Follow the leader - Has it ruined offshore racing?

It was off the Outer Hebrides, August 2022, during the Sevenstar Round Britain and Ireland Race. We were on Morning After , my rebuilt S&S 34 from 1967, working our way around the top of Scotland, and the next big decision was at the Butt of Lewis before making course towards Muckle Flugga, up at the northern edge of Shetland. It is the sort of place that concentrates the mind. The land is running out. The sea feels bigger. The weather has room to do what it wants, and the decision is no longer theoretical — get it wrong there and you can spend a long time paying for it. Our weather intelligence setup was, by modern standards, comically analogue. A satellite phone connected to a mobile phone and portable router — which had to have a clear view of the sky to work at all — pulled GRIB files down into the onboard PC running Seatrack. Old software, certainly, but the mathematics behind it was doing the same fundamental job as the routing engines in Expedition and Adrena today: take the weather, take the boat’s predicted performance, run the options, find the quickest way through the problem. Alongside the screen there was a barometer — still, in my view, the single best instrument for anticipating a weather change — the BBC Radio 4 shipping forecast on longwave, and a small black box that had started life as a fax machine and downloaded forecast data onto what amounted to a tiny green-screen display. Not exactly a professional weather routing operation. Then again, I am not exactly young. We used what we had, and we used all of it.

We were well out of AIS range of any other boat in the class. The brief windows when we had enough satellite signal to pull down the Yellow Brick tracker gave us a rough sense of where the boats in our fleet were, and not much more. For practical purposes, we were on our own. The sea off the Outer Hebrides was telling a slightly different story from the model. The wind was not behaving quite as cleanly as the GRIB suggested, and the pressure, the feel of the boat, the sky and the water all gave that familiar offshore warning: the screen may be right in principle, but wrong enough in practice to cost you the race. There was no fleet to cross-check against, no AIS contact, nobody else’s tack to follow. It was just the boat, the weather, the software and the question of which one to believe.

So before committing north-east towards Muckle Flugga, we made the tactical decision that the routing line had not quite earned. We trusted what we were seeing over what the software was drawing. The next forty-eight hours were brutal — hard upwind in twenty-five to thirty knots, thirty degrees of heel, no relief — and then we ran into a wind hole off Muckle Flugga. A satellite phone call had told us we had been leading the race overall, yes overall on IRC, for eight days. That is the power of the classic S&S design: upwind just fast, straight downwind fast, if you can hold your nerve and stop the roll. Reaching always a nightmare. We drifted around the most northerly point of the course and set off for Fair Isle and the Scottish east coast on a punishing reach. We knew we had lost the overall lead. The GRIBs had not predicted the hole accurately — but to be honest, we could not have sailed around it regardless. We did enough to win Class 4 in the Round Britain and Ireland Race, and a good overall finish for the oldest boat in the fleet. What that race taught me — or perhaps confirmed — is something that has been growing more relevant ever since: the technology that helps us find better answers can also persuade whole fleets to ask the same question, accept the same answer and sail into the same piece of water. Getting the weather picture wrong, or getting too many conflicting pictures and following none of them properly, is not really a technology problem. It is a thinking problem. And if you understand it properly, it is also an opportunity — which is what this article is about.

1. How We Got Here Cast your mind back to the Admiral’s Cup era. Routing was done with paper charts, a pencil, pilot charts, hand-drawn tidal vectors, weatherfax, radio forecasts and the accumulated judgement of whoever you could persuade to sit behind the chart table at two in the morning. The competitive edge in navigation was genuinely scarce — it lived in the heads of a relatively small number of people who had enough miles, enough failures and enough weather scars to read a synoptic chart with real conviction. The RORC (Royal Ocean Racing club), always progressive, sensibly introduced technology into the offshore rules as they became affordable for all. Offshore racing remained competitive and relevant and so it continues Software did change things in a leap alongside reliable hardware. Expedition became the navigation package of choice for many serious offshore racing programmes. Adrena developed its own deep following. PredictWind, Squid, SailGrib, LuckGrib, qtVlm and TimeZero widened access to routing tools that would have seemed like science fiction to the navigators of the 1970s and 80s. At the same time, the weather models improved and became easier to access. GFS, the American global model, became freely available and widely used. ECMWF, called “EC” by most sailors, became the high-quality benchmark that professional navigators wanted to compare against. Higher-resolution local models, ensemble forecasts and now AI-assisted forecasts have added further layers. AI assited auto-helms are now going to start having a big impact this year as well (2026). The result is, in many ways, extraordinary. A club IRC boat leaving Cowes for a Channel Race, a Class 40 heading across Biscay and an IMOCA 60 preparing for the Southern Ocean may all be looking at the same global model families, broadly similar routing outputs and broadly similar strategic choices. That is genuine progress. But it has a side effect that the sport has been slow to reckon with honestly: when the same information is available to everyone, the information itself stops being the edge. The old navigator could win by knowing something the fleet did not know. The modern navigator is more likely to win by interpreting shared information better than the fleet, by recognising when the information is incomplete, and by having the courage to trust the evidence on deck when the software is pointing somewhere else. That is a very different game — and most crews have not yet adjusted their thinking to match it.

