- From http://lamag.com/cover.htm Dawson
April 2001 Los Angeles Magazine
Imagine a traffic jam. It is ten o'clock on a Thursday evening and you are
stuck somewhere on the south face of the Sepulveda Pass, stuck behind what
must be the mother of all accidents. Well, stuck isn't the word: There is
some movement, a sort of seasick, stop-and-go swaying inching you down the
405. Twenty minutes pass, then another 20 minutes, and almost another, when
suddenly . . . boom, it ends, the concrete opens up before you--you're free.
But there is no mother of all accidents, no accident at all, not even a
flat tire--nothing to signify the preceding hour's unmoored misery. And you
think to yourself: What the hell is happening out here?
What happened was that three hours earlier, in a different traffic jam, a
woman late for a meeting answered her cell phone and, momentarily
distracted, plowed her Explorer into the next car. That accident has been
cleared for two hours. What you have been caught in is what engineers
believe is the kinetic equivalent of a spring expanding and contracting. But
more on that later . . .
Caltrans district 7, which comprises the 405 and every other freeway in
L.A., is arguably the most notorious road system in the country. If you want
to drive its borders, first get into your car at the 405-605 interchange.
Drive north on the 605 to the 60, then head east to the 71 north, jog west
momentarily on the 210, and then hang a right on the 39 into the San
Gabriels. After negotiating a few mountain gullies, merge onto the 2 heading
west to the 210, which leads you to the 5 and over the Grapevine. From here
hunt your way on back roads to the northern Ventura border and the Pacific,
where District 7 eventually runs out. You have just skirted 5,956 square
miles of the broadest coastal lowland in California, within which sit 98
cities housing 10.3 million people who, on an average day, drive a total of
99 million miles. Embedded in every lane of every freeway, charting the
flows, are 20,000 copper wire loops, generating data every 30 seconds,
monitoring volumes and speed. The data is shot over phone lines and across
satellite systems to the corner of Spring and 2nd, where the District 7
Transportation Management Center--or more intimately, TMC--sits.
Traffic slows for all kinds of reasons in District 7. The 405-10
interchange, the nation's worst bottleneck, squeezes down on traffic that is
already roiling and turning chaotic in the face of so many on-ramps and
off-ramps. The San Fernando 14-5 "Y" is a bottleneck; so is the 57-10
connection in Pomona. The largest bottleneck in Los Angeles was
inadvertently created by the Spaniards, who settled the pueblo at the
confluence of two valleys and a sediment plain; today five major freeways
converge there in an area of just 12 square miles. Bottleneck 90023.
These are stationary bottlenecks. Inside the TMC, on a black screen that
fills an entire wall, engineers watch them every morning on live feeds sent
back from District 7's 140 cameras. There are, however, floating
bottlenecks--freeway incidents--that haunt engineers' thoughts. They are
erratic and arrive without notice, and they account for one-half of the
congestion in Los Angeles.
In a city based on imagination, these incidents have no common trope.
There are the flat tires and crashes due to speeding, the tickets due to
speeding. There are the spills: Fuel-oil spill. Antifreeze spill. Orange
spill. Lemon spill. Cattle spill. "We have had cows out there running loose
on the freeway," confirms a Caltrans dispatcher. The traffic backs up. Other
animals run loose on the freeway. In the county's northern precincts, across
the 210 and the 118, coyote and deer cross the concrete unsuccessfully. The
traffic slows. On July 5, traffic snarls after the county's domestic animal
population, spooked and unhinged by our country's patriotism, runs onto the
freeways and is killed in record numbers. Sometimes animals are deliberately
introduced into the freeway network. Lacking rivers and bridges, owners of
the city's unwanted puppy litters leave them by medians in open bags to be
killed. On mornings after cockfights, Caltrans work crews are sent out to
recover trash bags of lacerated roosters that block lanes. The traffic backs
up. Motorists dump trash into the lanes. They leave couches, chairs,
refrigerators, stoves on the freeways. In the week after Christmas, Douglas
firs sprout in groves across the network, complete with stands. The traffic
slows. Other things are thrown away. Around February 14, the number of
pedestrians jumping off bridges into the network spikes, jamming traffic. In
the days following a well-publicized suicide, fatality crashes rise, copycat
suicides using Chevrolet instead of Nembutal. Pedestrians drop bricks,
bowling balls, pieces of the coastal range off overpasses. People are
dropped into the network. Story told by a Caltrans dispatcher: "I had a
friend of the family a long time ago. Someone I kind of grew up with. This
was my first year in the district. I don't know what happened to her, but
she threw her kids off the freeway. Off an overpass of the 110. Two
children. This is somebody I knew."
