# Surfs up!

Hello,

I have been thinking about something recently and thought I'd get back on the list (after 10 years).

The situation I have been thinking about is what variables are being controlled when surfing, or sailing.

To simplify the issue think of a small boat with a very basic sail that could be raised or lowered. The goal of the sailor is go forward, and so she raises the sail when the wind is from behind and lowers it when the wind is from the front, and so, respectively moves forward or stays still. In PCT terms the variable being controlled is forwards movement and the output is to raise or lower the sail. The wind, then, is a disturbance to the perception of forward movement. Is that the correct way to think about this system, in PCT terms?

If so, then the disturbance is being used as a means of control, and the goals can't be achieved without it. Is that right, and are there other examples control where a disturbance is necessary?

···

--

Regards,
Rupert Young

[From Erling Jorgensen (2012.04.09 1010 EDT)]

Rupert Young (Mon, 9 Apr 2012 13:53:05 +0100)

In PCT terms the variable being controlled is forwards movement
and the output is to raise or lower the sail. The wind, then, is a
disturbance to the perception of forward movement. Is that the correct

Hello Rupert,
I suspect it would be useful to think in terms of degrees of freedom.
Wind has degree of intensity, as well as direction. That allows
partitioning those two, when it comes to analyzing things from a PCT
perspective.

from the little sailing I have done, it seems to me that the degree of
intensity of the wind is part of the Environmental Feedback Function
in the standard PCT control loop. As such it is part of the means of
attainment of getting the sailboat to move. The direction of the wind,
however, can sometimes be a Disturbance to the controlled direction of
sailing.

That is why, when the wind direction changes, the customary response is
not simply to lower the sail, because that removes the biggest component
of the EFF, & the sailboat remains dead in the water (or indeed, starts
drifting backwards, because of course the wind hits not just the sail,
but the side of the boat as well.) Rather, when the direction of the
wind itself becomes a Disturbance, as you undoubtedly know, it can be
compensated for by other means at the sailor's disposal, such as the
rudder.

Leaving the EFF in play by leaving the sail aloft allows some of the
wind's intensity to still be utilized to move the boat forward, even
though a good deal of the wind's intensity spills out of the sail. The
art & fun of sailing is finding that optimal turning of the rudder, so
that one can still maintain one's heading _despite_ what the wind is
doing.

In the process, the concept of "forward" gets redefined, as a tacking
back & forth that "on average" gets you where you want to go.

If so, then the disturbance is being used as a means of control, and the
goals can't be achieved without it. Is that right, ...

So technically, in the basic system that you set up, the wind fills two
roles in a basic PCT analysis, according to these two degrees of freedom --
that of a Disturbance to the direction of sailing, & that of an
Environmental Feedback Function as a means of control itself.

However, at a higher level, that engagement with the disturbing qualities
of the wind is part of the whole point of going sailing! That's where
the fun is. People usually don't wait for there to be no wind, so they
can row their sailboat around. As a first guess, I'd call it controlling
for some System Concept, composed of multiple level inputs, from the
Sensations & Transitions of the wind & the water, to the mechanics of
sailing itself (various Program decision nodes), to how daring you are &amp;
on how much of an edge you want to carve through the water (possible
Principle perceptions), to perhaps some kind of Narrative of being out on
the water like one's Irish(?) (Nordic in my case) ancestors. Yes, it
is all perception, & we control on all those levels at once.

All the best,
Erling

Well as a srfer controlling fear was always a biggie

···

On 4/9/12, Erling Jorgensen <ejorgensen@riverbendcmhc.org> wrote:

[From Erling Jorgensen (2012.04.09 1010 EDT)]

Rupert Young (Mon, 9 Apr 2012 13:53:05 +0100)

In PCT terms the variable being controlled is forwards movement
and the output is to raise or lower the sail. The wind, then, is a
disturbance to the perception of forward movement. Is that the correct

Hello Rupert,
I suspect it would be useful to think in terms of degrees of freedom.
Wind has degree of intensity, as well as direction. That allows
partitioning those two, when it comes to analyzing things from a PCT
perspective.

From the little sailing I have done, it seems to me that the degree of
intensity of the wind is part of the Environmental Feedback Function
in the standard PCT control loop. As such it is part of the means of
attainment of getting the sailboat to move. The direction of the wind,
however, can sometimes be a Disturbance to the controlled direction of
sailing.

That is why, when the wind direction changes, the customary response is
not simply to lower the sail, because that removes the biggest component
of the EFF, & the sailboat remains dead in the water (or indeed, starts
drifting backwards, because of course the wind hits not just the sail,
but the side of the boat as well.) Rather, when the direction of the
wind itself becomes a Disturbance, as you undoubtedly know, it can be
compensated for by other means at the sailor's disposal, such as the
rudder.

Leaving the EFF in play by leaving the sail aloft allows some of the
wind's intensity to still be utilized to move the boat forward, even
though a good deal of the wind's intensity spills out of the sail. The
art & fun of sailing is finding that optimal turning of the rudder, so
that one can still maintain one's heading _despite_ what the wind is
doing.

In the process, the concept of "forward" gets redefined, as a tacking
back & forth that "on average" gets you where you want to go.

If so, then the disturbance is being used as a means of control, and the
goals can't be achieved without it. Is that right, ...

So technically, in the basic system that you set up, the wind fills two
roles in a basic PCT analysis, according to these two degrees of freedom --
that of a Disturbance to the direction of sailing, & that of an
Environmental Feedback Function as a means of control itself.

However, at a higher level, that engagement with the disturbing qualities
of the wind is part of the whole point of going sailing! That's where
the fun is. People usually don't wait for there to be no wind, so they
can row their sailboat around. As a first guess, I'd call it controlling
for some System Concept, composed of multiple level inputs, from the
Sensations & Transitions of the wind & the water, to the mechanics of
sailing itself (various Program decision nodes), to how daring you are &amp;
on how much of an edge you want to carve through the water (possible
Principle perceptions), to perhaps some kind of Narrative of being out on
the water like one's Irish(?) (Nordic in my case) ancestors. Yes, it
is all perception, & we control on all those levels at once.

All the best,
Erling

Erling,

Thanks for your response. I purposefully wanted to avoid talking about wind intensity (or fun) so as not to cloud the issue.

A disturbance is, I would say, anything in the EFF that cannot be controlled. In this case the wind direction. It seems odd that a disturbance can also have a different role, perhaps you could expand on that. Though I guess I've always thought of a disturbance as a hindrance to control rather than an aid.

Do you have any other examples of control where this is the case?

