complicated game

Kevin Hall, one of the best and successful nice guy sailors ever, gives you plenty to think about…
“Mr. Herreshoff agreed that computers provide certain information about a boat’s performance and the action of the wind, but a ”super sailor will sense a lot of other things – all those nuances that exist in your mind after 20 years of sailing.” In an age of winged keels, plastic sails and black boxes in the cockpit spewing out digital data, it is comforting to know that the feel of the wind on the cheek still counts.” Read more.
Sailing is so cool. You can do it in a tiny boat with no instruments. You can do it in a huge boat with lots of instruments. You can enjoy the subtle challenges on big boats around whether the instruments are helping – they must be! It says “TRUE” right on the display! – or hurting: well, Bob, it seems like every time we sail target speed we just dribble into the boat to leeward and don’t go forward much at all….
Here is a fact: we are using fundamental concepts and hardware from circa 1987 on the most cutting-edge keelboats in the world. While the way we use them now is highly evolved, what we are using has barely changed in over thirty years. Perhaps my biggest lament in my sailing career is that I never got to sail with Juan Vila or Stan Honey to see how they do it. (Most of the rest of us are just bluffing and hoping nobody notices!) While you and I may never get to sail together either, I can certainly jot down a few things I’ve noticed going on between sailing instruments and their end users, and hope your time reading about those things feels well spent.
In this multi-part series, I’ll share a little of my experience as a racing navigator and AC testing manager. You will come away with a better understanding of your instruments as fundamentally flawed tools. These tools can help win races if working well and used as one of many types of input. However, they can also mislead if they are working poorly. Even trickier, they can be “working perfectly” on a synoptically unusual day and give terrible advice. Advances in technology make some dreams about a new way to do the whole thing worth considering. I’ll share my personal vision for this in the final article.
For now, we will assume that everything is calibrated perfectly. Not because that is a reasonable assumption, but because we’re going to geek out enough without worrying about calibration.
Fluids and fields. The boat moves through and disturbs a what-you-see-is-what-you-get fluid, evidence of which we can easily see – bow waves, or a wake which betrays things like resistance and leeway. We can easily measure, or reliably model, the angle to centerline of the water moving under the boat. We can easily measure, and reliably calibrate, the velocity of the water moving past the boat. Now we have everything the instrument system needs to know about what’s going on down there.
Basically, we measure boatspeed and do something about leeway, and get on with it. If for some reason, we sailed in a place where there was 6 knots of current going one way at the bottom of the keel and 4 knots of current going sideways to that at the boatspeed sensor depth, and 2 knots headed somewhere 30 degrees different again at the waterline, it would be much harder to decide what to call leeway or boatspeed. Fortunately, that sort of thing is very rare.
The sailplan also moves through a fluid, called air. Also three dimensions. Harder to see, but there are things like luffing sails, and tipping boats to remind us some forces are at play. Here’s where the rubber comes right off the road and goes on vacation: we still sample that highly dynamic field in one place, do some trigonometry, and call it “truth”, just like we did in 1987.
Emirates Team New Zealand owns a data set from an America’s Cup Act in 2005 in Malmo, Sweden which will help make the point. I ran around after navigating a day’s racing saying “hey, guys, you should see the data! We sailed all day without ever tacking! It says right here in the data!!” The sweaty, exhausted grinders looked at me like I had a propeller, or something, coming out of my forehead. Of course we had tacked – the boat tipped left, it tipped right. The genoa snapped in on the port side, then the starboard. But meanwhile, the “TRUE wind angle” volleyed from -80, to -10, and back again. “Ah! Well, that must have been quite a lot of Shear!” I hear some of you saying. I guess the water was really cold, or Thor’s hammer was spinning backwards, or something that day.
The instruments told one version of the story, as measured way up high in the sky, while far below, the telltales told a different story. On that particular day, it was easy to walk away from the TRUTH, because down on deck it was so obviously a lie.
Let’s pretend for the rest of the article that God decides there is never again directional shear, in either hemisphere, in reality or in the instruments’ depiction of that reality. One problem with 1987-but-also-2017 would be solved. There’s still another problem. You know how some days the water is almost glassy but the boat tips right over and sails anyway? That’s “Wind Weight” or “Wind Profile”. It behaves with regard to our instrument system and historical data something like this: “Compared to an average day, a heavy windweight wind is pushing in total on the sails more than the number we are calling ‘TRUE wind speed’ says it should.