2. The Mechanics of Convergence To understand why fleets bunch up, you need to understand what routing software is actually doing. In simple terms, the software takes three main inputs: a weather model, a boat polar, and a start and finish point. The weather model provides forecast wind speed and direction across a grid and through time. The polar tells the software how fast the boat should sail at different wind speeds and angles. The routing engine then calculates, thousands or millions of times, the combination of headings, tacks and gybes that should get you from A to B in the shortest elapsed time. The output looks persuasive because it is precise, a line on a chart, an ETA, a sequence of manoeuvres, a neat answer to a messy offshore problem. The difficulty is not that the algorithm is wrong. Often it is broadly right. The problem is that when most navigators in the class are using the same global models, broadly similar routing engines and broadly similar polars, their answers will tend to converge. Routes cluster, boats cluster, and thinking clusters and when thinking clusters, the fleet starts to look less like a race and more like a convoy. That is where the danger begins, because a navigator can easily mistake a common answer for a correct answer. The fact that the fleet is going somewhere does not prove the fleet is right. It may only prove that the fleet has asked the same question of the same data and accepted the same output. In the 2026 Myth of Malham Race, the routing was almost unanimous: take the mainland shore in the shallow water towards Yarmouth, push north into Christchurch Bay, then come south to Durlston Head off Swanage. I took the opposite view and tacked along the Island shore instead. By the time we reached Hurst Narrows and the Sconce cardinal, we were around one and a half miles behind the leaders in our class — behind the fleet, behind the routing solution, and with a crew who were beginning to wonder whether their skipper had lost the plot entirely. The pressure came when I held my line past the Shingles bank and took a shallow track close in to Totland and Alum Bay, heading south-west. Everyone on board wanted to tack onto port and follow the fleet north. The questions were polite, the faces were not. And off the Needles fairway buoy, I gave in. We tacked north to join the convoy. Less than a mile later, looking at the tracker, I could see the fleet to the north had slowed. We tacked back and resumed the route I had planned before the pressure got to me. From the Needles to Portland Bill we were one of the fastest boats in the fleet. Post-race analysis told the rest of the story clearly enough: roughly ninety-five per cent of the fleet had followed each other into Christchurch Bay, the routing software had pointed them all there, and most of them had been slower for it.