Regards,
Rupert

···

On 09/04/2012 16:00, Erling Jorgensen wrote:

[From Erling Jorgensen (2012.04.09 1010 EDT)]

Rupert Young (Mon, 9 Apr 2012 13:53:05 +0100)
In PCT terms the variable being controlled is forwards movement
and the output is to raise or lower the sail. The wind, then, is a
disturbance to the perception of forward movement. Is that the correct

Hello Rupert,
I suspect it would be useful to think in terms of degrees of freedom.
Wind has degree of intensity, as well as direction. That allows
partitioning those two, when it comes to analyzing things from a PCT
perspective.

From the little sailing I have done, it seems to me that the degree of
intensity of the wind is part of the Environmental Feedback Function
in the standard PCT control loop. As such it is part of the means of
attainment of getting the sailboat to move. The direction of the wind,
however, can sometimes be a Disturbance to the controlled direction of
sailing.

That is why, when the wind direction changes, the customary response is
not simply to lower the sail, because that removes the biggest component
of the EFF,& the sailboat remains dead in the water (or indeed, starts
drifting backwards, because of course the wind hits not just the sail,
but the side of the boat as well.) Rather, when the direction of the
wind itself becomes a Disturbance, as you undoubtedly know, it can be
compensated for by other means at the sailor's disposal, such as the
rudder.

Leaving the EFF in play by leaving the sail aloft allows some of the
wind's intensity to still be utilized to move the boat forward, even
though a good deal of the wind's intensity spills out of the sail. The
art& fun of sailing is finding that optimal turning of the rudder, so
that one can still maintain one's heading _despite_ what the wind is
doing.

In the process, the concept of "forward" gets redefined, as a tacking
back& forth that "on average" gets you where you want to go.

If so, then the disturbance is being used as a means of control, and the
goals can't be achieved without it. Is that right, ...

So technically, in the basic system that you set up, the wind fills two
roles in a basic PCT analysis, according to these two degrees of freedom --
that of a Disturbance to the direction of sailing,& that of an
Environmental Feedback Function as a means of control itself.

However, at a higher level, that engagement with the disturbing qualities
of the wind is part of the whole point of going sailing! That's where
the fun is. People usually don't wait for there to be no wind, so they
can row their sailboat around. As a first guess, I'd call it controlling
for some System Concept, composed of multiple level inputs, from the
Sensations& Transitions of the wind& the water, to the mechanics of
sailing itself (various Program decision nodes), to how daring you are&amp;
on how much of an edge you want to carve through the water (possible
Principle perceptions), to perhaps some kind of Narrative of being out on
the water like one's Irish(?) (Nordic in my case) ancestors. Yes, it
is all perception,& we control on all those levels at once.

All the best,
Erling

--

Regards,
Rupert Young
Mobile: +447795 480387
Moon's Information Technology Limited

I don't mean to sond like a devils advocate, bt cold it be that self
preservation instinct (fear) is a distrbance on the perception of a
surfer. Controlling for that fear by developing an appreciation for
it, cold explain the thrill seeking enjoyment. And perhaps this is why
the big wave srfers often develop a thrill seeker complex-pshing the
limits of their fear and threading the thin line between self
preservation and controlling the body in harmony with a sblime natral
force.

···

On 4/9/12, Rupert Young <rupert@moonsit.co.uk> wrote:

Erling,

wind intensity (or fun) so as not to cloud the issue.

A disturbance is, I would say, anything in the EFF that cannot be
controlled. In this case the wind direction. It seems odd that a
disturbance can also have a different role, perhaps you could expand on
that. Though I guess I've always thought of a disturbance as a hindrance
to control rather than an aid.

Do you have any other examples of control where this is the case?

Regards,
Rupert

On 09/04/2012 16:00, Erling Jorgensen wrote:

[From Erling Jorgensen (2012.04.09 1010 EDT)]

Rupert Young (Mon, 9 Apr 2012 13:53:05 +0100)
In PCT terms the variable being controlled is forwards movement
and the output is to raise or lower the sail. The wind, then, is a
disturbance to the perception of forward movement. Is that the correct

Hello Rupert,
I suspect it would be useful to think in terms of degrees of freedom.
Wind has degree of intensity, as well as direction. That allows
partitioning those two, when it comes to analyzing things from a PCT
perspective.

From the little sailing I have done, it seems to me that the degree of
intensity of the wind is part of the Environmental Feedback Function
in the standard PCT control loop. As such it is part of the means of
attainment of getting the sailboat to move. The direction of the wind,
however, can sometimes be a Disturbance to the controlled direction of
sailing.

That is why, when the wind direction changes, the customary response is
not simply to lower the sail, because that removes the biggest component
of the EFF,& the sailboat remains dead in the water (or indeed, starts
drifting backwards, because of course the wind hits not just the sail,
but the side of the boat as well.) Rather, when the direction of the
wind itself becomes a Disturbance, as you undoubtedly know, it can be
compensated for by other means at the sailor's disposal, such as the
rudder.

Leaving the EFF in play by leaving the sail aloft allows some of the
wind's intensity to still be utilized to move the boat forward, even
though a good deal of the wind's intensity spills out of the sail. The
art& fun of sailing is finding that optimal turning of the rudder, so
that one can still maintain one's heading _despite_ what the wind is
doing.

In the process, the concept of "forward" gets redefined, as a tacking
back& forth that "on average" gets you where you want to go.

If so, then the disturbance is being used as a means of control, and the
goals can't be achieved without it. Is that right, ...

So technically, in the basic system that you set up, the wind fills two
roles in a basic PCT analysis, according to these two degrees of freedom
--
that of a Disturbance to the direction of sailing,& that of an
Environmental Feedback Function as a means of control itself.

However, at a higher level, that engagement with the disturbing qualities
of the wind is part of the whole point of going sailing! That's where
the fun is. People usually don't wait for there to be no wind, so they
can row their sailboat around. As a first guess, I'd call it controlling
for some System Concept, composed of multiple level inputs, from the
Sensations& Transitions of the wind& the water, to the mechanics of
sailing itself (various Program decision nodes), to how daring you
are&amp;
on how much of an edge you want to carve through the water (possible
Principle perceptions), to perhaps some kind of Narrative of being out on
the water like one's Irish(?) (Nordic in my case) ancestors. Yes, it
is all perception,& we control on all those levels at once.

All the best,
Erling

--

Regards,
Rupert Young
Mobile: +447795 480387
Moon's Information Technology Limited

[Martin Taylor 2012.04.09.17.31]

Hello,

I have been thinking about something recently and thought I'd get back on the list (after 10 years).

The situation I have been thinking about is what variables are being controlled when surfing, or sailing.

