Ramblings on Routers

29 11 2010

Laughing Sally posted a question on SOL’s forum recently asking how much of a difference use of software routing makes and whether using one is a pre-requisite to winning in online sail-races. I drafted a reply, getting partway into it before completely losing track of what I had started out trying to say. Editing wasn’t going to save the response, so I moved it here… It’s still long and rambling, but at least the page format doesn’t highlight that quite so badly.

I’ll take a stab at this one. Nicely posed question.

There are several sub-classes of routing software (and I suppose routing or potential-routing SOLers):

A handful of “super-routers” are using custom software. This gives two distinct advantages – they can take into account the way SOL works, and they understand what the software is doing from the inside out. “Can” doesn’t translate to “do”…

It used to be that the big (huge?) edge was the weather. It’s well known that SOL’s weather doesn’t match up perfectly with public sources and this means that although you can use commercial software for overall strategy you shouldn’t steer by it. Two changes have erased this – brainaid is making SOL’s “grib” available, and SOL has changed the way it sources and resamples the wind. Both are very welcome! Using brain’s weather (and polars for that matter), the gap to commercial software is pretty much erased. There are some very powerful free alternatives now too.

A “perfect” router will give you a perfect solution inside the 6h update interval. This means that races that can be broken down are easier to route with certainty. If there is a waypoint before the next update both the course and arrival time can be “known”. This is maybe closest to the “they just do what the computer tells them” complaint, though a fair number of boats that “could” win these quite handily either don’t bother with this sort of race, don’t route at all in these circumstances, or only use software minimally. I guess it needs to be pointed out that inside of 6 hours the good SOTP guys are very fast as well – it’s a much simpler problem. I can intuit a short leg awfully close to the perfect track, and if you want to work things out with more precision the ideas presented in my blog will squeeze out the last tenth of a percent.

If the forecast didn’t update, there would only be one solution for every race – many boats could make a pretty decent go sailing by feel, while someone with enough ambition could work it out fully by hand. The software guys could just push a button and set some DC. If all three methods were perfectly executed (this decides many short races) the results would be near-identical. I’m pretty sure that with *really* accurate timing you could tell what software someone was using… zero difference in real life.

The fun comes when the changing nature of the weather is taken into account. Every time the weather updates a new “perfect” solution is created, and the old one becomes sub-optimal. On a small shift, a boat that that was “behind” by the old weather can get promoted. On major changes whole races can be flipped upside-down. From this viewpoint, there is little reason to sail the supposedly perfect routed track – it’s guaranteed to be wrong. If you watch some of the best sailors go toe-to-toe (oh for replay) you can see a layer of tactics/strategy at work that makes for most of the game. It can be pretty subtle, but there is constant jockeying for position, gambling when there are decent odds, hedging against loss… It really never makes sense to sail the routed solution. You can’t pass a boat that way and you can’t hold a lead either. There’s a whole lot more racing going on than might appear.

Learning how weather systems behave can have a big impact on how you use routing software. I don’t think that any of the current routers have much in the way of statistical analysis built in – they give a single solution for weather that is assumed to be fixed. Pretty crude. If you can see that a weather event 5 days off is impacting a decision in the next hour or so you need to be pretty sure that it’s a stable/predictable  before you wager too much. A random windhole in the wrong place at the wrong time can disqualify a winning line through some islands for no reason. The software doesn’t know any better. The wind we use is a pretty impressive forecast, and betting *against* the weather rarely works, but there are lots of ways to add value to the routed solution. What are the odds if you sail a little higher or lower than the calculated track? Where does that put you relative to the fleet? Some routed tracks are very risky (though fastest at the time) but going slightly wide is often almost as fast but much less volatile. The unpredictability of the weather is somewhat predictable. Google “ensemble forecast” for some fun reading. How serious to take this?

Long ocean races are particularly interesting due to the lack of accurate long range forecasts. SOL weather only goes out 7 days. It’s far more important to understand what is likely to happen when you get to the trades than what the current model says. There’s a place in all this high-tech stuff for pilot charts.

Depending on the router, identifying options may be particularly difficult. If the router presents the fastest track but there is another wildly different one that is predicted to arrive just a little later you had better take a pretty close look. Even worse are solutions that alternate between two approaches on every update. Think of some of the races we’ve had with competitive splits – the fastest boats are the ones that set a strategy and then look to the router, not the other way around.

There are a few tricks to working with gribs that routing glosses over. Everyone knows that the NOAA data doesn’t match real life perfectly (or at all!) on a small scale, but there are clues “hidden” in the nice smooth weather that are relevant to offshore sailing strategy. The solution needs to be adjusted based on the navigator’s knowledge of how weather works and what is really likely to be there. Shear lines show up as impassable walls, unsettled areas interpret as jumbled low-wind areas where they likely have squalls. If the Navigator is avoiding a 400nm stretch of ocean because the software thinks there is a phantom obstacle there he’s not doing a very good job. The situation in SOL is different in that we sail the actual forecast and that impassable wall really is impassable. It’s somewhat predictable though, and knowing how SOL works can present options that the routing is blind to.

