A Fuel-Price Wire Story Landed in the Tennis Drawer: On Dirty Data in the Sports Newsroom

A Fuel-Price Wire Story Landed in the Tennis Drawer

The clock on the Sydney newsroom wall read 2:47 a.m. I was tidying the folders tracking the ATP's late-season swing when a wire report dropped into my inbox. Short headline: Pakistan's government raised petrol by 4.42 rupees a litre and high-speed diesel by 6.10 rupees a litre. Beneath it were a few lines about Brent crude, about WTI, about the Oil and Gas Regulatory Authority, about an effective date of September 15, 2026.

The final line of the report was what made me stop. It carried a classification tag: tennis.

I read it three times. No player. No tournament. No ranking, no surface, no scoreline, no set. Just fuel prices, crude-oil markets and a national energy regulator. And yet it sat in my drawer.

Some things only surface when you agree to sit still longer than one set. That night I sat still far longer than one set.

Context: how a story loses its way

In nine years of covering tennis from the stands and from the newsroom, I have grown used to the fact that a single report passes through at least four stations before it reaches a reader. Station one is the wire or the press release. Station two is the automated tagging system, where software reads the headline, the opening and the recognised entities to guess what the piece is about. Station three is the section editor. Station four is the writer, which is me.

At station two, the software does not read to understand. It reads to match patterns. It counts how many entities it recognises, what type they are, and how close together they sit in the text. When a piece is short, thin on proper nouns, and heavy on numbers and acronyms, the system tends toward what I call a shallow match: it latches onto a surface signal and labels from there.

Here is the point most newsrooms leave unsaid: a domain label is not a judgement about content, it is a guess about probability. And every probability guess carries an error rate.

The trouble is that the error rate only gets caught when a human sits down to cross-check. In a sports desk, that human usually does not exist. The workload is designed for speed, not for verification.

I once freelanced for a local football site in Sydney while the A-League was suspended during the pandemic. Over three weeks I tracked five players, logging sprint output every session and comparing it with their own previous season. Sprint efficiency had fallen 12 percent, while the coaching staff had projected a fall of roughly 5 percent. That 12 percent means nothing on its own. It only means something because I know where it came from, how it was measured and what it was measured against.

A number without provenance is like a story with the wrong domain tag. It still exists, still reads, and can still do damage.

The null audit

That night I did something I still do whenever a report looks suspect. I opened a blank spreadsheet, laid out the columns that any tennis analysis requires, and filled them with what the report actually contained. I call this a null audit.

Numbers do not lie. You simply have to ask the right question.

The first column is technique and tactics. A serious tennis analysis needs to know where a player sits on their technical arc, whether the forehand is advancing or retreating, how the second serve is being protected, how clutch-point handling has moved across tournaments. The fuel-price report contains not one line in this category. The column is empty.

The second column is data and form. Here you need first-serve percentage, points won on first serve, points won on second serve, return points won, break-point conversion, deciding-game win rate. The report does carry numbers: 4.42 rupees, 6.10 rupees, Brent up 2.6 percent to 107.33 dollars a barrel, WTI up 2.5 percent to 102.56 dollars a barrel. Those are real figures, with units and timestamps. But they measure fuel, not a single person on court. The column is empty on tennis, full on economics.

The third column is tournament system and schedule. A proper analysis needs the tier, the points on offer, whether entry is mandatory, where the event sits in the calendar, and what path the draw opens. The report carries two dates: a review date of September 12 and an effective date of September 15, 2026. Those are pricing cycles, not tournament cycles. The column is empty.

The fourth column is the wider landscape and player positioning. To judge a player, I need to know which of the four familiar tiers they occupy: title-contender, top-10 seed, top-30 backbone, top-100 fringe. I need to know whether their generation is strengthening or fading. The report names no person at all. The column is empty.

The fifth column is rules and governance. Here I need the issuing body, the precedents, the compliance risk level. The report cites Pakistan's Ministry of Energy and the regulator OGRA. Both are state energy institutions. Neither is the ITF, the ATP, the WTA or a Grand Slam committee. The column is empty.