"We didn't plan for incidents when we built the freeways," says Dick
Murphy, who has worked on the network since it was a single lonely segment,
the Arroyo Seco Parkway, looking for a union. "Our goal was for lanes to
carry 1,800 vehicles an hour. So we looked at the demand, then the lanes,
divided the two, and came out with the freeways. Incidents were never
figured into capacity."
It's not too many cars, then. Fifty percent of all traffic jams, says
Caltrans, are the result of floating bottlenecks sucking up the network's
capacity. But what is capacity?
Let's go back to the Sepulveda Pass jam. Imagine that before you slowed
to a crawl, you were moving at a respectable 45 mph. Forty-five miles an
hour, according to every highway manual in the country, is the ideal speed
to move traffic through a network, when the laws of traffic behavior allow a
steady flow of 2,000 cars an hour through each lane. Increase the lane speed
to 60 mph, say the manuals, and the cars space themselves out, reducing the
flow; slow down to 35 mph, vehicles lose their mobility, the flow drops.
Thus, 45 mph is the optimum network speed.
This was true, anyway, before two recent discoveries. The first occurred
on an eight-mile stretch of I-15 just outside San Diego. Cars were doing
things there that had never been seen before--moving in platoon formation,
with two-second leads between them, at 60 mph. Twenty-four hundred cars an
hour were moving through each of the 15's lanes, a number the engineers had
thought was impossible before the copper loops picked it up. The second
discovery was made last October by a group of engineers led by Pravin
Varaiya at UC Berkeley's Institute of Transportation Studies. Looking into
the data generated by District 7's 20,000 loops, the engineers discovered
that peak flows--2,000 cars per lane per hour--actually occur in L.A. at 60
mph, not 45 mph.
Taken together, the discoveries mean two things. Pravin Varaiya will
tell you they mean that the Highway Capacity Manual--the bible of all
California freeway engineers--"is pure formula, wrong, based on no empirical
formula." Or maybe we're just beginning to understand how our freeways work,
just starting to take the first steps toward a unified field theory of
congestion. A San Diego Caltrans engineer might tell you something else
altogether: that freeways force drivers to evolve, that while 45 mph may
have been the limit for commuters to move themselves through a lane three
decades ago, that number has risen in the San Diego rush hour and may keep
on rising in coming decades, allowing freeways to move more cars faster than
There is a third phenomenon the twin discoveries point to--the
elasticity of capacity. Capacity, it turns out, is about as stable as
element 103 on the periodic table. The position of the sun will alter
capacity, dropping the number of cars that can pass through a freeway. A
rain cloud, a tunnel, a clear day in Los Angeles revealing the San
Gabriels--all alter capacity, imperceptibly affecting drivers. Varaiya found
that segments of L.A. freeway fluctuated in their capacity to move cars on a
daily basis by as much as 20 percent, even in the absence of incidents.
Varaiya's discovery means that on a Wednesday a freeway like the 405 could
move 300,000 cars, yet on a Thursday carry only 240,000 cars. No one has an
explanation for this yet.
It also means that traffic, as well, is inherently unstable. It means
that the 405, carrying 300,000 cars in a day, is continuously poised to
seize up, freeze, at the slightest fluctuation. Mark Hallenbeck, who teaches
engineering at the University of Washington, calls this phenomenon the Wile
E. Coyote Effect. One day, while walking his cat in his Redmond backyard,
Hallenbeck explained Wile E. Coyote to me.
"There is this mesa of performance," said Hallenbeck, "2,000 cars an
hour--the magic component of capacity. It's very delicate. Then something
happens. Nude sunbathers. Someone leans over to change a CD. A woman answers
her cell phone. Boom. You go right off the mesa, drop from 2,000 cars an
hour to 1,100 cars an hour at 25 mph. Just like that. You hit that ledge of
1,100 cars, then drop again to 600 cars an hour. You can't stop it. That's
This is what happened in our jam on the Sepulveda Pass. Three hours
before you arrived, a woman running late distractedly answered her cell
phone and hit the next car. The coyote went off the cliff. Yet why are you
still in traffic hours later?
"This is the problem of springs," said Hallenbeck. "At the accident,
traffic is starting to compress for all sorts of reasons--not just people
looking. And once it slows, it's off the cliff. More cars are piling in
faster than the bottleneck--the accident--can release them. You get this
compression, a shockwave that actually travels up the freeway, and then
release. Compression, then release--like a spring. That's the stop-and-go
traffic. And the spring will keep operating until the number of cars
arriving at the back is less than the release flow."
When does that number become smaller? I asked Hallenbeck.
