In this blog I will discuss a recent move by the National
Trust to ‘macro-manage’ land in an attempt to get things broadly right over a
large geographic scale. Attempting to
manage land for nature can be difficult because nature is so complex, in the
face of such complex systems the National Trust’s strategy is not only cost efficient,
it is extremely sensible. To explain why
I think this, I first need to write a little on the nature of randomness. I will distinguish between three types of
randomness, inherent, game and apparent randomness. All three types of randomness are alike in a
crucial way: they make the future uncertain.
Inherent randomness belongs in the domain of physics, more precisely
quantum mechanics which states, amongst other things, that it is impossible to
know both where a (sub-atomic) particle is and where it is going. There is pure randomness at the root of
reality. Fortunately, as we move to
larger scales e.g the size of a cell all of the randomness averages out in a predictable
way hence this randomness has no influence on our lives. Game randomness is the randomness we are most
familiar with, the moment before the die is rolled we are uncertain about what
the outcome will be, if we pick a card at random from a deck we do not know for
certain which card we will pick. Now, if
the die is fair and the deck is a complete one then we can know with certainty
the probability of rolling a ‘2’ (1/6) or picking an Ace (1/13), we know how
uncertain the future is, how much we don’t know. The last type of randomness, apparent
randomness, is the most important to this topic. Apparent randomness is the case of facing an
uncertain future due to a lack of knowledge and understanding, it can be
thought of trying to play cards with a weighted deck (say, one in which all
clubs have been replaced by hearts) the composition of which the player does
not know. Here, if one knew the makeup
of the deck then they would be able to predict the probability of different
outcomes but the deck, the randomness generator, is not known to the
player. For example, though the weather
may be determined by laws, just as the movement of a planet is determined by
(Newton’s) laws of motion, our understanding of the weather laws is such that
we are unable to predict it, thus, for all purposes, the future weather appears
to us to be, at least slightly, random.
For another example consider the many precise (but often wrong) predictions
made by the Bank of England regarding future inflation/interest
rates/unemployment (see
here
for the result of a very quick Google search)
The weather is so difficult to predict because it depends
upon many ‘units’ (water droplets, the solids around which they form, local air
pressures and more) which can interact to form positive feedbacks. The result is a system which is rendered
hugely difficult to predict, in part because a tiny mis-estimation can have far
reaching consequences. In fact it was a
study of the weather which gave birth to the mathematical field of ‘chaos
theory’. Chaos theory can be understood
by imagining a tennis ball sat atop a large exercise ball in the middle of a
sports hall. If the perfectly balanced
tennis ball is instead placed slightly to the right then it will roll away to
the right, perhaps as far as the end of the sports hall and vice versa if the
ball is placed slightly to the left.
Thus a tiny mis-estimation of the starting position of the tennis ball
has far reaching consequences for predictions of the future position of the
ball. That the weather forms such an
unpredictable system is a problem for ecology, conservation and agriculture
because the weather plays such a large role in determining what happens at the
very base of every foodweb. Even if the
weather was perfectly forecastable, predicting the precise impact of altering a
complex foodweb via carrying out a conservation intervention would be nigh on
impossible as is predicting the impact of a government’s economic intervention.
Conservation interventions (outside of academia) are
normally carried out on the belief that doing A will result in a change in
B. If A is expensive then it will need
to be justified. The normal approach to
this problem is to make a prediction about the change in B which will result
from action A. Unfortunately this usually
means making predictions about complex systems which are not completely
understood. The more precise one
attempts to be, the more likely one is to be wrong, herein lies the
problem.
There are two
solutions to the difficulties of forecasting presents:
making vaguer
predictions and making no predictions at all.
Both of these solutions run counter to our nature and
also counter to the media’s handling of prediction making in the face of randomness. Firstly we cannot help but attempt to predict
the future by imagining various scenarios, not making predictions requires a
great mental effort and self-restraint.
Secondly, when we cast our minds forwards we do so by imagining one
possible scenario at a time. Such a
forecasting system is deeply flawed. A
useful forecast is not the sum of one or a few imagined scenarios, is the
(weighted) average of all the possible scenarios which includes in it a measure
of uncertainty.
The National Trust’s Wicken Fen (Cambridge) (see
here)
and High Peak Moors (Peak District) (see
here and
here) projects are both attempts
to manage land whilst making the minimal possible number of predictions. The National Trust is doing this by utilising
fairly unspecific tools (i.e. livestock graze land instead of fine scale
intervention by hand in Cambridge and blocking ditches in the Peak District) in
the hope that these will bring about broad benefits to the ecosystem. Precisely what those benefits will be, there
has been little attempt to precisely predict.
Instead the introduction of the livestock, ditch blocking and other
broad-stroke interventions, are forming lessons from which the National Trust
will learn in order to inform future efforts.
Restraining the extent of our meddling with nature,
basing this meddling on minimal ecological theorising and instead looking for
examples of what worked in the past may seem a non-scientific approach but it
isn’t. Learning what works is
scientific. If broad and approaches can
be shown scientifically to work best, scientists are obliged to put aside any
inherent preference for complexity and evaluate these approaches in the same
way as more complex ones.
Being broadly right means dealing with uncertainty, with
vagueness, a situation which may not sit well with us at first but it is surely
better than being precisely wrong.