To simplify the issue think of a small boat with a very basic sail that could be raised or lowered. The goal of the sailor is go forward, and so she raises the sail when the wind is from behind and lowers it when the wind is from the front, and so, respectively moves forward or stays still. In PCT terms the variable being controlled is forwards movement and the output is to raise or lower the sail. The wind, then, is a disturbance to the perception of forward movement. Is that the correct way to think about this system, in PCT terms?

If so, then the disturbance is being used as a means of control, and the goals can't be achieved without it. Is that right, and are there other examples control where a disturbance is necessary?

Rupert, for the convenience of later reference, it is conventional on this list to precede your text by an "Author plus date-time" message ID, as I have done here. Some people say "From" before their name, but that's a matter of personal choice.

I see it a little differently. You can't tell what is a disturbance unless you know the controlled perception, and even then, the actual disturbance is only the effect on that perception, whatever the cause of the disturbance. What perception are you trying to control? Is it perceived wind direction relative to you or is it the direction of motion of your boat? It can't be both, because if the wind shifts geo-centred direction, say from North to Northeast, you will feel it as coming from a different direction unless you turn the boat the equivalent amount.

Usually, one would be controlling a perception of the boat's movement. That is actually two controlled perceptions, one being the boat's heading, the other being the component of the boat's velocity in the reference direction. Wind provides what I call an "environmental affordance" for each. The wind speed and direction determines the range of speeds the boat can go on any momentary heading. By tacking, the sailor can effectively add vectors for speed in two directions, so that averaged over sufficient time the boat will progress in the reference direction, though possibly not as fast as the sailor desires.

What actually will the sailor perceive at any given moment? And what actions are available to alter those perceptions that feed into the main controlled perceptions -- average heading and velocity in the reference direction? You say that the action available is to raise and lower the sail, but that is only one degree of freedom. You need another, and two rather obvious ones spring to mind, namely to alter the angle of the sail relative to the boat heading and to alter the angle of the rudder (I guess a wind-surfer doesn't have the latter :-).

As for the available perceptions, the wind velocity and relative direction are quite often observable, at least to a first approximation, but suppose they are not, and the only available perceptions are the controlled ones, boat heading and velocity in the reference direction. In other words, imagine that the situation is as is usually considered in a PCT analysis. The value of the perception is observable, the cause of any disturbance to that perception is not.

Suppose that at some moment, the sailor has the boat going in the reference direction at the reference velocity. Now the wind shifts direction. What happens to the controlled perceptions? Presumably the boat heading and/or velocity in the reference direction change. Those are two disturbances, one to each controlled perception. What action does the sailor take? Whatever combination of sail and rudder actions return the boat to the desired heading and velocity, a combination that originally would have been determined by reorganization, but that in a practiced sailor is already organized fairly effectively.

The problem now is that not all combinations of wind speed and direction are compatible with any specific reference direction and velocity pair. Some wind shifts cause insuperable disturbances. With that wind, the boat cannot physically move in that direction at that velocity. But it is always, with modern sail patterns, possible to move the boat with in a direction that has a vector component in the reference direction.

We are assuming that the sailor can't perceive the actual wind, but the perceptions she is controlling are getting further from their reference values no matter what actions she takes. More reorganization! But what reorganization can solve this problem if the wind conditions make it simply impossible to head in the desired direction at the desired velocity? We call the end-point of this new reorganization "tacking", but perhaps we should look a little more closely at what "tacking" involves.

When on either tack, the sailor controls a perception of the heading with a reference value that is different from the reference for the desired forward motion of the boat. This might be thought to indicate a conflict between the two controlled direction perceptions, whereas the sailor actually sees the "false" heading as contributing to motion in the reference direction, the error introduced by one tack being corrected (and over-corrected) on the other. If the perception of forward motion is averaged over long enough times, the error will be close to zero.

In hierarchic perceptual control, control is necessarily fastest at levels closest to the physical action. If a higher-level control system attempts to control faster than a lower-level supporting control system can function, the whole complex is likely to go into oscillation. For the sailor, when the gross motion of the boat is instantiated by changes in the momentary heading, control of the gross motion has to be slow compared to the changes in heading. The environment, in the form of the wind, actually determines how these two perceptions interact, but as we posited, the sailor does not know the wind direction. All the sailor needs is to be able to perceive the boat's heading and separately its overall motion direction and velocity component in the reference direction. Normal error correction for the three controlled perceptions will generate a suitable tacking sequence, if tacking is necessary. Too long on one tack, and the perception of the overall motion direction will drift toward the direction of that tack, resulting in a shift in the reference heading direction onto the other tack.

To return to the question that started this essay, the wind is not itself a disturbance to anything. Variations in the wind alter the relationships among the output variables of sail quantity, sail orientation, and rudder orientation, and the ability of the boat to move in different directions at different speeds. This means that if these three action quantities remain static when the wind changes, the boat's motion direction and velocity will change. Those changes are the disturbances to the controlled perceptions, and those two changes are what alterations in the output variables (sails and rudder) will counter.

Of course, in the real world, the sailor will be able to perceive momentary changes in the wind, and will be able to control perceived relationships that involve the wind, the sails, and the rudder. But I think that was not the sense of your question, which was about how to conceive the effects of the wind as disturbances to controlled perceptions. I hope this screed goes a little way toward helping to clarify the issue.

Martin

···

On 2012/04/9 8:53 AM, Rupert Young wrote:

[From Erling Jorgensen (2012.04.09 2030 EDT)
(I drafted this last night, but then a glitch kept me from posting it.)

Rupert Young (Mon, 9 Apr 2012 18:29:45 +0100)

Different thoughts occur to me, in response to what you raise. I have
a small disagreement about how you phrase things, but I'll wait with
that because I realize that is not the main part of what you are asking

...I've always thought of a disturbance as a hindrance
to control rather than an aid.

Technically, a disturbance can work in either direction, to help or to
hinder. I remember back in the early 1990s, Michelle Duggins (one
of Tom Bourbon's students) gave a presentation at the CSG Annual
Conference. She was trying to model "helping behavior", using
various configurations of interacting tracking simulation tasks.

If I'm remembering the set-up correctly, she routed the output function
of the so-called "Helper's" mouse movements so that they would
independently move the "Helpee's" cursor position, and thereby
cancel some of the independently occurring disturbance to the target
the Helpee was trying to follow. Even though this was supposed to be
helping them to control, as I recall she said some of the test subjects
were upset, because their cursor was moving differently from how
their own output was seeking to have it move. In this case, it seems,
"helping" wasn't much help.

