Passage 1 · How a canal lock works
A canal has a problem that a river does not. A river finds its own way downhill; a canal
is dug, and the land it is dug through is rarely level. Somewhere along its length the
ground will rise or fall, and a boat cannot climb a slope. The lock is the eighteenth-
century answer to that problem, and it has barely changed since, because nothing simpler
has been found.
A lock is a chamber, usually just wider than a boat, with a gate at each end. The gate at
the upstream end holds back the higher water; the gate downstream holds back the lower.
The chamber between them can be at either level, because it is filled and emptied by
opening small openings in the gates themselves. These are called sluices, and on older
locks they are worked by winding a handle on the lock side. Each gate usually carries a
long wooden arm, the balance beam, which projects over the towpath; the crew push against
it to swing the gate, and its weight balances the gate so that one person can move several
tonnes of oak.
Consider a boat travelling downhill. It approaches with the chamber full and level with
the upper reach, so the upper gates can be opened; it enters, and the gates are closed
behind it. The sluices in the lower gates are then wound open, and the chamber empties
into the reach below until the two are level. The lower gates can then be opened and the
boat leaves, now several metres lower than it was. A boat travelling uphill does the same
in reverse, and the two directions are worth distinguishing, because a lock set against
you must be emptied or filled before you can even enter it, which is where most of the
waiting on a busy canal comes from.
No pump is involved at any point. Every litre that raises a boat has come downhill under
gravity from the reach above, and every litre that lowers one runs away downhill below.
This is the design's great virtue and its central weakness. It needs no power, which is
why locks built in the 1770s still work; but each operation spends a lockful of water,
typically two to three hundred cubic metres, and that water is gone from the upper level
for good. A canal summit — the highest section, with locks falling away on both sides —
therefore loses water every time a boat crosses it, in either direction, and has to be fed
from a reservoir or pumped back up.
Several solutions to that loss have been tried and mostly abandoned. Side ponds, which
catch half the water from a descending lock and return it to the next boat going up, work
but double the time an operation takes. Boat lifts, which carry a whole caisson of water
and boat vertically, use almost no water and were built at a handful of sites; they are
expensive and, being machines, they break. The staircase lock, in which several chambers
share gates so that the bottom gate of one is the top gate of the next, saves construction
cost rather than water, and has the drawback that two boats cannot pass each other in one.
There is one further constraint that shapes the whole system, and it is a human one. A
lock is operated by the crew of the boat using it, not by an employee, which means the
design has to survive being worked by somebody who has never seen one before. Everything
about the gearing reflects that: the sluice handle turns the same way on every lock in a
network, the beam is long enough that a child can move it, and a gate left open wastes
water but cannot flood anything, because the two gates are too far apart to be open at
once. Systems that assumed a trained operator — the lifts, the inclined planes — are
precisely the ones that no longer run.
The lock has survived, then, not because it is efficient but because it is robust. It has
no moving part that a boat crew cannot operate by hand, no component that cannot be
replaced with timber and iron, and no failure mode more complicated than a leak. On
Britain's canal network, which was built for freight that stopped being carried in the
1960s, roughly sixteen hundred of them remain in use, almost entirely for leisure, and the
oldest working examples are over two hundred and fifty years old.
Passage 2 · The rise and fall of the shopping mall
AThe enclosed shopping mall was invented, in its recognisable form, by an Austrian
architect who disliked almost everything American retail became. Victor Gruen's design
for Southdale Center near Minneapolis, which opened in 1956, put shops on two levels
around a covered court with a fountain and a garden. His intention was to give the new
suburbs a town square, since they had been built without one. What he produced instead was
a machine for selling things, and he spent the rest of his career saying so.
BThe commercial logic was immediate and it explains almost everything that followed. Two
large department stores placed at opposite ends of a building draw shoppers past every
smaller unit between them, which is why the big stores paid almost no rent and the small
ones paid a great deal. The climate was controlled, so the shopping season was no longer
weather. There were no roads to cross and no parking meters. Between 1960 and 1990 roughly
fifteen hundred enclosed malls were built in the United States alone, and the model was
exported to every country with a growing suburb and a car.
CDecline set in before the internet, which is the part usually forgotten. American retail
floor space per head roughly doubled between 1970 and 2000, far faster than either the
population or retail spending, and a great deal of that space was mall. By the late 1990s
a new mall was frequently killing an older one twenty minutes away rather than creating
new trade, and the industry knew it. The phrase used at the time was over-storing.
DOnline shopping then removed the categories the mall was least able to defend. Books,
music, film and electronics went first, and they went almost completely, because a
shopper buying a known item has no reason to travel. Clothing, which is the mall's
historical core, proved far more resistant than anyone predicted, for a reason the
early forecasts missed: fit. What has changed there is not where people buy but how often
they return what they bought.