"Well, that's the problem with L.A. flows. It happens at four o'clock in
the morning, okay?"
From the air, approaching out of the south by helicopter, the arc of the
110-105 commuter lane looks like a wan concrete smile that seems to shrug at
the sky and say, Who, me? The helicopter passes. In one seat, Rod Berensen,
retired police sergeant, current Fox 11 traffic reporter, talks into a
camera that beams his face onto 275,000 television screens. Beside Berensen
sits his pilot, Bryan Andrus, formerly of Alaskan helicopter tourism, now a
traffic hunter who looks remarkably like a robust Ken Burns in earphones.
Berensen's favorite adjective is boneheaded without the ed. Hovering
over downtown's jammed 110, eyeing a commuter who has abandoned his car to
walk through morning traffic, Berensen opines, "That's a bonehead move." It
is 6 a.m. on a winter morning, but this is not the first bonehead move of
the day. Already the 60 is jammed because of an overturned vehicle, the 405
is clogged behind a fuel spill, the Cahuenga Pass commute has slowed around
a truck gone into the sound wall. Upwards of 15 helicopters and fixed-wing
craft monitor District 7's traffic daily, and yet it is actually a lonely
beat for Berensen. L.A., the city of freeways, is a myth. Just 615 miles of
freeway make up District 7; out of 64 urbanized areas in the country, Los
Angeles ranks 63rd in the number of roadway miles per capita. Put simply,
relative to the rest of the union, we have no freeways.
It was not always to be thus. The early plan for L.A.'s freeways, the
1956 Freeway and Expressway Plan for Southern California, called for
approximately twice the number of freeways we have today. The Whitnall
Freeway, the Venice Freeway, the Mulholland Expressway, and Manhattan
Parkway, however, were never built; nor were the Hawthorne Freeway, Slauson
Freeway, Little Tujunga Expressway, and Industrial Freeway. Freeways were
planned to cross Laurel Canyon, Beverly Hills, and the Angeles Crest,
on-ramps to sit no more than four miles from wherever you happened to find
yourself in L.A. County. The money and the will, it turned out, were not
nearly as convenient.
Excluding the 210, which is plowing through the San Gabriel alluvial fan
on its way east to the 15, the Freeway Era of Los Angeles is over. There are
no more on-ramps to build. What is left is a study in pressure and volumes,
dazzling volumes--328,000 vehicles a day passing across the 101 where it
meets the 405; 317,000 cars moving through downtown on the Harbor Freeway;
322,000 vehicles daily on the 405 at Santa Monica Boulevard, one mile north
of the nation's worst bottleneck. Recording it all are the 20,000 copper
loops embedded in District 7's pavement. There are 12 Caltrans districts in
the state producing traffic data like this. In sum, California generates
around 3 gigabytes of data a day; District 7 accounts for half of that. The
data feed into the TMC on 75 modems, 10 gigabytes a week, organized and then
shot out again across bandwidth to Caltrans engineers and UC Berkeley
physicists, or to long-term storage in what are called optical jukeboxes.
Three jukeboxes sit inside the TMC, each holding 240 CDs, each CD capable of
storing 2.6 gigabytes. The jukeboxes began collecting data in April 1999;
today they are just about ready to reach the limits of "long-term."
From Berensen's helicopter, looking out over District 7, you cannot help
thinking you are gazing down on the geography of God's mind--a dendrite
system as complex and baffling in operation as the Deity itself. If the
metaphor is apt, then District 7's 20,000 copper loops make up the circuitry
of that mind, the electronic impulses they carry proof of the network's
thinking. It is thought in foreign code, the way Lexuses talk to Hondas, an
Esperanto of axles. When Berkeley physicists speak of scanning the
district's loop data for bottlenecks they call it "looking into the noise."
If the Freeway Era has ended in Los Angeles, the Congestion Era has just
begun. In the next quarter century an estimated 7 million people will move
onto the grid, almost doubling its volumes. And in the Congestion Era, the
engineers and physicists are up against a network that is not kidding: All
they are looking for are capacities, freeing capacities to move and direct
the flows into, and the bottlenecks that block the network's capacity
reservoirs. Engineers are fond of comparing traffic flow to water flow. The
hydrologic equivalent of the TMC's task would be managing the Sacramento
delta during a peak flood warning if peak flood warnings lasted 52 weeks a
year. No one in Caltrans can tell you where the capacities are because they
are usually invisible until you find them. But they think they know where to
look for the bottlenecks. They find them in the system's noise.
I went to see michael cassidy at Berkeley's Institute of Transportation
Studies after Pravin Varaiya said that maybe the physicist could answer
questions that baffled even Varaiya about his own research into the noise.