The moment I tacked north off the Needles is the moment I want you to remember as you read this article. Not the result — the moment of doubt. That is where the follow-the-leader problem actually lives: not in the software, but in the human being looking at the tracker and losing confidence in their own reasoning because everyone else seems to disagree. Simon Fisher — one of the most respected professional navigators in offshore racing, winner of the Ocean Race and the Magnus Olsson Prize — described the trade-off with characteristic precision in a 2025 interview with Yachting World: “Good routing comes down in large part to good risk management. A big part of that is how much risk you place on the weather forecast or a routing option, versus what the rest of the fleet is doing. It’s a strategy versus tactics trade-off.” When the models agree, when the ensemble routes are tightly packed, when the weather pattern is stable and the fleet is compressed around a rational solution, the answer may genuinely be to stay with the fleet and execute better. But when the models disagree, when the ensembles scatter, when the deck is telling you something the GRIB is not, the game changes entirely — and that is when races open up. 3. Automation bias: The quiet enemy on the chart table There is a name for part of this problem, and it is worth knowing: automation bias. It is not stupidity and it is not laziness, it is what happens when a system is right often enough that people stop interrogating it properly. The routing software gives you a clean line, a tidy ETA, a neat sequence of tacks and gybes and a number that looks more objective than the uncomfortable evidence building on deck: the darker patch of water under the land, the pressure trend that has not matched the last GRIB, the rival two miles away sailing a different wind angle, the cloud line developing faster than the model suggested, the sea state knocking half a knot off the speed the polar promised. The danger is not that navigators use routing software. The danger is that they allow it to become the senior decision-maker on board. The best navigators I have sailed with do not worship the route — they interrogate it. They ask what assumption is buried inside it, what happens if the front is six hours late, whether the headland breeze fills differently than the model expects, whether the boat is actually sailing to the polar being used, and whether the fleet is making a genuine strategic choice or simply copying the same first answer from the same data. That distinction matters more than any single piece of software. The software can calculate a route. It cannot take responsibility for the race.

4. The Three Layers of Convergence The follow-the-leader dynamic is not one thing. It operates on three levels simultaneously, and understanding which layer is in play at any given moment is the key to exploiting it. Layer one: weather model convergence. Most serious offshore navigators are looking at the same major model families. GFS is global, freely available and runs out to sixteen days, though useful detail reduces with forecast horizon. ECMWF is widely respected and heavily used. Higher-resolution local models may refine the picture, but in many offshore races the strategic framework is still being shaped by the same broad model guidance. When the models agree, routing solutions stabilise — but it is not enough to look at the latest GRIB. You need to look at model behaviour over time. Has one model been drifting consistently? Has one been persistently late with a front? Are the ensembles tightly grouped or spread across half the Channel? The route is not the forecast; it is only the software’s answer to one version of it. Layer two: software convergence. Expedition, Adrena, PredictWind, Squid, SailGrib and their various competitors do not all work identically — the interfaces, features, routing controls and underlying assumptions differ, and some are considerably deeper than others. But if you feed similar software the same weather data, similar polars and the same start and finish, you should not be surprised when the outputs begin to look similar. The weakest navigator asks what the route is. The better navigator asks why it is the route. The best navigator asks what would have to be wrong for that route to lose — and that final question is the one most sailors never get to. Layer three: polar convergence. Boat polars have improved enormously. ORC’s VPP, manufacturer polars, sailmaker performance data, logged boat data and modern instrument calibration have all made it much easier to model how a boat should perform at different wind speeds and angles. That is useful, but it is not magic. Your polar is still a prediction, and it may not know your sail inventory, the tired leech on your old No.3, the way your boat bogs down in short chop or the fact that your crew loses ten per cent of performance after midnight. The boats that share similar rating profiles and similar real-world performance will often be routed towards similar waypoints — but the boat that has built better real-world performance data can sometimes break the pattern. Most amateur crews think of polars as something they download. Serious crews think of them as something they build.

5. AIS, trackers and the end of hiding Here is where the follow-the-leader dynamic becomes genuinely complicated — because in most modern offshore races, the fleet is no longer sailing in the informational darkness of the old days. Back in the day the fun was the French boat with the dimming stern light on a resitor that meant you thought they were sailing away. It’s just a story but sums up what sailing at night used to be like! AIS is now standard practice in offshore racing. In RORC races it is mandatory, with competitors expected to transmit the boat name, not just an MMSI number, and at its useful range of ten to twenty miles every boat in your vicinity is visible in real time. That changes behaviour in ways that cut both ways. If your nearest rival is two miles to leeward and their bow is lifting, that is information — and if their AIS track has shown a consistent ten-degree header over the past twenty minutes, you may be looking at better real-time wind intelligence than any model. Using AIS that way is not copying. In the right hands, it is navigation. But the darker side is real. When every boat is watching every other boat, herd behaviour is reinforced in real time. The navigator who was considering a split may lose conviction when the tracker shows the fleet going the other way. The rational individual response, repeated across the fleet, becomes collective conservatism — and the convoy tightens further. Satellite trackers — YB Tracking and similar systems — have added another layer. In many offshore races the whole fleet picture is now visible to competitors, shore teams and armchair tacticians, and some race tracker feeds can be pulled directly into navigation software. The tracker is no longer just entertainment for people at home. It has become part of the tactical environment, which means the fleet race has quietly split into two simultaneous contests: the race on the water, and the race to interpret shared information better than everyone else. Everyone can see the dots. Not everyone understands what the dots mean. 6. The AI Weather Layer There is another layer arriving quickly, and it is worth understanding what it does and does not change. ECMWF’s Artificial Intelligence Forecasting System, AIFS, is now operational, with an ensemble version running alongside the traditional physics-based model. That is a serious development. It may give sailors faster forecasts, different model behaviour and, in some situations, better guidance at shorter lead times. It will almost certainly become part of the offshore navigator’s standard toolkit within the next few years.

