The promise and pitfallsof geothermal power

TechnologyQuarterly

Rewiring the nervous system

Bottom-up ways to deliver energy

The sceptical cyber-guru

September 4th 2010

Hot rocks andhigh hopes

TQCOV-Sept-04-10.indd 1 23/08/2010 16:20

w i d e o p e n.YoUR poSSiBiLiTieS. now

© 2010 Dell Inc. All rights reserved.

Introducing the new Dell Services. dell and perot Systems are

now one company with one purpose. Your purpose. instead of

coming in with a one-size-fits-all approach, we listen to what’s

on your mind. And then we put our technology expertise and

domain knowledge to work to create effective solutions that

will produce measurable results. So you can achieve whatever

your vision of success looks like. it’s why more industries, from

healthcare to financial services to manufacturing, trust us every

day. whether it’s managing your iT operations, improving

your business processes, or deploying cloud computing,

dell Services delivers. where would you like to go? As far as

we’re concerned, the future is wide open. dell.com/services

Applications Business process Consulting infrastructure Support

Services

Monitor

00 Better photo�t images, remotemedical monitoring, a new kindof jet engine, acoustic �bres,monitoring drivers’ behaviour,hacks to cut printing costs,improved speech recognition,using magnetism to measurefood quality, cracking quantumcryptography and crowdfundingcreative projects

Mining social networks

00 Untangling the social webWhat your network of friends canreveal about you

Inside story

00 Energy from the earth’s coreThe promise and peril ofgeothermal power

Bionic limbs

00 Rewiring the nervous systemA new approach makes arti�ciallimbs easier to control

Energy in the developing world

00 Power to the peopleHow bottom-up approaches canaddress the �energy gap�

Brain scan

00 The virtual curmudgeonA pro�le of Jaron Lanier, acyber-contrarian

2 Monitor The Economist Technology Quarterly September 4th 2010

THE human brain is hard-wired torecognise faces. Babies learn to identify

their parents’ faces within hours of beingborn, and even in old age people canremember what their childhood friendslooked like. But remembering faces is notthe same as being able to describe them.This is particularly apparent when wit-nesses are asked by the police to create acomposite picture of a suspect. Even whenthe result is thought to be a good likenessby the witness, that does not mean thatother people will also be able to recognisethe face and thus identify the suspect.

Indeed, even when working from afresh memory, the composite picturespeople produce are, on average, recognis-able to others only 20% of the time. Andthis falls to just a few per cent if the wit-ness is working from a memory more thana few days old. The problem is that facerecognition is a holistic process: people aregood at recognising faces as a whole, butstruggle when it comes to identifying ordescribing individual facial features, suchas a person’s eyes, nose or mouth.

But now police forces in Britain andother countries are using a new tech-nology that solves this problem by chang-ing the task from one of recall into one ofrecognition. Charlie Frowd at the Universi-ty of Central Lancashire, in Preston, andPeter Hancock at the University of Stirling,

in Scotland, have spent more than a de-cade re�ning their new approach, but itnow appears to be paying dividends.

Instead of asking witnesses to selectfacial features their system, called EvoFIT,initially presents them with a grid of 18randomly generated faces that match therace, gender and general shape of thesuspect. From these the witness is asked toselect the two that most resemble thesuspect, however vaguely. The softwarethen takes these two selections and�breeds� them, treating the facial featuresof their selections like genes, mixing themtogether and making random changes, toproduce 18 new o�spring which are dis-played in a new grid. As this process isrepeated, with the witness again choosingthe closest two matches, the faces generat-ed quickly converge on face that bears aresemblance to that of the suspect, says DrFrowd. In fact it usually takes only a fewcycles, he says.

But just how recognisable are thesefaces to others? To �nd out, Dr Frowdrecruited subjects to act as witnesses whowere shown a photographs of an unfamil-iar face, who happened to be a footballplayer. Two days later the witnesses usedboth EvoFIT and a traditional compositesystem commonly used by the police,called PRO-�t, to recreate this face frommemory. The resulting pictures were then

Total recall

Software: A novel approach to generating images of suspects uses a range oftricks to achieve a dramatic improvement in accuracy

On the coverGeothermal power stationsharness underground heat togenerate electricity involcanically active places suchas Iceland. A new approachuses drilling to allowgeothermal power to betapped almost anywhere. Butit has several obstacles toovercome: page 00

Contents

1

User: johnneeson Jobname: TQ1 Zone: UKPB ProofState: 20-08-2010----09:35

ART CART EDITORIAL READ BY: SPELL CHK:

2

1

ites were recognised by 11% of the subjects,against 5% for the PRO­�t composite.

Drs Frowd and Hancock realised that ifthe results were to be improved furthertheir evolutionary approach was notenough on its own. What was needed wasa dash of psychology. Research suggeststhere are fundamental di�erences in theway people perceive familiar and unfamil­iar faces, says Dr Frowd. With familiarfaces it is the internal features of the face�regions that include the eyes, nose, eye­brows and mouth�that matter most. Withunfamiliar faces, by contrast, the externalfeatures�the hair, ears, face shape andneck�play a more important role.