By this time I had been on the Berkeley campus, surrounded by engineers and
physicists, long enough to realize that I wasn't really in Berkeley; I was
in Mathland, an abstract of Berkeley, where everything is explained and
understood in algorithms and fractional equations. In Mathland I sat in on
seminars where it was normal for postdocs to walk in barefoot from rain in
40-degree weather because that wasn't really rain out there--it was a
complicated yet easily understood series of computations. In Mathland, it
wasn't traffic that clotted the I-80 every afternoon at 5 p.m., merely a
string of abstractions to be absorbed in the life of the mind. At the
Institute of Transportation Studies, every misery you and I experience on
freeways can be represented numerically. Here, for instance, is a
representation of you waiting at a signaled on-ramp, saving the commute time
of everyone else on the network:
Varaiya stood out in Mathland. He was, I thought, the coolest professor
of electrical engineering I had ever met--East Indian, handsome, chugging
through the rain with an ever-present cigarette dangling from his lips. In
his cardigan, he looked like a man David Lean could have made good use of.
He was also a heretic, having rejected the bible of all traffic engineers
because it was wrong. Varaiya knew this because the voices in the electric
noise of the L.A. desert told him so. In the noise, Varaiya shared with me,
he had found something else: Just before a jam hits, just before traffic
drops from 60 mph to 20 mph, something strange happens. Varaiya had
discovered a spike in capacity. Right before the coyote goes off the cliff,
Varaiya noticed, a freeway lane's capacity momentarily jumps--from 2,000
cars an hour to 2,400 cars. No one had seen this before, and no one could
explain it, but Varaiya thought that if the capacity spike could be
understood, then maybe, just maybe, engineers might find a hidden reservoir
of capacity tucked away in the network.
"Go see Cassidy," Varaiya instructed me. "He's a physicist--maybe he can
"Varaiya doesn't know what he's talking about," Cassidy told me upon
arrival at his office. "He's not seeing the invisible bottlenecks."
It's like this at the Institute of Transportation Studies--you get the
feeling you've been blown back to some pre-Hellenistic Alexandria, set down
among gnostics arguing over which wind said what howling out of the
boundless wasteland. Cassidy was a Mad Monk of a separate ilk, his thinning
hair slightly tousled, his burning eyes boring through sheaf after sheaf of
paper to be scribbled over, his voice rising and falling over the mountains
of algorithms he had to share.
"I know some things about how to deal with the noise," Cassidy told me.
"What I know, most people don't know. But it requires processing the noise
in careful ways that haven't been done before. You want to figure out what's
going on out there in District 7? Look for the invisible bottlenecks--that's
where the action is."
To understand what Cassidy is talking about, go back to our jam on the
Sepulveda Pass and the accident that has caused the bottleneck. Now, throw
the spring out of your mind. A spring, any physicist at Berkeley will tell
you, is actually a poor analogy for what is happening in stop-and-go
traffic. The kinetic action of two parting waves, they say, really describes
the behavior of traffic passing through a bottleneck. Imagine the accident:
Traffic leaving the downstream side of the crash is freeing up, cars spacing
out like loosed electrons, increasing their speed. This is what is called a
wave of acceleration. Watching it move in Mathland, it progresses just as a
wave across water does. Breaking from the crash in the opposite direction is
an upstream wave of deceleration, growing in its effect on traffic the
farther upstream it moves. What does this mean? It means traffic actually
slows and grows denser the farther you are from a bottleneck, not the closer
you are to it. In the case of our accident, or the 405-10 interchange for
that matter, this is not an important fact--we know where the bottleneck is.
But what if the cause of the bottleneck is undetected? People will look to
where traffic slows to its densest formation to fix it. Yet they will be
looking, Cassidy says, as far as five miles from the true bottleneck.
"No one knows how many bottlenecks are out there," Cassidy told me. We
sat inside his fourth-story office, looking out on a north campus unfocused
by Pacific storm. "The bottleneck comes in so many varieties that we don't
have a unifying theory to explain it. We can explain the jam; we can model
its evolution. But bottlenecks will perhaps always be unique. Right now
we're just trolling the noise, looking for where one wave moves away from
another. It's very boring."
In Mathland, Cassidy and his team of postdocs are looking for spots
where waves of acceleration and deceleration diverge--the equivalent of a
point break in surf. The point break, they know, will sit in the noise
between two copper loops--a distance of about half a mile. On the 5 they
found a point break. It took them a few weeks to discern its cause--of all
things, an off-ramp. On the 22, another point break: a malfunctioning ramp
signal was releasing cars in synchronicity with heavy surges in traffic,
which seem as endemic to the 22 as currents to the Mississippi. That
solution took another several months. In the coming year Cassidy and his
team are hoping to solve a third bottleneck that has confounded them for
some time. In fact, they've been looking at it--an on-ramp and a curve
outside Toronto, Canada--for two years now.