EThe anchor model then turned out to be a trap. A mall's smaller tenants commonly held
leases with a clause allowing them to reduce rent or leave if an anchor closed, which had
seemed a fair protection when written. When the big department-store chains began closing
branches, one closure could therefore trigger dozens of rent reductions in the same
building, and the mall's income fell far faster than its occupancy did. Several large
American mall owners went through bankruptcy in the decade after 2010 without their
buildings ever standing more than a quarter empty.
FWhat is happening to the buildings is more interesting than the decline itself. A few have
been demolished. Rather more have been converted, and the conversions have very little to
do with shopping: a mall is a large, flat, structurally simple building with enormous
parking and good road access, which turns out to describe a distribution warehouse, a
medical centre, a college campus or a data centre almost exactly. One of the largest in
Ohio is now a logistics depot serving the online retailers that emptied it.
GNone of this means retail floor space is disappearing. What is disappearing is a
particular kind of it. Retail parks, where each shop opens onto its own car park and
nothing is enclosed, have held their occupancy through the same period; so have the
highest-end malls, which have responded by becoming something closer to a day out, with
restaurants, cinemas and gyms taking the floor that clothing used to. The mall that is in
trouble is the middle one: too large to be a convenience, not distinctive enough to be a
destination, and in a town with one more of them than it has ever needed.
HGruen, who died in 1980, disowned the form entirely in his last years and said he refused
to pay alimony to those bastard developments. He had wanted a town centre and had built a
shop with a roof on it. The irony is that the conversions now under way are producing
something closer to what he wanted than the malls ever were — mixed buildings with
housing, clinics and colleges in them — and they are producing it by accident, because the
shell happened to fit.
Passage 3 · What the octopus knows
The octopus is the standard example in any argument about animal intelligence, and it is
the standard example for a reason that has nothing to do with how clever it is. It is the
example because it is not related to us. Every other animal we call intelligent — the
chimpanzee, the dolphin, the crow, the elephant — sits on a branch of the vertebrate tree
and shares an ancestor with us that already had a centralised brain. The last common
ancestor of a human and an octopus lived something over five hundred million years ago and
was, on the best evidence, a flattened worm with a nerve net and possibly a light-sensitive
patch. Whatever the octopus has, it evolved separately. That makes it the only large-scale
natural experiment we have in whether minds have to be built the way ours is.
They are not built the way ours is. A common octopus has around five hundred million
neurons, comparable to a dog, but fewer than half of them are in the brain. The majority
lie in the arms, in clusters running down each one, and the arms retain a striking degree
of independence. An arm severed from the body will still reach for and pass food towards
where a mouth would be. In the intact animal, experiments by Hochner's group in Jerusalem
showed that an arm can execute a reaching movement without the central brain specifying
its path — the brain appears to issue something closer to a goal than an instruction. What
the animal's own sense of its body is like under that arrangement is not known and may not
be knowable.
The behavioural evidence is easier to gather and harder to interpret. Octopuses solve
puzzle boxes, open screw-top jars, and escape from tanks with a persistence that aquarium
staff describe in terms usually reserved for colleagues. They have been recorded carrying
coconut shell halves across open ground to assemble as shelter later, which is one of the
few uncontested cases of tool use in an invertebrate. Individuals in a laboratory
distinguish between human keepers who are dressed identically, and behave differently
towards the one who has previously handled them roughly. Godfrey-Smith has argued that
such behaviour is best read not as a list of isolated tricks but as evidence of a general
capacity to treat novel objects as things with properties worth investigating.
Against this, the sceptical case is respectable and rests on method rather than on
disbelief. Much of the published behavioural work uses very small numbers of animals,
sometimes a single individual, and octopuses are extremely difficult to test in the
controlled, repeated way that a psychological experiment requires. They are short-lived —
a common octopus lives one to two years — solitary, and sensitive to handling, so a design
that would be routine with rats is often impossible. Mather, who has worked on the animals
for decades and is not among the sceptics, has written that the field's central problem is
that the most striking observations are the least repeatable.
There is also a deeper objection. Almost every task we set an octopus was designed for a
vertebrate, which means it measures how well the animal does something we already thought
was difficult. An octopus can taste with its arms, change the texture as well as the
colour of its skin in under a second, and has no fixed body shape at all. If there is a
cognitive achievement in coordinating that body, our tests would not detect it, because it
has no analogue in anything we do.
One line of work does get round part of the problem. Instead of setting the animal a task,
several groups have begun simply recording what octopuses do when nothing is asked of
them — in the wild, over long periods, with cameras. The results have been unexpected: at
a handful of sites off Australia, animals of a species long described as solitary have
been filmed living within metres of each other, signalling with colour changes, and
throwing silt at neighbours who come too close. Whether that constitutes a social life is
contested. That it went unrecorded for a century of laboratory study is not.
Where the argument has arrived is a long way from where the popular accounts leave it. The
serious disagreement is no longer about whether octopuses are clever; it is about whether
intelligence is the useful word. A mind distributed through eight semi-autonomous arms,
with a lifespan too short for culture and no social life to speak of, may be doing
something for which our single word is simply the wrong shape — and on that point the
researchers who disagree about almost everything else are in unusual agreement.