Getting back to your own example of sailing, there are times when
wind direction is mostly a help, and other times when it is more of a
hinderance, to the direction the sailor wants to go. What makes it a
"disturbance" in a technical sense is that it arises independently of the
actions of the controller, and that it affects something that matters to
the controller. Usually it doesn't exactly coincide with the preferred
state of the controller's perception, and so it is counteracted as
needed. As I said before, one can still maintain one's heading
_despite_ what the wind is doing, whether helping or hindering.

Do you have any other examples of control where this is the case?

What about sunshine? We use the extra light of daytime sunshine
to move better and do all sorts of things that would be much harder
at night. And yet, depending on our direction of movement, it can
also be somewhat blindingly in our eyes, and thus a hindering
disturbance.

It seems odd that a
disturbance can also have a different role, perhaps you could expand on
that.

This was a comment, I believe, on my distinction that one aspect of the
wind (i.e., its intensity) could serve as the means of attaining control,
and thereby be part of the Environmental Feedback Function as such, on a
classical diagram of a control loop, even as another aspect of the wind
(i.e., its direction) could serve as a Disturbance on that same diagram.

I think it's helpful to remember that these arejust analytical distinctions
we are making. The wind doesn't have to switch positions, in order to
appear in two different spots on the diagram. It is how it functions that
determines how we diagram it. If we are leveraging its energy, we add it
to the overall gain of the control loop, at the point marked EFF, as part
of the environmental dynamics _through which_ we achieve control. If it is
getting in the way of something that matters to us, we diagram it as coming
from outside the loop proper, and add its effects to whatever else is
affecting the so-called controlled variable. The "whatever else" there
is customarilly the counteracting effects of our own behavioral output,
so that control is achieved _in spite of_ the action of the disturbance.

A disturbance is, I would say, anything in the EFF that cannot be
controlled. In this case the wind direction.

Here is the small quibbling about your choice of words. I do find it
useful to keep those two functional designations distinct. If something
is "in the EFF," it is a multiplier in the equations of the loop, part of
the overall gain term, and multiplied by the behavioral output to achieve
its effect.

By contrast, if something is "a disturbance," it is an additive term in
the equations of the loop, added with a sign opposite to the behavioral
output (which ordinarily serves to largely cancel out the impact of the
disturbance term.)

These are conceptual distinctions, which make a difference in how
the equations are set up. That is why I invoked the notion of
"degrees of freedom" to have the same outside force (in this case, the
wind) function in two very different manners at two distinct points in
the control loop diagram.

Moreover, I wouldn't speak of a disturbance per se as something that
[can or] "cannot be controlled." Remember, it is the _effects_ of a
disturbance that we may or may not try to control. I find it a helpful
discipline to keep inserting that word, because it reminds me that,
despite the various things that we do, it is perceptions (or the effects
of what we do) that we care about.

All the best,
Erling

[From Rick Marken (2012.04.10.1255)]

The situation I have been thinking about is what variables are being
controlled when surfing, or sailing.

Thanks to Rupert for suggesting my next project to me. I'm moving from
studying how people catch baseballs to how they sail to Catalina
Island. I actually started some field research on it last Sunday; it
turns out a friend of ours has a very nice 43 foot sailboat so we took
it out for a test drive. The wind was a very helpful disturbance;-)

Best

Rick

···

On 2012/04/9 8:53 AM, Rupert Young wrote:
--
Richard S. Marken PhD
rsmarken@gmail.com

[Rupert Young 2012.04.10.19.00]

Responding to both Martin and Erling.

I'd like to avoid getting into tacking or other degrees of freedom, as this is not actually about sailing. What I am interested in is when control is entirely dependant upon an environmental "function"; in this case the wind. For modelling purposes, and my simple brain, I'd like to keep things simple so I'll restate the situation, though let me know if it's too simple.

Think of a small boat with a very basic sail that could be raised or lowered. The goal of the sailor is go forward, and so she raises the sail when the wind is from behind and lowers it when the wind is from the front, and so, respectively moves forward or stays still. The sail then can be in one of two states, 1 (up) or 0, the wind, -1, 0, or +1 (from behind) and the movement of the boat -1, 0, or +1 (forwards). The overall reference, then is +1, for the boat movement.

It would seem that it is necessary for the system (the sailor) to perceive the state of the wind (though not control it) in order to "decide" what to do; that is, if the wind is from behind (+1) raise the sail. So considering comments made is it better to think of controlling the relationship between the perceived state of the boat movement and the perceived state of the wind? In this way the system would use the wind to propel the boat forwards.

So, is the wind an additional perceptual input to the system? If control of the relationship between wind and movement (so both equal +1) how would this be modelled? I'll have a go at a spreadsheet model. Ideas welcome.

Regards,
Rupert

[From Bill Powers (2012.04.11.0946 MDT)]

Hi, Rupert (and welcome back!) --

[Rupert Young 2012.04.10.19.00]

RY: Think of a small boat with a very basic sail that could be raised or lowered. The goal of the sailor is go forward, and so she raises the sail when the wind is from behind and lowers it when the wind is from the front, and so, respectively moves forward or stays still. The sail then can be in one of two states, 1 (up) or 0, the wind, -1, 0, or +1 (from behind) and the movement of the boat -1, 0, or +1 (forwards). The overall reference, then is +1, for the boat movement.

It would seem that it is necessary for the system (the sailor) to perceive the state of the wind (though not control it) in order to "decide" what to do; that is, if the wind is from behind (+1) raise the sail. So considering comments made is it better to think of controlling the relationship between the perceived state of the boat movement and the perceived state of the wind? In this way the system would use the wind to propel the boat forwards.

BP: That is the logic behind the compute (or predict) -and-execute model. It says that if you want to achieve a goal, you have to analze the variables that are involved and compute the action that would be needed to achieve the goal, then execute the action.

Of course that would work, though it would require some accurate analyses of the environment and a lot of computing power, and still might not work very well if there are unexpected disturbances or other changes. A rather advanced level of intelligence is needed to make it work.

But there is a much simpler way for the sailor to operate which doesn't require knowing anything about sailing at all, except that raising the sail can make the boat move and lowering it can stop the boat. He doesn't even have to know that it's the wind that makes the boat move.

If the sailor wants to go forward, he raises the sail. If the boat moves the wrong way, without even a component in the desired direction, the sailor takes the sail down again. If the boat doesn't move, the sailor leaves the sail up until the boat starts moving. If it starts moving in a good direction he leaves the sail up. And that will eventually get him where he wants to go. It may take years, but he will get there unless he starves or dies of old age first. He does need to know what a good direction is.