Cassidy's discovery presents a radical idea of what is happening on
L.A.'s freeways. He is saying that the jam you're stuck in is not
necessarily caused by too many cars, or even an incident cleared up hours
ago. Flying over the 5 one morning in Fox's traffic helicopter, Rod Berensen
looked down on traffic that surged then jammed, surged then jammed every few
miles and said, "That's that spring motion that hangs on the freeways for
hours after an incident has been cleared out." Cassidy, looking down on the
same scene, might say, "Well, maybe--but what you might have here is a
series of invisible bottlenecks that have yet to be detected." You have to
troll the noise that pools between loops to find the bottlenecks, and there
are 20,000 copper loops out there. On a good day about a fifth of those
loops are malfunctioning, sending out no data, silencing the noise. And
remember: Once you find the point break, get ready to spend up to three
years discerning its cause.
"It's like detective work," Cassidy told me as I was leaving his office
for another seminar. At the end of our talk, a bottleneck sat in my mind.
Cassidy imparts so much information so fast that it quickly backs up
somewhere behind your eyes, leaving you feeling disoriented, trapped--as
uncomfortable as someone stuck in the worst congestion. "It's like solving a
crime. There could be dozens of undiscovered bottlenecks out in District 7.
You have to go where the clues lead you. And there is no cookbook for
bottlenecks, no unifying theory. If Einstein couldn't do it, how is a
traffic engineer supposed to do it?"
Outside, my bottleneck cleared up without warning. Suddenly, the storm
light looked a little more beautiful, the smell of damp leaves was a bit
more distinct. It was a traffic epiphany. The Mad Monk had told me how to
fix the world, and why it could never be done--as close to enlightenment or
a beatific vision of the traffic jam as I'd come. I wanted to keep feeding
I skipped the seminar for the prix fixe lunch at Chez Panisse.
To get to Portland, Oregon, from Santa Fe you head northwest on I-40,
then merge left after some miles onto a road that rises and drops through a
series of red rock canyons and piñon forest, eventually climbing onto a
plateau where Los Alamos sits beneath the now-burnt Bandelier National
Forest. You drive down Oppenheimer Drive, past the Trinity Apartments,
turning right into the Pueblo Complex parking lot. Pueblo Complex is a
satellite of the national laboratories--a dun-colored Quonset hut with two
outstretched office wings that look out onto boundless acreage of blackened
trees which, in winter with snow on the ground, appear less like forest than
elements of a pointillist canvas. Enter the north-facing office wing, ask
for Steve Eubanks, and when Eubanks appears, ask to see his computer--not
really one computer but a string of 64 linked 500-megahertz computers
running one program, TRANSIMS. Portland is inside the computer: every
freeway, every street, every house and building, every person, every car and
truck. Because the cars and trucks are actually driving, 24 hours a day,
every Portland traffic jam is in Eubanks's computer, too.
Eubanks has a storm of red hair, a sunburned face, and a Western style
of dress. In New Mexico's outback he looks not unlike a man who might run
the underground cell of some militant environmental group, inspiring
pandemonium. In fact, Eubanks has a degree in randomness from the University
of Texas; as a postdoc in the physics department, he investigated chaos,
looking into sunspots, dripping faucets--any phenomenon lacking harmonious
function--then attempted to model said phenomena in cyberspace. Eubanks
joined the TRANSIMS group with a goal to model traffic flows around chaos
theory. TRANSIMS, if it worked, would be able to tell cities all kinds of
things: what will happen to your freeway traffic if you build a new highway,
if you add a commuter lane, if 7 million people move onto your grid. First
they had to rebuild Portland in cyberspace. They solicited travel diaries
from 3,000 of the city's residents, then distributed the freeway biographies
to the 1.5 million people in their computer. What they saw when they turned
Portland on was what traffic demographers have known for years: The
biographies of a city's residents say a lot about where traffic will form
and how intense congestion will be.
Asking drivers about their travel schedules is a delicate task. Alan
Pisarski, who helped devise the travel questions on the 2000 Census, calls
it "really getting into people's pants." When last year's census began
showing up in mailboxes, drivers around the country began sensing Pisarski
in their pants and dialed their congressman. Pisarski refers to our trips
across the freeway network as "lines of desire" and traffic congestion as
"people with the economic means to act on their social and economic
interests bumping into other people who have the means to do the same
thing." Last year, Pisarski bumped into the republic. But like a pickpocket,
he came away with valuable currency.
