Mexico's 'CURP for Pets' Debate: A Registry Model, a Mislabeled Record, and a Lesson in Data Integrity

Last week a seemingly ordinary dataset landed on my desk. The top-left cell read: Domain Label: football. Below it sat twenty-two information points, two core viewpoints, and a handful of organisation names. I checked every one of them. Not a single item concerned football. Nowhere was there a club, a player, a coach, a competition, a transfer, a tactical system, or a governance matter. What was there instead was the registration of pet dogs and cats in Mexico — the so-called 'CURP for pets' (CURP para mascotas) — state legislation, and a national registry bill pending in the Senate.

This is not an unfamiliar scene to me. I have spent years building data-classification models — match data, transfer valuation, load models. The biggest lesson from that work is simple: a label is never proof; a label is only a hypothesis that demands verification. Here the hypothesis failed. And that failure is the real subject of this piece.

I could have written a football analysis here. Believing the label, dressing the error in football language, and producing a tidy analysis would not have been difficult. But the cleaner the numbers, the more uncomfortable the story. This story is about registration, and the question is about data.

Context: What Is Actually Happening

The word 'CURP' is familiar in Mexico — Clave Única de Registro de Población, the single citizen-identity registry key. It is a centralised national registry that gives every citizen a unique key. But there is no separate CURP for dogs and cats. Companion-animal registration runs through state-level systems.

Mexico City (CDMX) operates RUAC — Registro Único de Animales de Compañía. The state of Nuevo León has a Ley de Protección y Bienestar Animal, under which registration obligations also exist. At the federal level, a bill is pending in the Senate that is contemplating a national companion-animal registry framework.

A popular rumour has spread that a 'CURP for pets' is now mandatory. The reality is subtler. In CDMX the registration procedure is free of charge, and a national mandate has not yet taken effect. The federal framework is still being defined; where state systems exist, they continue to operate under each entity's own rules.

Here lies the first subtlety: a proposed federal bill and an operating state rule are not the same thing. Confusing the two breeds misunderstanding — much as a single mislabeled domain can send an entire analysis in the wrong direction.

Why is companion-animal registration said to matter? Public health, vaccination, the recovery of lost animals, and accountable ownership are the stated reasons. But these very reasons raise the real question: which reason is primary, because the model's design changes depending on the answer.

This is not a Mexican problem alone. Proposal → rumour → clarification: the cycle appears everywhere, in sport, in health, in taxation. Only the subject differs, not the method.

Registration Is Actually a Model

Registration is a kind of data model. Like any model, it has parts: what gets recorded (variables), who records it (the responsible institution), and who verifies it (governance). The questions in companion-animal registration are complex. What key identifies the animal — the owner's CURP, or a separate ID? Which animals are included, and which excluded? Which state's rules apply where? What happens to the record when ownership changes? What happens when the animal dies?

The spreadsheet is my monastery; the patch notes are scripture. I wrote that line about football models, but the principle is identical. A system's strength lies in how its inputs are defined, where its records end, and how rigorously it verifies.

A larger question hides here — the purpose of the data. Is registration for animal welfare, or for penalties and revenue? If the purpose is to hold owners accountable, the data model differs; if it is public-health statistics, it differs again. A model is never neutral; its purpose fixes its architecture. To me this is the most neglected question — not the technology of registration, but its purpose.

Federal Versus State: The Geography of Governance

Mexico's debate is really about the geography of governance. Between a national bill and local implementation there is always a gap. I stopped asking who won and started asking which state allowed it. The same question applies: which level applies which rule, and who falls outside that rule?

The federal-state split is not merely a boundary on paper; it is a concrete data problem. If each state runs a separate key system, separate rules, and a separate database, building a national picture becomes hard — and errors surface late.

I will not hand down a verdict on whether centralised or state-based is better. The question is procedural: which level of rule, which data boundary, which verification. A model works only when its boundaries are clear.

The Name Is Data: The Trouble With 'CURP'

The name 'CURP for pets' is convenient but misleading. CURP is a specific national citizen-registry key; no such thing exists for pets. Once the name spreads, it creates public expectation — and the gap between expectation and reality is where misinformation is born.