The sixth column is team and personnel management. No athlete, no coach, no agent, no support staff. The column is empty.

The seventh column is risk. The report does state real risks: oil-supply disruption of up to 4 percent of global supply, alongside attacks on shipping in the Middle East. Those are serious risks, but they belong to the energy supply chain. There is no injury risk, no points-defence risk, no doping risk, no match-fixing risk. The column is empty.

The eighth column is media narrative. The report builds a clear storyline: a sixth consecutive increase, global oil above 100 dollars, cost-of-living pressure. That is a story about household budgets and national energy security. No player is hyped, no greatest-of-all-time debate, no expectation gap. The column is empty.

The ninth column is industry transmission. This is the only column where I can imagine a plausible link: fuel costs feed into the travel costs of tennis delegations. But I have to be blunt. The report never mentions tennis. Any link of that kind is my inference, not the source's data. I log it under "low confidence" and I do not use it to reach a conclusion.

Nine columns. Eight entirely empty. One containing nothing but an inference I built myself.

The audit result was unambiguous. This story did not belong in my drawer. It belonged somewhere else entirely.

The contrarian angle: the fault is human, not mechanical

My first instinct was to blame the tagging system. That instinct is comfortable, because it hands you an invisible culprit to criticise without having to look in the mirror.

But when I traced the story's path, I found the machine had only done its job. It applied a wrong label, and a wrong label can always be fixed with one click. What actually pushed the story to me was not the machine. It was a chain of human decisions: the person who set a daily article quota, the person who approved an automated ingestion feed with no verification step, the person who chose not to re-read a headline before passing it on.

In my trade there is a very specific temptation. When a story tagged 'tennis' lands in your lap, and you have a quota, and the story has enough numbers to fill a long piece, you will want to write it. You will find a way to turn oil prices into a metaphor for financial pressure in tennis. You will hunt for a player with a sponsorship tie to an energy conglomerate. You will produce something that sounds persuasive, sounds analytical, and is entirely hollow.

I nearly did it. I opened a fresh page and typed the first line.

Fans have every right to live in emotion; my job is to live in data. If I write a tennis analysis from a source containing no tennis, I have not made a technical error. I have made a professional-ethics error.

One lesson came from the Douglas Costa affair two years ago, while I was covering Sydney FC. A well-placed source put the transfer fee at two million dollars. Colleagues published the figure at once. I checked the registration paperwork and found the real number was 1.2 million. I waited. Two days later the club confirmed exactly 1.2 million. The wrong reports had to be corrected.

The lesson was not that I am better than my colleagues. The lesson was that speed is a kind of cost, and that cost is usually paid by the reader, not the writer.

There is a paradox here that anyone in this job will recognise. The wrong story that spreads fastest is the one that sounds most interesting. A piece about fuel prices dressed as tennis will carry a headline that provokes curiosity, and it will be shared more widely than a correct piece about a dull second-round match. The algorithm cannot tell truth from fluency. It only measures engagement.

So what needs fixing is not the software. It is who we reward, and for what. If a newsroom pays for volume, it will get volume. If a newsroom pays for accuracy, it will get accuracy, along with fewer pieces.

Signals to keep tracking

I do not remember what I wrote. I remember what I counted.

That night I counted nine columns, eight gaps and one inference I discarded myself. I saved the spreadsheet, tagged it "domain misclassification" and routed the story to its proper pipeline. Then I wrote a three-line internal note to the editor who owns the ingestion system.

The three lines read as follows. One: re-audit the domain classifier currently labelling energy and macro-economy reports. Two: add a manual cross-check step for any story tagged as sport that contains no person, event or sports organisation. Three: treat every label failure as a test case, not an isolated incident.

Three months later I still have no reply. But I am still counting.

There is one question I want to leave with people who do this job. When a mislabelled story lands in your lap, just long enough, just rich enough in numbers, just convenient enough for your quota, what will you do? Will you write, or will you open a blank spreadsheet?

Your answer decides whether you are a reporter, or merely a signal relay. The beat keeper does not write the music, but without him everything falls out of time.