Women, for example, create more traffic than men. "A real cause of
congestion conflict," says Pisarski, "is the process of women joining the
labor force and the pressures of time placed on them." Those pressures form
what demographers call "trip chaining"--multiple stops to complete domestic
errands during the commute time. "Trip chaining generates more trips by
women in the worst period possible," says Pisarski. "It tends to suck into
the peak periods of congestion trips that historically would have happened
outside heavy travel time, slowing freeways." Another phenomenon Pisarski
found was that the link between children and women's travel patterns
contributes heavily to congestion. "While women's travel tends to be more
erratic," Pisarski says, "men's trip making is stable and totally
independent of the kids--like they're disconnected from the family."
If travel is the expression, and answering, of fundamental human needs,
it follows that people of a certain social status are better equipped to
answer those needs; or, to put it another way, the well-off create more
traffic than the poor. "Increasing income in a household," says Pisarski,
"leads to increasing car ownership, increasing trips per household,
increasing lengths of individual trips. You can map a city like L.A. and
locate certain neighborhoods that are heavy traffic generators along lines
of per capita income." Pisarski's findings may explain the 405's
near-endless cycle of congestion: On a north-south compass, the 405
traverses a majority of the city's affluent communities. That congestion may
dissipate somewhat in the coming year. Traffic rises and falls with economic
recession and boom periods. In the depths of L.A.'s last recession, Caltrans
engineers noticed that traffic improved on the 405. Suddenly, fewer
individuals were able to answer their fundamental human needs.
Steve Eubanks's computer is able to see all this. Which is why certain
people in L.A. are very interested in fitting L.A. onto his screen. At the
moment, the Southern California Association of Governments, or SCAG,
predicts congestion futures in District 7. Theirs is an unsure science.
Looking out from SCAG's 12th-floor windows above 7th Street downtown, a
gridlocked dystopia shaped by ballooning suburb populations always sits on
L.A.'s horizon. Yet traffic often confounds SCAG. A 1979 report stated that
the city would be in the grips of perpetual gridlock by 1995. SCAG failed to
take into account that traffic has a malleable shape, fanning itself out
across ever-widening time periods and geographies. In addition, recent
counterintuitive theories about our suburbs now paint sprawl as a solution
to, not a cause of, traffic. The suburbs are in a dynamic process of
becoming, shortening commuter trips or even reversing the flows, as the TMC
screens are beginning to show on the 101 and 5 freeways leaving the Valley
every morning. If SCAG eventually does look into Eubanks's computer to
predict L.A.'s future, however, it won't be cheap. Turning on L.A., the
association estimates, will cost $2 million.
The phenomenon of traffic readjusting itself, maintaining flows, is a
good thing. The dark side to this, the black blotch at the back of every
engineer's mind in District 7, is a separate phenomenon known as latent
capacity. Latent capacity says that if traffic is expressed desire, then in
a congested city like Los Angeles there is a lot of unrequited love out
there. Latent capacity explains why the moment Caltrans opens two new lanes
on a freeway they are immediately jammed. It explains why engineers believe
if you really wanted to accommodate all the potential trips in L.A., some
freeways would have to be widened to 16 lanes each. Drivers are just waiting
for space to open up on the network. The 1984 Summer Olympics are the best
example of latent capacity. A week in which L.A. was expected to experience
complete gridlock in fact turned out to be a period of the city's lightest
flows. Why? Drivers who expected jammed freeways withheld their desire,
conserving trips. Engineers who think only 16 lanes can contain all the
trips waiting to be made are saying that, in effect, we are still living in
the shadow of the Summer Olympics--drivers are still conserving trips,
waiting for the network to open up. It is a dark vision. It says that
whatever District 7's TMC does--add lanes, fix bottlenecks, remove
accidents--latent capacity is always poised to fill the space, negating any
positive effect, keeping the freeways in perpetual un-motion.
Let's return to the traffic accident on the Sepulveda Pass one last
time. If our driver was a working mother living in one of the communities
adjacent to the 405, we have some idea of the impetus that might have placed
her on the freeway at 7 p.m. But why did she plow her Explorer into the next
car while answering her phone? To answer that, we need to understand the
accident in terms of what psychologists call psychophysics--or how humans
make mathematical relationships between what's happening in the world and
how they perceive it.
Answering the cell phone was her first mistake. Most drivers, if asked,
would say that dialing a cell phone in moving traffic is the riskier
behavior. In fact, studies in Japan--the world's leader in the investigation
of cell phone-related accidents--show that drivers answering their ringing
cell phones cause the majority of such accidents. "Drivers suffer from
cognitive distraction because they are prethinking who is calling them,"
says Daniel V. McGeHee, a researcher at the University of Michigan. "It's
that cognitive distraction, not the mechanical task of dialing, that
exacerbates the chances of accidents. Cognitive task is the critical task."