If the sailor can perceive the difference between a favorable wind and an unfavorable one, he may realize that there is a logical connection between the kind of wind and the direction the boat moves when the wind blows. If the relative wind is blowing from any direction other than from straight ahead plus or minus 30 degrees or so, the wind is favorable. The controlled variable is a logical variable: Unfavorable wind XOR sail up. XOR is "exclusive OR." A XOR B means (A and not B) OR (B and not A)

This logical perception is true if the wind is unfavorable and the sail is not up, OR if the wind is not unfavorable and the sail is up. The reference signal is set to TRUE, so the state of the sail would be adjusted to keep the perception of this logical variable TRUE. If the logical state is false, take the sail down if it is up or raise it if it is down.

If there is no wind, that could be perceived either as favorable or unfavorable. If favorable, the sail could be left up with no harm done. If seen as unfavorable, an obsessive-compulsive captain would insist on lowering the sail. Eventually he would reorganize because lowering it is a waste of effort.

RY: So, is the wind an additional perceptual input to the system? If control of the relationship between wind and movement (so both equal +1) how would this be modelled? I'll have a go at a spreadsheet model. Ideas welcome.

BP: My approach, as you can see, is to assume total ignorance from the start and rely on random changes in perceptions (and a propensity for controlling to achieve good outcomes) to produce more and more effective control. While the objective speed and direction of the wind are not affected by the sailor's actions, the relative direction is affected by the rudder and by the way the boat is trimmed to react to crosswind components. So turning the boat can be used as a way to turn an unfavorable wind into a favorable one. However, it is not necessary to perceive the wind, because turning the boat affects its forward speed, and if turning it toward the destination leaves enough forward speed, that is sufficient to get there for all wind directions but those coming from nearly dead ahead.

Learning about tacking would be a more advanced kind of sailing -- over some range of objective wind direction, you get to the destination faster by falling off alternately to left or right to get more forward speed. I think that in principle the optimum strategy could be evolved still without ever perceiving the exact role of wind speed and direction. Put the destination some distance left of the destination direction for a while, then to the right of it, and so on, and you will get there. As you repeat this experience over and over, you learn how much tacking to do. You try using the rudder, then try raising and lowering the sail, and keep reorganizing until the time to the destination is made as short as possible.

Intelligent systematic control is always more efficient than random reorganization, but the latter is needed to bring the former into existence.

Come to think of it, you could get to the destination just by using the E. coli strategy. If the destination is getting closer, do nothing. If it starts getting farther away, start randomly changing things like rudder angle and sail position, and keep changing them until the destination is getting closer again. Eventually, you will get to the destination. If you live that long and don't collide with Africa or an iceberg.

Best,

Bill P.

···

At 05:36 AM 4/11/2012, Rupert Young wrote:

[From Erling Jorgesen (2012.04.11 1100 EDT) - i.e., when started, amidst
interruptions]

Rupert Young 2012.04.10.19.00

What I am interested in is when
control is entirely dependant upon an environmental "function"; in this
case the wind.

This reminds me that all control loops (except for what has been designated
"the imagination connection") are closed through the environment. Items
that are 'out' in that environment, diagrammed in a standard control loop,
include the physical dynamics which we term the Environmental Feedback
Function (EFF for short) through which the controller's behavioral output
is made effective, as well as any Disturbance to the perception that is
being monitored.

For modelling purposes, and my simple brain, I'd like to
keep things simple ...

Sounds good. I always like returning to first principles, so I can notice
hidden un-accounted-for assumptions (like, "then a miracle happens") that
are creeping into a given analysis.

The sail
then can be in one of two states, 1 (up) or 0, the wind, -1, 0, or +1
(from behind) and the movement of the boat -1, 0, or +1 (forwards). The
overall reference, then is +1, for the boat movement.

So there is only one degree of freedom for the sailor's behavioral output.
And by what you have said, the reference for movement is always kept at
+1 (forward), and anything other than that -- i.e., no boat movement or
backward boat movement -- would generate error that the sailor would
attempt to correct.

Even the simplified environmental aspect of the wind in your proposed
model, still has two degrees of freedom: intensity, 0 or 1, and direction,
- or +. Since it is the only means of propulsion in this simulation,
it forms part of the EFF and must be multiplied by the state of the sail,
which itself is either 1 (up) or 0 (down). So the six combinations of Sail
x Wind lead to 0 on four occasions (whenever the sail is down, or when the
wind is motionless even if the sail is up), -1 on one occasion (when the
sail is up, but the wind is from the front), and +1 on only one occasion
(when the sail is up, and the wind is from behind.)

So I'm making here the behavioral Output = Sail x Wind.

The sign of the wind, which indicates its direction, enters into the
equations as a Disturbance (or not) to forward motion of the boat. The
only way to go forward is if the wind is from behind, but that outcome is
conditional on the state of the sail. All other conditions lead to, at
best, remaining still in the water, which of course is still a disturbance
to preferring to go forward.

Because you are using discrete variables that are either 1 or 0, (in other
words a given variable cannot be more "on" than 1), don't you have to use a
Boolean operator to compute the perception as Output AND Wind, where you
only get forward motion if both Output and Wind are +1? That would seem to
be the way to enter the Disturbance portion into perceived Motion. I am
not well versed in this realm.

What bothers me is that there does not seem to be a way to counteract the
wind when it is from the wrong direction, other than to lower the sail
and wait it out. So I think you would need some higher system that can
generate the reference "Wait" if the wind conditions are not right.

It would seem that it is necessary for the system (the sailor) to
perceive the state of the wind (though not control it) in order to
"decide" what to do; that is, if the wind is from behind (+1) raise the
sail.

Ordinarily, as I think Martin said, it is not necessary to perceive a
disturbance directly or its origin, in order to control the variable that
is affected by the disturbance. Just monitor the perception, and if it
is not how you want it to be, "do something," and adjust the sign (+/-) of
what you do according to whether the outcome is getting worse or better.

I think because you only have a single degree of freedom in the output
(i.e., raise the sail or not), you would need to get another degree of
freedom into the perceptual stream. That would be a way of constructing
a higher level of perception than just perceiving the motion of the boat.
Otherwise, an error is still generated when the sail is up but the wind
conditions are not right, yet there would be no effective action that
could be taken.

As you indicate, introducing a perception of the state of the wind allows
introducing it as a condition in a higher level of if-then logic, which
could vary the reference as to boat motion, with its behavioral output of
lowering vs. raising the sail. You then have two possible reference states
that could achieve zero error, in essence, "Go forward" or (at a higher
level?) "Wait, before going forward", respectively.

So, is the wind an additional perceptual input to the system? If control
of the relationship between wind and movement (so both equal +1) how
would this be modelled?