By contrast, a light windweight wind is tipping us over less than the windspeed says it should”. That glassy day, the windspeed measured only at the top of the rig would output a number much higher than one suggested by the seat-of-the-pants sailor looking only at the water to choose a jib code before hoisting sails. Sails would go up, sailing heel would be compared to historical averages of target heel for that windspeed, and target boatspeed would be reduced by the trimmers and helmsman to reflect the lower wind weight day.
You might hear things onboard from the old salts like “the wind really isn’t filled down to the water today” and, soon after, “let’s start with target a knot under posted”. On boats with more elaborate systems and reliable historical data, you might hear “Heel WindWeight 80%” (where Heel WW = Heel/Target Heel reverse-looked-up as windspeed) and then “Heel WindWeight target eight tenths under posted”. With any luck, this would happen before coming off the start line and dribbling into the boat to leeward.
Here’s where it gets good: using live heel compared to historical heel creates a feedback loop. So, if we suspect a lower wind weight day and adjust our target speeds down just a little, our averages are shaded toward less actual heel from sailing slightly thinner on the telltales, which corroborates the lower wind weight. Without a boat to leeward of us creeping ahead with a better mode, we could easily talk ourselves with data into sailing way too high and slow. The same happens in reverse: we suspect strong wind weight so we sail pressed, the average heel numbers go up so we gain confidence in higher wind weight, etc.
Many keelboat classes have lots of practice using excellent data but, also, boat on boat relatives. Then, these feedback loops have balancing additional information coming from other boats to help find what really works and avoid spiraling toward poor performance which the data says is good.

Now, check this out! There is a secondary feedback loop for targets in the 1987-but-also-2017 system. If we decide to shade our sailing toward the narrow/high-mode side of targets, and before even changing course we trim our sails ever-so-slightly tighter, the mainsail with reduced twist bends the windfield narrower ahead of it and the windvane reports that we are sailing higher. So, the upwash on the tighter mainsail causes the “TRUE wind angle” to read (artificially!) slightly narrower just from the trim change. Similarly, if we shade our sailing toward the wider/fast mode side of targets, the more-twisted sail upwash causes the “TRUE wind angle” to read artificially slightly wider just from the trim change.
Splitting hairs, “artificially” narrow TWA causes the TWS to under-read, artificially wider TWA causes the TWS to over-read. So now we have data which says the tighter-sail-lower-heel-super-high-mode is working even better, because not only have we shifted the baseline of the TWS lookup slightly to show us performing well in “less wind”, we have also made the wind angle read over-narrow. When this “tighter-higher test” is compared with the “looser-wider test” (where the instruments over-sold the windspeed and over-widened the wind angle), it’s a double double smackdown for the high mode.
So you know how I said “God said no more shear”? He was kidding. Have a look at the above paragraph again, but imagine what starts to happen when we cause the input to the entire system (measured at one point) to have actual, real-world asymmetry. Now you’ve got a TWS which is part of a wind-triangle solution measured at one point but in a wind-field which is actually bent from the deck to the masthead.
So, that TWS reads as higher on the wider tack, and the faster targets derived from looking up a target boatspeed from that TWS is added to the mix. So you respond to those targets by sailing more pressed…and…tacking away. But now the shear is making the targets under-read, and you can’t understand why you were lower and not much faster on the other tack sailing target, but now you’re higher and much much slower and just got rolled. You tell yourself you’re doing this for fun, but….
Phew. We’ve gotten to the top mark debating the authority of the targets, theorizing about where the errors might come from when they don’t seem to work right against the other boats, and wondering whether maybe the label for the wind angle displayed on the mast shouldn’t be something more like: “Representation of angle of the sailplan to a 3-dimensional field, sampled in one place, itself influenced by said sailplan and also invisibly variable in time and across the course, which we boil down to one number and egregiously misname as ‘True’”.
Later, we can go downwind and try to figure out what to do with our target boatspeed and TWA numbers, when they assume use of the tiller and crew weight to create dynamic apparent wind speed and dynamic apparent wind angle behavior which keeps the downwind heel and kite trim matching those assumptions.
For now, a quick summary: it’s way more complicated than it looks from a stack of six displays on the mast, of which two say “TRUE” and another one screams “Target” like it owns the place. We are still bringing a spatula to a lightsaber fight. The reason the very top boats don’t notice how dated their instruments are, is that they have gone to great effort and expense to add multiple, autonomous interlaced neural nets boasting highly-evolved, experience-assimilating algorithms and richly nuanced communication protocols to their onboard systems.
We call those: sailors. Title inspiration thanks to XTC.
Jump in the thread.