This is a familiar scene. In football a transfer rumour spreads and people treat it as final. Every transfer rumor is a variable waiting for a timestamp. The same holds for registration: a name spreads and people start treating it as law. But a name is never a rule; a name is only a hypothesis that needs a timestamp.

The name matters because the name is the classification. Using 'CURP' fuses companion-animal registration with a national citizen system — factually wrong. Yet this wrong name spreads fastest, because it is short and familiar. In data work that is the real risk: a simple name over a complex reality.

Rumours have an economy. A wrong name travels fast because it sounds like news. Service journalism then fills the gap — clarifying what is still a proposal and what is already law. The role is useful but limited; it does not stop the rumour, it only slows it.

Whose Data, and Where It Goes

There is another layer: who can see the data, and where it travels. In a citizen registry the state looks; in companion-animal registration one can see the owner's address, the animal's health, its vaccination status. Once an integrated national system forms, questions of privacy and surveillance arise.

I have seen many times that once data is collected it never stays only for its original purpose. A registration database can drift toward commercial use, insurance pricing, even flows to third parties. My position is clear: the more centralised the data, the greater the risk of misuse. Even with noble intent, the structure fixes the limits.

In the football market, intermediaries hide the true cost; likewise, in registration systems extra intermediaries add cost and complexity. A simple model with fewer layers is my preference.

What a Good Registry Model Looks Like

Several principles are clear to me. First, the purpose must be set first — welfare or control. Second, the boundaries must be explicit — which animals, which state, which key. Third, verification must be automated and transparent. Fourth, data flows must be limited — only as much as is needed.

These principles hold for football models too. When I build a live xG model, the same questions arise: what is the purpose, where are the boundaries, who verifies. Different domain, identical method.

A Lesson in Data Integrity

Now to the real question. Why did a pet-registration story enter a pipeline carrying the label 'football'?

The most likely explanation: an automated classifier tripped on a spurious keyword or a feed-routing glitch. In football analysis this is not trivial. A mislabeled record entering a large dataset manufactures false positives, corrupts entity graphs, and poisons aggregate conclusions.

A clean dataset can still lie when the crowd is missing. Here 'crowd' means context. The label was clean; the context was false. And data without context can lie.

There is a large danger here — for the analyst himself. An analyst who believes the label and dresses the error in football language will in fact produce a fabricated analysis. My rule is clear: no football entity, no football analysis.

A Contrarian Angle: The Mislabel Is the News

The natural reaction is to skip the error — just write the real subject. To me the opposite is true. The number was clean; the match refused to be. The label is clean; the content is not.

This error is not a mere accident; it is a systemic signal. If such errors recur at scale, the precision of classification models declines, and doubt grows around every downstream decision. The question is not 'who is guilty'; the question is 'what process made this error possible'.

A confession is needed here. My analysis rests on limited information — one document, twenty-two information points, a few organisation names. The sample is small. So I write about mechanism rather than numbers, and I attach a confidence level to each claim. Issuing firm verdicts without a large sample runs against my method.

From Registry to Ledger: The Technology Context

Globally a technology debate is underway about registry systems — centralised databases versus distributed ledgers or blockchain-based registries. Land, identity, supply chains: blockchain registries have been trialled in many domains.

To be explicit: the source content of this piece contains not a single word about blockchain or any distributed ledger. This is general context, not a claim about Mexico. When registration is a model, its technology choice is also a decision — centralised control, or distributed verification. Which is better depends on purpose and governance, not on technological fashion.

Closing Thought: What to Watch Next

I am not ruling on whether Mexico's national registry bill will pass, or whether it would be good. What is clear to me: a proposal, a state system, and a wrong name — three distinct objects, each needing its own timestamp.

In the next phase I will watch three signals. First, the language of the Senate bill — whether the federal framework is defined. Second, whether state systems continue under their own rules or lean toward the centre. Third, and most important — how many more mislabeled records are hiding in the classification pipeline.

Is another mislabeled record waiting on the dataset in front of me?