Under the conditions you set up, it seems that perceiving the wind might
be needed. But I believe the key would be the relationship between Sail
and Wind, rather than wind & movement.

I'll have a go at a spreadsheet model. Ideas welcome.

See what you come up with. Hope these ideas have been helpful.

All the best,
Erling

[Rupert Young 2012.04.11.2300 BST]

> [From Bill Powers (2012.04.11.0946 MDT)]

BP: If the sailor wants to go forward, he raises the sail. If the boat
moves the wrong way, without even a component in the desired direction,
the sailor takes the sail down again. If the boat doesn't move, the
sailor leaves the sail up until the boat starts moving. If it starts
moving in a good direction he leaves the sail up. And that will
eventually get him where he wants to go. It may take years, but he will
get there unless he starves or dies of old age first. He does need to
know what a good direction is.

RY: Something like that did occur to me, though it seems that the
ability to perceive the wind enables greater efficiency; cutting out the
failed attempts.

> BP: If the sailor can perceive the difference between a favorable
wind and an unfavorable one, he may realize that there is a logical
connection between the kind of wind and the direction the boat moves
when the wind blows. If the relative wind is blowing from any direction
other than from straight ahead plus or minus 30 degrees or so, the wind
is favorable. The controlled variable is a logical variable: Unfavorable
wind XOR sail up. XOR is "exclusive OR." A XOR B means (A and not B) OR
(B and not A)

Ok, this sounds like an input with two perceptions; one for the wind and
one for the state of the sail. Is that right? What makes a wind
favourable or otherwise? Also there is no perception of the movement of
the boat, should there be?

I've attached an attempt at modelling this (xls and xlsx versions). I
may have things arse over tit, but it works very nicely. You can change
things for the boat to go backwards or forwards, and the sail goes up or
down accordingly. See description.

Basically the output of control system B indicates whether the wind is
favourable or not, and C changes the state of the sail accordingly. I am
concerned though that there is no control system representing the boat
movement.

Please let me know if this model is valid in PCT terms.

Regards,
Rupert

sailing v1.2.xls (30.5 KB)

sailing v1.2.xlsx (11.4 KB)

···

On 11/04/2012 18:05, William Powers wrote:

[Rupert Young 2012.04.11.2315]

[From Erling Jorgesen (2012.04.11 1100 EDT) - i.e., when started, amidst
interruptions]

So there is only one degree of freedom for the sailor's behavioral output.
And by what you have said, the reference for movement is always kept at
+1 (forward), and anything other than that -- i.e., no boat movement or
backward boat movement -- would generate error that the sailor would
attempt to correct.

Even the simplified environmental aspect of the wind in your proposed
model, still has two degrees of freedom: intensity, 0 or 1, and direction,
- or +. Since it is the only means of propulsion in this simulation,
it forms part of the EFF and must be multiplied by the state of the sail,
which itself is either 1 (up) or 0 (down). So the six combinations of Sail
x Wind lead to 0 on four occasions (whenever the sail is down, or when the
wind is motionless even if the sail is up), -1 on one occasion (when the
sail is up, but the wind is from the front), and +1 on only one occasion
(when the sail is up, and the wind is from behind.)

So I'm making here the behavioral Output = Sail x Wind.

The sign of the wind, which indicates its direction, enters into the
equations as a Disturbance (or not) to forward motion of the boat. The
only way to go forward is if the wind is from behind, but that outcome is
conditional on the state of the sail. All other conditions lead to, at
best, remaining still in the water, which of course is still a disturbance
to preferring to go forward.

In my reply to Bill just now I have modelled the wind as a perceptual input rather than a disturbance, I think. Perhaps, you could modify it with your concepts.

What bothers me is that there does not seem to be a way to counteract the
wind when it is from the wrong direction, other than to lower the sail
and wait it out. So I think you would need some higher system that can
generate the reference "Wait" if the wind conditions are not right.

Isn't the sail state of 0 enough for that.

Ordinarily, as I think Martin said, it is not necessary to perceive a
disturbance directly or its origin, in order to control the variable that
is affected by the disturbance. Just monitor the perception, and if it
is not how you want it to be, "do something," and adjust the sign (+/-) of
what you do according to whether the outcome is getting worse or better.

That sounds like what I am doing in my model.

See what you come up with. Hope these ideas have been helpful. All the best, Erling

Regards,
Rupert

···

On 11/04/2012 21:20, Erling Jorgensen wrote:

[From Bill Powers (2012.04.12,0624 MDT)]

[Rupert Young 2012.04.11.2300 BST]

RY: On 11/04/2012 18:05, William Powers wrote:
> [From Bill Powers (2012.04.11.0946 MDT)]

I've attached an attempt at modelling this (xls and xlsx versions). I may have things arse over tit, but it works very nicely. You can change things for the boat to go backwards or forwards, and the sail goes up or down accordingly. See description.

Basically the output of control system B indicates whether the wind is favourable or not, and C changes the state of the sail accordingly. I am concerned though that there is no control system representing the boat movement.

Isn't the favorableness of the wind judged from its effect on the movement of the boat? You can't tell whether the wind is favorable to moving the boat if you observe just the wind and not the boat. If the boat moves in a good direction, the wind is favorable.

The model does work nicely. FYI, I'm using the Open Office Calc program, which works with xls files.

Bill

···

At 04:02 PM 4/11/2012, you wrote:

[Martin Taylor 2012.04.11.09.35]

[Rupert Young 2012.04.10.19.00]

Responding to both Martin and Erling.

I'd like to avoid getting into tacking or other degrees of freedom, as this is not actually about sailing. What I am interested in is when control is entirely dependant upon an environmental "function"; in this case the wind. For modelling purposes, and my simple brain, I'd like to keep things simple so I'll restate the situation, though let me know if it's too simple.

No, it's not too simple. But your specific example admits of a simple solution, which I offer below even though Bill has already posted something similar.

Think of a small boat with a very basic sail that could be raised or lowered. The goal of the sailor is go forward, and so she raises the sail when the wind is from behind and lowers it when the wind is from the front, and so, respectively moves forward or stays still. The sail then can be in one of two states, 1 (up) or 0, the wind, -1, 0, or +1 (from behind) and the movement of the boat -1, 0, or +1 (forwards). The overall reference, then is +1, for the boat movement.

It would seem that it is necessary for the system (the sailor) to perceive the state of the wind (though not control it) in order to "decide" what to do; that is, if the wind is from behind (+1) raise the sail. So considering comments made is it better to think of controlling the relationship between the perceived state of the boat movement and the perceived state of the wind? In this way the system would use the wind to propel the boat forwards.