What running routing software does do very well is relieve some of the drudge work of calculating courses. The mechanics of HOW to get from A to B can be handed off, leaving the crucial decision of WHERE B needs to be to the sailor. The “game” that the best routers are playing is not “how to negotiate this shift” but how to set up for the best possible odds vs the fleet. “How to negotiate this shift” is an important lesson that SOL offers, but it’s not a tactical decision.

Routers are very useful for setting DCs, and knowing with some certainty where your boat will be in the morning. They also make snap decisions when you don’t have time for all of the above easy. It might not be the best track, but just doing what the voices tell you won’t lead you too far astray if you’re SOLing while juggling lunches and rushing people out the door in the morning.

Finally – routers are a fantastic way to learn more about sailing and navigation. The absolute biggest advantage that learning to use a router can give is better understanding, even if you never switch it on during an online race. If you like this sort of stuff, you can play for days with different scenarios, see how different boats handle the same weather, watch how the solution evolves over time. Some of the visualisations give you a whole different outlook on the game. Plotting isochrones shows all sorts of information – the weather, polar, obstacles and course are all handily summarized. Time contours plotted backwards from the finish let you assess fleet position. Combined, you can identify areas where potential gains might be made, where it’s safe to deviate from track to try an attack, evaluate risk on laylines… Some software allows you to simultaneously display the fastest track to every point on the course – this highlights the “seams” where there are multiple options, and those are critical areas on the map.

Do you need one to win – absolutely not.

How much difference do they make? – it’s a very useful tool. There’s nothing that the software does that can’t be done by hand, but you can solve in 30s what might take hours. Using the result effectively needs skill.

Disadvantage? – qualified yes? There is a reason that software routing has become a major part of offshore racing. That said, the difference between a well-routed and a good SOTP time is usually far less than the margins of victory while the strategic stuff really adds up quickly. The fleet has boats using identical software that finish very differently, and usually with dozens of places between.

Phew… back to the discussion on SOL.



13 responses

29 11 2010

This is an absolutely excellent post about what routers are and how they affect the game!

29 11 2010
Jeff Baumann

Thanks for an excellent review of this subject. I view routers as an advance in technology no different than the old sextant or today’s GPS. I have avoided using one, however, because some of my friends on SOL consider them cheating. My brother, for example, feels he wants to sail his own race, not “follow a computer.” Another friend of mine feels the same way. He feels it takes a lot of the fun out of a race. This topic has been debated not only on SOL but on VR, where we came from. Thanks Again, Jeff

30 11 2010

Hi 76T,
This is a fantastic post. So much to learn.
Thanks for posting.
Antonio (zero)

30 11 2010

Again, you find the right way (and the time) to explain in detail what we experience in SOL. My true compliments on the quality of your comments!

21 12 2010

Nice work. I’d be interested to know how many of the top 20 in the Christmas race used software, and how many were SOTP. Certainly 2 that I know of were SOTP….

I’m always of the opinion that we should work out for ourselves where we should be, and how we should get to where we are going. Understanding why we need to take a certain route and then applying it sucessfully is so much more satisfying then being told how to get there.

23 12 2010

Excellent exposé. Thank you for making your thoughts and knowledge available to us.

Because of this article I just realized an almost spiritual side aspect of SOL. One that I have overlooked so far – but one which is the reason I’m a fairly constant 15 – 20 at best:

It just occurred to me that I am sailing against the weather to maintain or improve position, whereas a mental attitude shift to sail with the weather is more likely to produce the desired results.

Now if I can just find this switch, I’ll flick it. The weather is friend – not foe.

7 01 2011

What a great explanation. Has inspired me to find out more. Thanks.

19 03 2012

Sailing a virtual boat that never breaks down or lack performance, using a router? May the best software win…

30 07 2012

La mejor explicación que he leido sobre el enrutamiento. Felicidades AGAGE. Te recordamos

14 03 2013
javakeda Bob

Outstanding post on the subject! Would like to continue the discussion to include:
— Specific router pitfalls [I have some examples to discuss]
— Where routers really help [I have examples here as well]
— “Objective Function” issues
— Applying router results, both virtually and IRL.
Would you like me to continue in these comments, or post in the SOL forum?

15 03 2013

Wow – Hi! Sure – SOL forum allows others to join in. I’m not currently adding new info on 76T/wordpressetc but would certainly follow up there/here as well. Shift to SOL and I’ll blog/copy the relevant bits?

The NY->SF is a great opportunity to explore how routing really relates…

22 03 2013
javakeda Bob

Given that I tend to write many words, I am thinking I should post in my own blog [javakeda.blogspot.com] lest I be perceived as either hi-jacking this thread or part of the SOL forum. Can you see that blog with the current ‘share’ settings?

12 04 2013

хи хи

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: