Nine Columns, Nine Times N/A: The Analysis That Refused to Invent a Race

It was eleven at night and nine tabs were open on the screen. Each tab was a dimension: technical and car, race strategy, team and driver, competitive landscape, regulation and governance, driver market, risk profile, public narrative, industry transmission. Inside every tab was a column headed 'Evidence'. Nine times out of nine, that column came back empty: 'N/A — insufficient information', 'assessment impossible'.

This was not a technical failure. The document that arrived for analysis carried not a single factual point — no title, no publisher, no team, no driver, no Grand Prix, no date. A framework built to compute sector times, tyre degradation, pit loss and aerodynamic testing restrictions was handed a payload with nothing in it to count.

My first reaction was not that the model had failed. My first reaction was that the model had done exactly what it was built to do — it declined.

Context

The framework runs in two layers. The first layer deconstructs the source: title, publisher, author, article type, information points, core viewpoints, entities involved, time sensitivity, source quality. The second layer builds nine dimensions on top of those facts. The full capability of the second layer rests on the payload of the first. This time the first layer came back empty, so the second had no ground to stand on.

It is worth writing down what arrived and what did not, because that is the only hard fact in this entire exercise. The list of information points was empty. Entities were never extracted — no team, no driver, no circuit, no technical lead. Article type stayed unclassified, which means it could not be established whether the source was breaking news, a technical feature, an opinion piece or a transfer rumour. Time sensitivity was never assessed, so there is no way to place the event within the season calendar or the regulation cycle. Source quality was not recorded, which means no ceiling was set on confidence.

The consequence of that last omission is direct. An experienced paddock journalist's exclusive, a general sports outlet's report, an aggregator's post and an uncredited social post should carry four different credibility grades. With the grade blank, all four sit on the same scale, and every conclusion then escapes its own limits.

Since 2026 every report I file opens with a methods box: source, sample, model error. The habit formed in the year I tried to understand Force India's best-ever season, fourth in the constructors' championship on 187 points. After rebuilding the Baku and Spa collisions from onboard telemetry, race-control logs and 1,240 logged lap times, I published an estimate that the intra-team contact cost the team between 24 and 31 points. I built the model alone and then had a data engineer audit it. After that I stopped taking assignments I could not reproduce myself. I rebuilt the 2026 cost model from the footnotes and the margin changed — which is why I no longer read an empty column as 'low'.

Core

Empty input and weak input are not the same thing. On weak input you can still reach a conclusion at low confidence and attach explicit uncertainty conditions to it. On empty input there is no conclusion, because the subject of the conclusion does not exist. Each of the nine dimensions needs at minimum one named entity and one named event. The technical dimension needs to know whose upgrade it is and which Grand Prix it was built for, otherwise it is impossible to separate following the benchmark from independent innovation. The strategy dimension needs circuit-specific pit loss and stint length, without which neither the net gain of an undercut and overcut nor the cost of a mistimed Safety Car can be priced. The team dimension needs two drivers in the same car — the only control frame the paddock offers. The regulation dimension needs a Technical Directive, a scrutineering outcome or a financial-regulation finding. The market dimension needs one confirmed or reported signing before a domino chain can be modelled; for technical staff, gardening leave length is the variable that sets how much knowledge is still current.

'N/A' does not mean 'low'; it means 'unknown'. The distinction looks small and costs a great deal. Writing a risk rating as 'low' becomes the claim that the subject carries little risk — when the subject itself is absent. So the rating was withheld deliberately, not defaulted. Likewise the four information-value scales — sporting value, industry value, timeliness value, reference value — are unrated rather than low. The only risk that can be stated with certainty concerns the input pipeline, not the article.

On the cost cap, regulation and enforcement must be separated in every sentence. The financial regulations are readable, quotable and endlessly re-litigable; the compliance data is not public. The published rule therefore gets mistaken for the disclosed outcome. With an empty payload the confusion is unusually visible: the text of the rule exists, and there is no fact in the file to say what any team actually spent.

The source-quality blackout is the most expensive gap. Without a source tier, an aggregator's post and a veteran paddock journalist's exclusive receive identical treatment. This is precisely where transfer rumours breed: a leak planted to pressure a contract renewal and genuine reporting look the same. The author's stance and the article's purpose are the two fields that detect the difference, and both were blank.

The minimum viable payload for resubmission rests on six items: title and source, including publisher, author and date; at least three discrete information points, claims rather than summary sentences; one team, one driver or technical lead and one Grand Prix; the three parts of the core viewpoint; an explicit time-sensitivity anchor; and a graded source tier. With those populated, all nine dimensions can run at full depth with confidence tags intact.

Contrarian angle

The paddock has an old habit: if the framework looks empty, fill it. Add a lap delta that reads as approximately right, add a pit window that sounds reasonable, and the reader will never be able to tell what was measured from what was assumed. The habit is more dangerous than a gap, because a gap is visible and a filled framework is not. The model looks elegant until you add the missing matchday revenue column.

The second habit is the borrowed model. Lifting a template from football, media or private equity and dropping it onto Formula 1 maps seductively well at the top line — race revenue, circuit hosting fee, broadcast rights all line up neatly. The danger sits in the denominator. Every borrowed model has to ship with its mapping assumptions, the case that would break it, and the one variable that would reverse the whole transfer. That discipline applied here too: no variable could be found that connected to any team at all.

So the publishable result this time is zero — and that zero is not a shield, it is a measured output. The administration ledger did not lie; it simply left a column unnamed. This document did not lie either — it said nothing. A quiet audit of the 2026 accounts explains more than the press conference did, and an honest 'unknown' carries more information than a decorative conclusion.

Takeaway

The real question is not why nine dimensions came back blank. It is why the pipeline returned empty-handed from the first layer, and how long it will take to fill. If the source is fresh breaking news, the cost of delay is asymmetric — the utility of breaking news decays fastest in the first hours. I will watch three triggers: information points at three or more, a non-empty entity list, and a graded source tier. If any one of them stays missing, the analysis stays where it is.

A framework that can say 'don't know' nine times — is that a weak framework, or the only one worth trusting?