I don't think it is necessary for the sailor to perceive the direction of the wind if she is controlling the direction of motion and can actually perceive the direction of motion directly (as opposed to deducing it from a perception of the wind direction). Suppose she keeps the sail raised when the motion is in the wrong direction. What happens? The error in the controlled variable is large and sustained. According to PCT, what is the result of that? Reorganization. In this simplified system, what could be reorganized but the sign of the sail up-down variable? So what happens if the sail stays up when the wind is ahead? The sailor's reorganization leads her now to drop the sail when the motion is backward.

The opposite case follows the same logic. Let us set aside the effect of the wind on the hull, and assume the boat is stationary if the sail is down. The controlled variable is the perceived motion of the boat. If the sail is down, the controlled variable is in error permanently, so reorganization would lead the sailor to rise the sail. If the wind is from astern, the error goes to zero and reorganization stops. If the wind is contrary, the state is as above, and she drops the sail.

This simple system oscillates if the wind is from ahead. But that's not wrong, since that's the best you can do if you can't observe the environmental state. Consider what you might do during a power cut to see if the power has come back on. From time to time you would switch on a light to see if it worked. How often you would do this depends on how critical you feel that it matters. The same with the sailor. If she feels that the wind usually doesn't change much over an hour, she might try raising the sail momentarily every half-hour or so.

However, the above doesn't get to why I said your question is not too simple.

So, is the wind an additional perceptual input to the system? If control of the relationship between wind and movement (so both equal +1) how would this be modelled? I'll have a go at a spreadsheet model. Ideas welcome.

You certainly could look at the relationship between wind and movement, but that couldn't be a controlled variable because it is a fact of the environment. If you perceive the sail to be down, then movement is zero. If the sail is up, movement is with the wind. The problem is that the wind direction has to be related to the reference value for movement. If the wind is from the north and you want to go south, you raise the sail, but if you want to go north you lower the sail and wait until the wind turns.

You could control a perception of the relation between sail setting and wind direction, as Bill P. suggests, but that is not controlling the perception of boat movement. Instead, you are imagining that particular relations between wind and sail result in particular boat movements. You have no idea whether other factors may be disturbing boat movement. The sail my be down and the boat going backward because of the current when the wind actually is behind you and you could make way if you raised the sail. If you are not controlling a perception of boat movement, but only are controlling a function of wind direction and sail setting, you wouldn't do anything about the problem. In the more general case, of which this is an exemplar, there are all sort of reason why the motion perception might be subject to disturbance.

Let's look at an HPCT structure to deal with this. At the top, you have a control unit that controls perceived boat motion. At the bottom, where, to mix a metaphor, the rubber hits the road, you have two sensors of the outer world, one of which is for boat motion, the other for wind direction. And you have one effector acting on the outer world, the sail up-down value. The sail state is available to be sensed, even though (in HPCT) the action to change its state is not. How can these all be connected to the controller of motion perception so that, if the boat is going slower than its reference value, the sail is up when the wind is behind and down when the wind is ahead? By itself, the relation between sail and wind is not a problem, since it doesn't involve the perception of motion. It becomes a problem when you are dealing with the control of motion perception.

Forget for the moment that the sailor is controlling a perception of boat motion, and consider the perception Bill described, the relation between sail and wind. The sail can have a value +1 or 0, and the wind can have a value +1, 0, or -1. The product Sail x Wind (SxW) can have a value +1, 0, or -1. A SxW value of +1 means that the sail is pushing the boat forward, and a value of -1 means the sail is pushing the boat backward. Now consider the output of the control unit that controls motion perception. If the boat is going too fast, you want SxW to be -1. If it is just right, you want SxW to be 0, and if it is going too slow you want SxW to be +1. These are the reference values that would be supplied by the output of the motion perception control unit.

If the wind is unfavourable (W = -1), you cannot achieve SxW = +1. The best you can do is SxW = 0, and you have to live with the ongoing error in the motion control unit -- or reorganize and buy a motor. If the wind is dead calm, SxW will be zero no matter what you do with the sail (the environmental feedback connection is cut). Finally, if W = +1, you cannot achieve SxW = -1. and again there is nothing you can do to slow the boat down unless you reorganize and introduce some other mode of action. That leaves you with the possibility of reducing error in the motion perception control unit only if the boat is going too fast and the wind is unfavourable or is going too slow and the wind is favourable.

Since you are using this situation as an example of a more general question of relating changes in environmental feedback possibility to changes in action, just redo the above analysis while allowing the "S" variable to go negative (a sail setting that would slow the boat when the wind is favourable or speed it when the wind is contrary).

The organization of this small hierarchy has a velocity control unit above a relationship control unit, which is unconventional but plausible. At some level, there must be an element that perceives the behaviour and the relevant aspect of the environment, and controls a relationship between them. I believe such a connection does not exist in standard HPCT, though it is quite consistent with Perceptual Control Theory.

The only PCT-correct alternative to this variation of HPCT that I have been able to make work is a more radical variation that requires some control unit to perceive the reference value for the boat motion and control a relationship between that and the environmental feedback possibiities.

I hope that Bill or someone can correct me on this.

Martin

[Rupert Young 2012.04.13.1400]

[From Bill Powers (2012.04.12,0624 MDT)]
BP: The model does work nicely. FYI, I'm using the Open Office Calc program, which works with xls files.

Is the model PCT-valid as far as it goes? Particularly control system B that outputs a signal if the wind perception is the same as the reference.

Isn't the favorableness of the wind judged from its effect on the movement of the boat? You can't tell whether the wind is favorable to moving the boat if you observe just the wind and not the boat. If the boat moves in a good direction, the wind is favorable.

In my model the favourableness of the wind is an output. How would I rearrange things so that it is an input to a control system?

I will look at the suggestions in Martin's response; he may have answered this.

···

On 12/04/2012 13:40, William Powers wrote:

--

Regards,
Rupert Young
Mobile: +447795 480387
Moon's Information Technology Limited

[From Bill Powers (2012.04.13.0844 MDT)]

[Rupert Young
2012.04.13.1400]

[From Bill Powers
(2012.04.12,0624 MDT)]

BP earlier: The model does work nicely. FYI, I’m using the Open
Office Calc program, which works with xls files.

RY: s the model PCT-valid as far as it goes? Particularly control system
B that outputs a signal if the wind perception is the same as the
reference.

BP:If the model is doing what you intended it to do, it’s valid!

BP earlier: Isn’t the
favorableness of the wind judged from its effect on the movement of the
boat? You can’t tell whether the wind is favorable to moving the boat if
you observe just the wind and not the boat. If the boat moves in a good
direction, the wind is favorable.

