The Data Vacuum: When ShotLink Goes Silent, Golf Reveals What Cannot Be Measured

Early afternoon in Florida, the big screen in the media center blinked off and came back with a single grey line: “ShotLink feed unavailable.” The fourteen signal towers along the course were still standing, the radar still turning, but the link back to the processing centre had been severed. In the broadcast booth, two commentators had to keep talking with no numbers in hand. The Strokes Gained board froze at round two. The shot-dispersion heat map vanished from the screen. Forty-one minutes later the system returned, and the room exhaled as if it had just escaped a minor technical fault.

I sat in the third row and recorded all forty-one minutes. The interesting part lay elsewhere. Through that entire stretch, nobody in the room could answer a very simple question: where did the second shot at the 15th hole land, and why did the player choose that club.

ShotLink is not software. It is infrastructure. From the early 2000s, the PGA Tour built a network of signal towers, radar and volunteers carrying laser devices along the fairways, recording the coordinates, distance and lie type of almost every shot in every round. From that raw dataset, Columbia University professor Mark Broadie developed the Strokes Gained system around 2026, splitting the game into four areas: off the tee, approach, around the green and putting. Every serious piece of golf analysis in America today stands on those four pillars.

Above it sits a third layer: the storytelling layer. Broadcast graphics, prediction models, betting markets, and writers like me. A multi-layer structure like that has an inherent weakness few people mention. If the bottom layer returns a null result, the layers above are not wrong. They simply become meaningless.

I once received exactly such a report. A long document, with section headings, tables, an assessment framework, even sections titled “hidden risks” and “reference value.” But every data field was empty. No player name. No tournament name. No timestamp. The only surviving label was a single word: golf. The report declared its own failure, and it was so honest that it became useful. It taught me that most of the certainty in this industry is built on pipelines nobody inspects.

In America, golf writing shifted from description to measurement over roughly fifteen years. A writer of the 1990s told stories through the sound of impact and the reaction of the crowd. A writer of the 2020s opens a data table first and writes afterwards. Both have a reason to exist. There is one large difference: the writer who works with sound knows he is subjective, while the writer who works with metrics often believes he is objective.

Those forty-one minutes in Florida were not a rare failure. They were simply one of the rare moments we could see it.

Golf metrics do not generate themselves. Every Strokes Gained value is the product of a chain of human decisions. A volunteer standing behind the green has to decide whether that nine-metre putt belongs to “around the green” or “putting.” A chip from the collar must be assigned the correct lie type. A shot pushed off line by wind must be logged at the correct landing coordinate. Mislabel one entry and an entire column of metrics for the whole round shifts with it.

Which means the numbers we cite on television every Sunday are produced by people holding laser devices, starting at seven in the morning, in the sun, in silence. They appear in no graphic.

The most dangerous thing in golf analysis is not missing data. It is data that looks complete.

When the pipeline breaks outright, you know immediately. Blank screen, empty table, no argument. When the pipeline is contaminated, you do not know. You have a beautiful Strokes Gained table, a colourful heat map, a line reading “according to ShotLink data” in the corner, and behind it a chain of small errors accumulated over six hours.

So I keep a private rule: never use a metric I have not seen fail. If I do not understand how a value is produced, I have no right to use it to judge a human being.

The clearest case is putting. For years, American analytics built the image of Scottie Scheffler as a player who dominates from tee to green but is fragile on the greens. That framing has real data behind it. It also ignores something anyone who has stood beside a green on a Thursday knows: green speed changes by session, by humidity, by wind direction, by when the greens staff mowed. A putting metric does not measure green speed. It measures the final outcome against an average calculated from that same dataset.

It is a closed loop: data creates the baseline, the baseline creates the judgement, the judgement reinforces the data. A player is locked inside a description he produced himself, with no way out by playing better, only by playing differently in the direction the model expects.

Nelly Korda offers the mirror case. In 2026 she won five consecutive LPGA events, matching a mark set by Nancy Lopez in 2026 and repeated by Annika Sörenstam in 2026-2026. Television explained that streak with putting and approach numbers. My notes from those weeks recorded something else: the way she and her caddie walked the last four holes, slower, reading more, talking less. No dataset stores that.

Ludvig Åberg arrived on tour with off-the-tee numbers among the leaders from his first season. The data is accurate. It does not explain why a golfer raised in Eslöv reads wind in Texas that quickly.

Rory McIlroy completed the career Grand Slam at Augusta in April 2026, fifteen years after his collapse on that same course. The data story compresses into one phrase: he finally knows how to close. The unmeasurable part is far longer: fifteen years relearning the same shot, under the same pressure, after losing in every way available. A single season is one sentence in a book that runs a decade deep.

There is one thing I have always distrusted: the heat map. A colourful rectangle, red here, blue there, captioned “shot dispersion.” It looks scientific. It is also the most misleading form of presentation in the entire industry. A red zone does not say the player aimed there, does not say the shot was good or bad, does not say anything about wind or pin position. It only says where the ball stopped. The heat map is golf's new form of divination: it offers an image of the truth instead of the truth.

I write about golf as an observer of silences. Tuesday and Wednesday at every event, before the gates open and before the ShotLink towers switch on, is when the real game appears. A caddie walks a few steps ahead, sets the bag down, presses a foot into the putting surface. A player stands behind the ball for a long time, does not swing, then lays the club down and steps away. Those cancelled decisions never enter the data, because data only records the shot that happened.

When the stands are empty, the match exposes what tactics conceal.

In July 2026, the LPGA returned from its pandemic shutdown at an event with no spectators. Players told me the same thing in different ways: they could hear the ball land, hear their own breathing, hear footsteps of a player on the adjacent fairway. Some liked it. Some lost rhythm. No ranking metric records that difference, but the tournament results did.

The counter-intuitive angle sits here: those forty-one blank-screen minutes in Florida may have been the best broadcast of the week.

With nothing left to display, commentators are forced to describe. They must talk about wind, about grass height, about club choice, about a player hesitating twelve seconds over the ball and then stepping away. None of it has ever been captured by data, and all of it is why viewers stay until the last hole.

Golf analytics has a small secret. Many pieces labelled “data-driven” are really “data-illustrated”: the conclusion is written first, the numbers selected afterwards. I have worked inside such a process and watched it produce reading that flows beautifully, convinces completely, and is wrong at the root.

So I set myself a rule. When a story is eighty per cent clear, I lock it and publish, and the remaining twenty per cent is labelled as probability, never as certainty. Waiting for one hundred per cent is the surest way never to publish anything at all. Coldness is a long-term strategy, not a character flaw.

Once, in a major championship media centre, an older male colleague cut me off while I was asking about a short-hole tee strategy. He suggested I should ask about the golfer's family instead. I did not argue. I went away, rebuilt the entire off-the-tee and approach dataset for the leading group across three rounds, and wrote a piece using only what could be verified. They doubted the voice before they heard the argument. I learned to gather the evidence first and the expectation afterwards.

Golf will not return to a low-data era. Every new season brings more sensors, more cameras, more models, and sooner or later automated analysis systems will return results before a writer has opened a laptop.

That does not make the work easier. It widens the gap between what is measured and what is understood, and puts the analyst back in the same old place: standing between two shores, where the pipeline can break at any moment.

That day in Florida, when the screens came back on, I added one line to my notebook: A blank screen forces me to read the match the way I read an unedited manuscript. That manuscript is still there, even after every metric has returned to its proper place.