But it will not remove the herd problem. It may sharpen it. Once the same AI-assisted forecasts are distributed through the same routing platforms, the fleet will simply converge around a new set of answers. The names of the models will change; the human problem will not. The edge will still belong to the navigator who understands uncertainty rather than just output and the relevant question will not be “what does the AI say?” but “where is the AI likely to be blind, and what evidence do we have on deck that it has missed something?” Modern offshore racing is not moving from human judgement to machine judgement. It is moving towards a harder version of human judgement, because the machine will often look more confident than the situation warrants. And confidence, as any experienced offshore sailor knows, is particularly dangerous when it comes from a screen. 7. Where the Model Fails — and Why That Is Your Edge Here is what routing software cannot do. It cannot see the line of dark water under the headland that tells you the land shadow extends further than the model shows. It cannot feel the pressure building on the backstay twenty minutes before the wind shift appears in the GRIB. It cannot read the tide rips off the Casquets well enough to tell you whether the west-going stream will give you half a knot or a full knot of lift on the final approach. It cannot know that your headsail, cut slightly flatter than the class norm, carries better in twenty knots than the polar suggests. It cannot tell you whether the crew on deck are sharp, cold, hungry, seasick or quietly losing concentration. These are the gaps. And they are not small gaps — they are the gaps that separate the fleet. Simon Fisher’s advice to keep your head out of the boat is not a throwaway line. It is the central discipline of modern offshore navigation. The screen is only one source of truth, and it may be the most organised and most seductive source, but it is not the only one. Coastal gates. The Fastnet Rock approach, Portland Bill, the Casquets, the Lizard, Start Point, the Mull of Kintyre, St Catherine’s, the Needles, Anvil Point, the Butt of Lewis, Muckle Flugga anywhere the coastline distorts the broad-scale weather picture, you need to be careful. A global model may be resolving the atmosphere at tens of kilometres; a tidal race may punish or reward you at the scale of a boat length. No routing software bridges that gap reliably, and the navigator who has sailed those waters fifty times knows something the algorithm does not. Transition periods. The approach of a front, the edge of a ridge, the collapse of a gradient breeze, the arrival of a thermal, the moment the tide changes but the wind has not yet decided what it wants to do, these are the moments when the fleet becomes most vulnerable to lazy routing. When the models diverge, the question to ask is not simply which model gives the best route but which model has behaved most consistently over the last three or four runs, which one has been drifting and which one best matches what you can observe now. The fleet will often chase the most attractive routing line. Good navigators ask whether it is attractive because it is likely, or because it is flattering. Polar reality versus polar theory. Your routing is only as good as the polar you feed it. If your boat cannot sail to the numbers, the software is optimising a fiction. If it beats the numbers in a particular condition, the software may be too conservative. If the polar assumes a sail you do not carry, a crew weight you do not have or an instrument calibration you have not checked recently, the route may be clean and still wrong. The best routing input is not a pretty polar diagram, it is a brutally honest one.

