or one that I often highlight when talking to young entrepreneurs: Blockbuster Video’s clinging to siloed systems and bad data habits that opened the door for Netflix to clean their clock with streaming.

Building Better Aqueducts: Catalogs, Knowledge Graphs, and Context Engines

Across industries, a familiar pattern emerges. Failures aren’t caused by isolated bad actors — they’re bred in environments where systems don’t talk, definitions don’t match, and decisions are made without full context. Just like with the Romans scratching their heads at water loss.

At data.world, we’ve built a platform that replaces outdated assumptions with real-time, governed understanding. I’ve written about this often, but it’s worth repeating how we turn data chaos into clarity:

1. The Data Catalog

Your central nervous system. It indexes, classifies, and annotates all data assets — creating one shared source of truth. But unlike a static list, ours is powered by intelligence. If Samsung had been using one in 2018, it would have helped in many ways, but most simply by triggering an alert that share issuances were about to exceed market cap.

2. The Knowledge Graph

Your brain. This is where we lead the industry, with 82 patents and counting. Our knowledge graph semantically maps your data, policies, people, and systems. It understands not just what data exist — but what it means, how it relates to everything else, and where the risks lie. Had Uber been using our suite of tools, the knowledge graph would have enforced the policy that commissions apply only to net fares.

3. The Context Engine

Data without context is dangerous. Our Context Engine reads metadata, relationships, and governance rules to surface insights — like trust scores, lineage paths, and policy violations — in real time. I can think of many ways we might have helped Blockbuster avoid its famous fate of dependence on late fees for 20% of its revenue. One might have been a forecast simulation showing how a digital pivot would affect store performance and staffing needs. Another could have been a mapping of data lineage showing where digital readiness was blocked by outdated infrastructure.

I could go on, but the point is that most data governance is reactive — cleaning up messes after the fact. We flip the script. With data.world, organizations can block incompatible transactions in real time. We help clients run simulations that reveal what will break before a code deployment. Users of our tools automate compliance audits and policy enforcement. Our teams help yours standardize semantics so every team speaks the same data language

This isn’t theory. For instance, NASA’s implementation of its knowledge graph platform led to a 10x increase in engineering productivity. Banks using similar knowledge graph approaches report audit times up to 72% faster. Enterprise agility improves because clarity replaces confusion.

The organizations that win in the next decade will be the ones that treat data not as a passive asset, but as a living system — governed, mapped, understood. At data.world, we’re proud to power this shift. We’re building a world where every decision is informed by trusted data — and where billion-dollar blind spots disappear in the rearview mirror.

The Romans taught us that infrastructure without understanding eventually breaks down. In their final days in the 5th century, the Romans resorted to harsh fines and punishment for water theft that wasn’t. They even contrived ceremonies to appease the “water gods” without apparent result. Let’s not repeat their mistake in the digital age. Let’s make data governance our next engineering marvel — and one that works.

As my grandfather wrote in a lesson plan in 1952: “Quantitative thinking is composed of at least three ingredients: a well-developed sense of numbers, an ability to estimate accurately the size of objects, and a knowledge of the formulas relating the quantities known to the quantities sought.”

It would have been good advice for the Romans. It’s excellent counsel today for enterprises thinking about the future and their optimal use of their precious data assets.

Thanks so much to my grandfather whose inspired me at such a young age to love math and who lives on through me and my love for data (and family). I miss you and your many teachings very much.