In answering her phone, our driver experienced what is referred to as a
"loss of situation awareness." Most of us experience this on a weekly basis
as we run through the radio dial while driving the freeways. The point where
we suddenly look up and ask "How did I get here? How long was I out?" is
where we return from a loss of situation awareness. For our driver, this
made a bad situation worse; her Explorer's front bumper was closing in fast
on the next car. The process by which drivers realize the gap between them
and the other cars is shrinking is not necessarily set off by brake lights.
In fact, eye-camera studies show that drivers' attentions are more fixated
on the rear window of the preceding car than the taillights. Yet our brains
are fairly competent in detecting a shrinking gap between bumpers. When that
distance shortens by 12 percent--or, say, the first 12 feet of a 100-foot
opening--there is a 50 percent probability that we will detect the next car
slowing, even without brake lights. What we are not so good at, however, is
comprehending the rate at which that gap is disappearing. Closing velocity,
it turns out, is difficult for human brains to judge; we apprehend speed by
processing the growing image of the car in front of us. So while our brains
quickly register the fact that we are coming up on a slowing car, they are
poor at understanding the speed of that approach. Weak perception of
relative velocity, then, is one reason our freeways suffer so many rear-end
crashes during drive time, and why our commuter crashed her Explorer.
"Humans are just not terribly good control mechanisms," says James
Moore, an associate professor of civil engineering at USC. "Our wiring is
electrical-chemical, which makes transmission of information between our
brains and muscles pretty slow. We have what the controlling engineers call
a 'gain problem.' Our poor wiring forces us to drive conservatively. A tap
on the brakes in front of us, and we smash our brakes--leading to an
accident and a traffic jam. But if we could replace the physiological
control mechanism which the driver represents with some mechanical means of
control, we could have cars that react much more quickly to traffic, and
freeways that move many more vehicles than they do today."
Why do we have traffic jams? Moore is asking. Because we put humans in
the cars. Take our hands and feet off the controls of the cars, and
congestion could become a thing of the past.
Most of the nation's research into keeping human feet off the brake
pedal takes place at the PATH headquarters in Richmond, California. PATH, a
transit research center sponsored by the University of California and
Caltrans, sits on 150 acres of eucalyptus groves that, in springtime, suffer
their own congestion when visiting monarch butterfly flocks form bottlenecks
among the fragrant leaves. Steven Shladover, PATH's deputy director, knows
the downside of our electrical-chemical physiologies. Twice in the past year
Shladover has been hit in traffic by a following vehicle. He spends his days
at PATH dreaming up ways to let cars drive themselves. If you could
disconnect our electrical-chemical systems from braking and acceleration
procedures, says Shladover, you could improve District 7's flows. Instead of
2,000 vehicles an hour, each lane could move 6,000 vehicles an hour. PATH
calls this Automated Highway Systems.
One day last winter Shladover showed me a video of cars driving
themselves. It was a difficult viewing because the VCR's technology kept
eluding Shladover, stopping and starting the video on a whim. Once the tape
player settled down, there was PATH's Intelligent Traffic on the monitor:
eight Buicks driving themselves in platoon formation down I-15. There were
passengers inside the Buicks, but they, in fact, were the drivers--doing
exactly what drivers are expected to do in videos like this, waving both
hands at the camera, reading the morning paper, sipping black coffee. In
formation the Buicks braked, accelerated, banked into curves. They talked to
themselves through computerized radios, and with the roadway itself,
communicating with magnetic plugs sunk into the pavement. Eventually, the
tape ran out.
Shladover leaned back in his chair. "Automated traffic as an idea," he
said, "goes back to the futurists of the 1930s, who were dealing with
notions of driver comfort and safety. Today we're dealing with congestion.
In traffic, at 60 mph, drivers are only comfortable with two-second gaps
between their vehicles. From the air, what you see is mostly empty space on
the freeway--wasted capacity. When we run the Automated Highway Systems, the
cars are operating with one-15th-of-a-second gaps between them." In
Shladover's vision of the future, drivers will control their cars up to the
on-ramp, at which point magnetized lanes of the freeway system will take
control of the vehicles, eventually depositing them at the proper off-ramps.
"Of course, a system check would occur at the exit to make sure, say, that
the driver is awake and can take control of the car to avoid crashing."
Magnets are already going into California roads. On the 80 through Lake
Tahoe, enough magnets exist to run snowplows through blizzard conditions.