8. How to Use the Herd The follow-the-leader dynamic is not only a problem. Used deliberately, it is a weapon. Cover smartly, not blindly. When the models genuinely agree, when the ensemble families are tight and the fleet is compressed for good reason with conditions stable, you are unlikely to gain much by being clever. That is not the moment for a heroic split — it is the moment to sail the boat properly. Trim better. Steer better. Change sails earlier. Hit the tidal gate. Eat before you get stupid. Sleep before you get proud. In a compressed fleet, consistent execution is its own form of strategy, and you may gain thirty minutes by not making the small, accumulated mistakes that everyone else makes when they get tired. Identify the moment of divergence. In every offshore race of meaningful length there is usually at least one moment where the race opens up — the models disagree, the tide gate becomes marginal, the ridge weakens, the breeze fills from the wrong side, the majority commits to one version of the future while another version is still available. That is the decision point, and your position in the race matters. If you are leading on corrected time with a sensible cushion you do not need to gamble like a boat in eighth place. If you are mid-fleet and need to move, the moment of model disagreement may be your chance. Risk is not a moral quality — it is a tool that destroys races when used stupidly and wins them when used properly. Use the tracker against itself. The tracker tells you where the fleet is, and more importantly it tells you what the fleet believes. When you see the convoy converging on a gate, you are looking at the collective routing output. The question to ask is not “where is the fleet going?” but “why is the fleet going there?” — and then the harder question: “do I agree?” Sometimes you will. Sometimes you will not. The difference between those two answers is where your race actually begins. Build your own polars obsessively. This is not glamorous work — it is logging, calibrating, checking instruments, writing down sail combinations, comparing target speed with real speed, reviewing tracks after races, asking honestly whether a slow leg was caused by a wrong route, bad trim, tired sails, a cold driver or a polar that was simply nonsense. But this is where amateur fleets consistently leave large amounts of performance lying around uncollected. If your routing inputs are more accurate than your competitors’ inputs, your routing output will diverge from theirs in the direction of reality. That is a real edge, not a fashionable one. Read conditions, not screens. The most underrated skill in modern offshore navigation is the oldest one. Look outside. Read the water, the sky, the pressure. Watch the boats around you. Listen to the helm. Feel the sea state. Ask the watch coming on what has changed. The navigator who can come on deck at 0300, look at the sky and say “the model is late” is still worth more than another laptop running another route. Miles still matter. Experience still matters. Judgement still matters. 9. Five Questions Before You Follow the Fleet Before committing to the routing line, ask five questions. First: are the models genuinely agreeing, or are you only looking at one attractive routing line without checking whether the ensemble supports it? Second: are the ensemble routes tightly packed, or is the software hiding significant uncertainty behind one confident-looking answer? Third: is the fleet going there because the weather genuinely supports it, or because everyone copied the same first solution from the same data? Fourth: does your boat actually sail to the polar being used, in the sea state you are in, with the sails currently set and the crew currently awake? Fifth — and most important — what does the deck say? Not the laptop. The deck. The clouds, the pressure, the water, the tide line, the helm load, the boats around you, the feel of the boat. If those five answers all point the same way, follow the route and sail the boat hard. If they do not, the race has just become interesting.

10. The Deeper Point — Offshore Racing Has Changed There is something genuinely interesting about what has happened to offshore racing over the past two decades, and it is worth discussing; The technology designed to distribute competitive advantage has, in many situations, compressed the old routing edge almost to zero at the top of the fleet. Everyone has the models, everyone has the software, everyone can see the fleet and everyone can look clever on the chart table. But that does not mean the racing has become easier. It means the opposite. The old offshore game had more informational dark corners a clever navigator could sometimes hide a strategic advantage over the horizon, and the fleet would not find out until the next sched. The modern game has fewer hiding places. The fleet can see more, know more, compare more and react faster. The premium has shifted. Less oracle, more judgement. Less mystery, more execution. The modern offshore navigator has to use the technology without becoming obedient to it, sail fast when the route is obvious, stay patient when the fleet compresses, and be brave only when the evidence genuinely justifies it. The era of the navigator as the one person on board who knows something no one else does is largely over — the technology democratised that knowledge. What it could not democratise is judgement. And judgement, as it has always been, remains the scarcest commodity offshore. The fleet will follow the routing software. Let them. Your edge is in the part of the race where the software is incomplete, the conditions are local, the fleet is nervous, and the only reliable answer comes from weather sense, boat feel, hard-earned experience and the courage to think for yourself. Offshore racing is never just about the boat, the forecast, or the routing software. It is about decisions made under pressure, in changing conditions, by tired people trying to keep the boat moving fast and safely in the right direction. That is where races are won and lost.