RY: In my model the favourableness of the wind is an output. How would I
rearrange things so that it is an input to a control
system?

BP: An output from what system? Doesn’t any system that can output a
signal indicating favorableness have to know where the boat wants to go?
Doesn’t favorableness have to be a perception?
Given the ability to sense whether there is a wind or not (without
sensing direction) and the ability to sense motion of the boat and
direction of motion relative to a destination (closer-farther), how would
you combine those three perceptions to produce a signal indicating that a
favorable condition exists? And what condition involving that perception
and the perception of the position of the sail would you then want to
maintain? I’m assuming that this is all happening at the program level so
we’re talking about logical true-false variables.
You can solve this problem using stimulus-response theory (no feedback,
just cause and effect), or using control theory (always controlling some
perception relative to some reference condition). If the problem is
simple enough, either way will work. But S-R theory depends on everything
you can’t perceive remaining the same, whereas control theory has more
latitude because a control system acts directly on the variable that is
to be controlled and senses whether that variable is in the desired state
or not.
Disturbances that have effects in directions that the controller can also
affect can be counteracted even if the controller can’t sense the
disturbance. If the controlled variable changes, produce an opposing
output. So if the controller can affect every dimension in which the
controlled variable can change, it’s not necessary to sense the
disturbance.
An S-R system always has to sense the disturbance in order to act on the
controlled variable in opposition to the effects of the disturbance, and
it has to calculate how to act because it’s not sensing the result. That
action can only be an estimate and has to assume constancy of
environmental conditions. And we have to ask how the S-R system figures
out what action is necessary. Doesn’t it have to sense the current state
of the controlled variable even to know if any action is necessary?
And how could it know exactly what action to use without experimenting to
see what effect different actions have on the controlled variable? Behind
every S-R system there is a control system. And, paying attention to what
Rick Marken keeps saying, often you will find that an S-R system actually
IS a control system if it actually works.
All these random-seeming thoughts are things to keep in mind as you try
solve the sailboat problem. I assume that when you ask how you
asking how you would find the answer if I got hit by lightning before I
could do it for you. I’m trying to tell you how. In mentioning S-R
theory, I’m also trying to tell you why we want to find a
usually many ways to solve the same problem. All you need is one that
works under all conditions.

Best,

Bill P.

···

At 06:43 AM 4/13/2012, Rupert Young wrote:

On 12/04/2012 13:40, William Powers wrote:

[From Rick Marken (2012.04.14.2007 EDT)

Bill Powers (2012.04.13.0844 MDT)--

Rupert Young (2012.04.13.1400)--

RY: Is the model PCT-valid as far as it goes? Particularly control system B
that outputs a signal if the wind perception is the same as the reference.

BP:If the model is doing what you intended it to do, it's valid!

I think a valid model maps properly to the variables involved. For
example, I think an important distinction between PCT and other
control models of behavior (which are mathematically identical to PCT)
is that PCT correctly maps the variables in the real system to
variables in the model. In PCT, for example, the input to the system
is the state of a perceptual variable; in other control models of
behavior the input to the system is an error signal. This, non-PCT
control models can say the perception controls behavior while the
valid mapping provided by PCT shows that it is actually behavior that
controls perception.

In other words, you can have a model that works exactly as you
intended, but is not valid.

Best

Rick

···

--
Richard S. Marken PhD
rsmarken@gmail.com

[From Rupert Young 2012.04.15.1900]

[From Bill Powers (2012.04.13.0844 MDT)]

BP: An output from what system? Doesn't any system that can output a
signal indicating favorableness have to know where the boat wants to
go? Doesn't favorableness have to be a perception?

Well, my control unit B compares the wind to the desired direction of
travel and outputs a signal accordingly, which could be said to indicate
that the wind is favourable. This then is used as a reference signal for
the sail state. Perhaps I am getting my terminology in a muddle.

Given the ability to sense whether there is a wind or not (without
sensing direction) and the ability to sense motion of the boat and
direction of motion relative to a destination (closer-farther), how
would you combine those three perceptions to produce a signal
indicating that a favorable condition exists? And what condition
involving that perception and the perception of the position of the
sail would you then want to maintain? I'm assuming that this is all
happening at the program level so we're talking about logical
true-false variables.

Ok I've had a go at this in the attached, but not sure I'm really doing
it correctly, see description. Any help appreciated. I don't yet have a
sail control unit, but can add that later. But as it stands I don't
think my perception combinations are correct and I don't see how the
overall goal of the boat direction can be incorporated, that is changing
the goal to -1.

sailing v1.3.ods (10.9 KB)

sailing v1.3.xls (11.5 KB)

···

On 13/04/2012 16:29, Bill Powers wrote:

--

Regards,
Rupert

[From Bill Powers (2012.04.16,0535 MDT)]

[Rupert Young 2012.04.15.1900]

BP: An output from what system? Doesn't any system that can output a signal indicating favorableness have to know where the boat wants to go? Doesn't favorableness have to be a perception?

RY: Well, my control unit B compares the wind to the desired direction of travel and outputs a signal accordingly, which could be said to indicate that the wind is favourable. This then is used as a reference signal for the sail state. Perhaps I am getting my terminology in a muddle.

BP: How do you compare wind to a direction of travel? You compare direction with direction. Only this is not the comparator, it's the perceptual input function looking at a relationship between two directions. The result is a perception of favorableness. How much favorableness do you want? If the direction of travel you're thinking of has a pile of nasty sharp rocks at the end of it, the direction is favorable for a shipwreck and you want to be sure you go in some other direction by manipulating the sail. The reference favorableness in that case is zero. If there is an error, you work the sail until there is less favorableness.

You have to put yourself inside the model to get its organization right. Favorableness is not something an external observer can see. Controlled variables are always defined by the behaving system. So you have to be the sailor, not the observer of the sailor. Then you will see what is a perception, and what is an output.

Best,

Bill

···

Given the ability to sense whether there is a wind or not (without sensing direction) and the ability to sense motion of the boat and direction of motion relative to a destination (closer-farther), how would you combine those three perceptions to produce a signal indicating that a favorable condition exists? And what condition involving that perception and the perception of the position of the sail would you then want to maintain? I'm assuming that this is all happening at the program level so we're talking about logical true-false variables.

Ok I've had a go at this in the attached, but not sure I'm really doing it correctly, see description. Any help appreciated. I don't yet have a sail control unit, but can add that later. But as it stands I don't think my perception combinations are correct and I don't see how the overall goal of the boat direction can be incorporated, that is changing the goal to -1.

--

Regards,
Rupert