The Difference Between an Empty File and a Zero: When the Cricket Data Pipeline Returns Nothing

19 December 2026, four in the morning, a rented flat in Walthamstow. A fourteen-column tagging sheet was open on the laptop. Thirteen columns were blank. One field held text: cricket_world. The feed coming out of Adelaide that night had passed through a pipeline somewhere between India's collapse to 36 and my screen, and the pipeline returned nothing.

I did not delete the file. Reading an empty file is a skill, and I did not have it before that night. When you have spent eleven years counting deliveries instead of reading scorecards, you learn to separate two things that look identical at a glance: the absence of information and the absence of an event. My sheet was empty. The match was not.

The sheet's columns are deliberately ordered so that the outcome arrives last — bowler, over, release angle, length band, line, batter's backlift, crease position, field quadrant, stump-mic sound. On Kuldeep's deliveries I keep one extra column: googly or stock ball, decided by wrist angle, never by result. That is the whole method. Log what was under control first, and what happened afterwards. My most valuable work as a tactical commentator has never come from a debate; it has come from freezing the same ball repeatedly until I could separately write down its seam angle, the position of the toe, and the quadrant of the field.

Here is the part nobody publishes. A ball-by-ball pipeline dies in three ways. Capture failure — the broadcast feed never arrived, or arrived at half length. Template mismatch — the feed arrived but the parser did not recognise the scorebug format it was written for, so it discarded everything and reported success. Language and encoding — the character set of the country where the match was played was never registered in the pipeline, and the output quietly described itself as clean and empty. All three look identical on screen. All three have different causes.

So my first rule is that I do not call an empty file a zero. A zero is a measurement — no runs off this over, no wicket in this spell — and a measurement requires a living instrument. An empty file is a death certificate for the instrument, not for the event. In the regular season, when I look at a team's numbers and one column is consistently and crisply populated, my first question is how that column is being populated. Clean data is usually just data somebody built plumbing for.

A scorecard is a lossy compression. It turns three hundred deliveries into one line. Take Kuldeep's spell at Trent Bridge on 12 July 2026: ten overs, six for 25, England 268 all out. The card gives you 6/25. It does not tell you how many of those sixty balls were wrong 'uns, how many were stock, or how often an England batter put his foot in exactly the same place he had put it the ball before. I tagged all sixty by wrist angle rather than by outcome. What that sheet showed was that England collapsed on reading, not on flight. That distinction never reaches a scorecard, because 'reading' is not a field in the scorecard's language.

And an empty feed is total loss — not even compression. What happened in Adelaide that night I rebuilt over nine days across a sheet of 1,100 deliveries, and to do it I had to hold three separate artefacts against each other: the host television feed, the stump-mic audio, the scorecard. The finding was that in the second innings, bat paths began roughly four centimetres wider outside off stump. The collapse was in the entry path of the bat, not in the pink ball. That conclusion did not come from the sheet. The sheet was empty. It came from rebuilding the sheet by hand.

This is where I have to catch my own worst habit. My last column is 'what changed', and I fill it in far too often before the ball has landed. The professional honesty is this: I wrote the frame down before I understood it. The frame goes down early, understanding arrives late, and in the gap my ego is busy defending the frame. In 2026 I built a full model around a domestic spell — 'he is dropping it into the slot, therefore he is getting hit' — and then the footage arrived. He was not dropping it into the slot. He was landing it two inches past the slot, and my sheet did not scroll at a zoom level capable of catching two inches. The model was not wrong. The model was the right question answered badly.

The real crisis in sports journalism is not the empty feed. It is the habit of filling it. When data does not arrive in the commentary box, adjectives do — 'lost his rhythm', 'under pressure', 'found his rhythm again'. These sentences sound solid and contain not one element that can be falsified by the next match. I banned the word momentum from my copy after 2026, because that word occupies precisely the space where data has not reached. What I watched from a twelve-pound standing ticket at The Oval on 18 June 2026 was not momentum. It was Mohammad Amir's seam angle, and India's batters holding the same crease position ball after ball. Pakistan made 338/4, Fakhar Zaman struck 114, Amir removed Rohit Sharma, Virat Kohli and Shikhar Dhawan inside his first three overs, and India were 158 all out. I went home, recorded the host feed, froze every delivery, and cut a nine-minute breakdown. Momentum does not appear in it once. The views did not arrive for momentum either.

Everything in my method depends on sequence: I log the spell first, then ask what it changed. Reverse the order and you are doing what the commentary box does — deciding the story, then hunting for deliveries to fit it. And when the deliveries cannot be found, the story stops needing them.

Now the part I could not solve. In the 1,100-delivery Adelaide sheet there are eight balls for which no bat-path explanation works. Three different batters, three different overs, the same straight ball left alone while it was heading at the stumps. I have no variable for them. I could have filled the gap by calling it psychology, except psychology is not a measurement. So those eight balls sit blank in the sheet to this day. They are the most honest part of it.

The notebook is not a museum; it is a warm-up. I do not keep a separate sheet for every match; I keep one sheet that grows, gaining a column each year and occasionally losing one. After Adelaide I deleted the field-quadrant column, because it changes too fast during a match for my tagging to hold it. A column you cannot fill is a column that flatters you.

At 4am in Walthamstow, the collapse sounded like a lesson. The lesson was not that India were bowled out for 36. The lesson was that my machine did not record the moment and I had to write it by hand. Next time a pipeline hands you a clean, empty output, do not read it as nothing happened. Read it as my instrument died, and go find the second artefact immediately — the stump mic, the host feed, the toe planted on the crease. One call, stated once, at 0.85 confidence: before the next series I will publish the field list of my tagging sheet, and at least two of those fields will be ones no broadcast graphic has ever carried. On the day they start carrying them, the word rhythm will have less room to stand in.