Yet Shladover feels a strange affinity with the wild acres of protected
wetlands on PATH's property. "Sometimes I feel like we are out in the
wilderness shouting about intelligent traffic," he says. "The commitment we
once had from government and industry has fallen away." Caltrans, however,
has just recruited ten other states to join in a study of intelligent
traffic, and Shladover is actually hopeful that cars will start driving
themselves within 15 years.
Fifteen years is a long time away. Latent capacity, psychophysics,
invisible bottlenecks--at the end of my stay in Mathland I was unsure if
there was any cure to traffic in the foreseeable future. But I had one final
seminar to attend--a foray into ramp metering run by Markos Papageorgiou, a
debonair Greek engineer whose visit to the campus was a shot across Pravin
Varaiya's bow, stylistically speaking. Half-dressed postdocs walked in from
the abstract of near-freezing weather, filling the hall's seats, settling
into silence. Freeways, the debonair Greek informed us, are the equivalent
of city streets before stoplights were invented. If we recognize that
removing our stoplights would result in total gridlock, he asked the hall,
why do we assume freeways will run successfully without similar metering?
The answer is not the archaic form of ramp metering we have today, which
simply lets cars on the network on average every four seconds, but active
metering that monitors and responds to real-time traffic conditions.
How could this debonair speaker be wrong? He made such obvious sense.
And by chance, it turns out that L.A. drivers will begin experiencing just
that as early as this year, when District 7's TMC turns on SWARM. SWARM
marks a departure in which our freeways begin thinking for themselves. It is
a computer program that actively senses traffic conditions up and down whole
segments of the network, activating and timing ramp meters so that they
respond to congestion before it can form.
"Congestion is not a natural phenomenon," Papageorgiou told his audience
as the lecture wound down. "It is something that can be controlled but is
not." A program like SWARM, operated correctly, was the solution. "With
proper ramp metering," finished Papageorgiou, "a city can reduce its
congestion delay by 50 percent."
A postdoc wearing Birkenstocks and a T-shirt stood up to ask the first
question of the afternoon.
"Won't your freeways be immediately swamped?" he wondered aloud. "I
mean, how do you fit latent capacity into your scenario?"
"That's not my business," said Papageorgiou, turning away. "That's
someone else's job. Next question."
- Endless silliness and stupidity.
We've spent untold billions of dollars, devoted decades of research and the
entire professional lives of thousands of researchers, crafted model after
model and tested innumerable permutations of possible solutions to the
problem of congested urban auto traffic.
Total progress? None. At all. It's been getting steadily worse every day.
When, one is entiled to wonder, will it occur to these intellectual giants
that they never get good answers because they persist in asking the wrong
Ishi, California's "last wild Indian," was perplexed and amused at much of
western technology. Generally, it seemed to him to be more trouble than it
was worth. He thought Europeans were "clever children."
He did approve of matches, though.
Dawson sent us:
> April 2001 Los Angeles Magazine
> Imagine a traffic jam. It is ten o'clock on a Thursday evening and you
> are stuck somewhere on the south face of the Sepulveda Pass, stuck behind
> what must be the mother of all accidents.
- The long article is intended to show that people need to be replaced as
but only on the parts of the road system that the most efficient
cyclists and walkers -- are banned (but, then, that requires drivers
emerge from their vehicles until they exit the freeways).
The vision of automated highways is not costed out; we are only assure
are plenty of pent-up demand for driving: further, faster,
"frequenter." It is
assumed that there is no higher state than traveling from one place to
occupying a footprint a thousand times larger than if one traveled by
foot. Is a
car trip more valuable or more important than a foot trip?
What it inadvertently shows is how much of today's road system is based
understanding about the behaviour of drivers, individually and
And the remark that the computer running the hundreds of thousands of
cars that their drivers turned over to is control will need to make sure
the driver is ready to re-assume control as it nears the driver's
prefered exist ramps suggests that, if the driver is not ready, the
computer would just keep the car on the road until he/she responds.
That could be forever, if the driver died; it would turn the freeway
system into portable cemetaries. Los Angeles would be the appropriate
place to have the world's first.
- Chris Bradshaw wrote:
>And the remark that the computer running the hundreds of thousands ofOr at least until it ran out of gas, then what?
>cars that their drivers turned over to is control will need to make sure
>the driver is ready to re-assume control as it nears the driver's
>prefered exist ramps suggests that, if the driver is not ready, the
>computer would just keep the car on the road until he/she responds.
>That could be forever,
>if the driver died; it would turn the freewayDrivers wanted? Dawson
>system into portable cemetaries. Los Angeles would be the appropriate
>place to have the world's first.