To account for this EvoFIT was adaptedto blur out any distracting external fea­tures when witnesses choose from thegrid, and restore them once their selec­tions have been made. And, to re�ne the�nal image and account for people’s gen­eral inability to gauge age accurately, theresulting image can be tuned to make theface appear older or younger, plumper orthinner, more or less attractive, or evenmore or less trustworthy. (This is done byaltering the size and shape of facial fea­tures in accordance with research in­dicating the traits that are generally regard­ed as attractive, trustworthy and so forth.)

To boost the performance of the pro­gram further, the researchers also decidedto address the way in which the compositeimages are presented. Typical mugshotstend to look similar to each other. Ratherthan seeing the unique features of theface, people tend to focus on similaritieswith other mugshots they might haveseen. So EvoFIT was designed to show ananimation of the �nal image, whichwould slowly accentuate the facial fea­tures, essentially creating a caricature,before morphing back to the original.

When these enhancements were put tothe test the improvement in performancewas staggering. In a second experiment,EvoFIT images led to correct identi�ca­tions 24.5% of the time with the blurringand feature toggling added. And when theanimated caricature was also added, thesuccess rate jumped to 42%.

So much for the theory. EvoFIT is nowbeing used in practice. The �rst formalevaluation was carried out by police inLancashire between 2007 and 2008. Dur­ing this time 30 EvoFITs were used, main­ly for serious crimes such as indecentassault. Although six arrests were madeduring this period�an apparent successrate of 20%�these cases also involved theuse of a traditional PRO­�t system. Andjust because someone is arrested does not,of course, mean that they were the per­petrator. But in a later 12­month trial withthe Derbyshire Constabulary in 2008,EvoFIT images were constructed in 57cases, resulting in 43 names being putforward by the public, 19 arrests being

made and seven people being charged. It is quite possible that the age­toggling

and caricature enhancements could beused to improve the success rate of tradi­tional non­evolutionary composite sys­tems, such as the PRO­�t and E­FIT sys­tems used by police forces in Britain, andthe FACES and Identi­Kit 2000 systemsused in America, says Dr Frowd. �But thereal problem is that about 70% of witness­es have such a poor recall of the o�ender’sface that a traditional composite cannot beconstructed in the �rst place,� he says. �Soyou need a system like EvoFIT that doesn’tneed good recall to produce an image.�

There are now 11police forces usingEvoFIT in Britain, and one in Romania. InAmerica, Dr Frowd is working with policein Boston to improve the software’s abilityto deal with Hispanic faces. One of theBritish forces, the Derbyshire Constabu­lary, has now started using EvoFIT exclu­sively. �It’s doubled our workload,� saysBeverly Hunt, the force’s facial identi�ca­tion o�cer, because it is now possible toproduce a useful image even when awitness has only a vague memory of asuspect’s face. 7

HALF way through a �ight from Mum­bai to London, a male passenger

complained of a swollen right hand andan inability to bend his �ngers. The �ightattendants were uncertain about what todo and hooked the passenger up to a smalldevice which took and transmitted vitalsigns, including his pulse, blood pressureand a picture of his hand, to a ground­based medical team.

As the passenger’s condition worsen­ed, the device was also used to transmit anelectrocardiographic (ECG) trace. Theresulting information was used to rule outheart problems, and the passenger wasstabilised and monitored with the assis­tance of a doctor on the �ight. The deci­sion was made to continue the journeyrather than divert to the nearest airport.

Medical emergencies like this oftenoccur on aircraft, and there are not alwaysdoctors on board. But in this particularcase the doctor says she would have askedfor the �ight to be diverted, had it not beenfor the ability of the portable vital­signsmonitor being carried on the aircraft torule out a heart attack. Such remote health

An online medic

Emergency medicine: Fieldmedicine, for soldiers and civiliansalike, gets smarter as medicalmonitoring technology improves

monitors are becom­ing even more sophis­ticated and are alsocapable of being usedin some of the mostextreme conditions.

The device usedon the Mumbai �ight was a Tempus tele­medicine monitor made by RDT, a Britishcompany based near Basingstoke. It sup­plies the devices for use on ships and atremote locations like oil wells, as well ason aircraft. The Tempus is similar to ahigh­end portable monitor used by someambulance crews, except that it is small,extremely rugged and capable of simulta­neously transmitting the data for vitalsigns, including ECG, blood­sugar andblood­oxygen levels, along with voice andvideo pictures. Moreover, it can do this inmultiple ways, with built­in Wi­Fi, mobile,satellite and Bluetooth communicationlinks. It is also designed to be simple tooperate using a touch­screen and pictorialinstructions, which means that the needfor training is minimal.

Now RDT has come up with an eventougher version, which has just gone intotrials with America’s special forces. It iswaterproof, soundproof (so as not to givea unit’s position away) and will operateover military radio systems. Because itwill be used by trained combat medics,new features are being added. Theseinclude a small ultrasonic probe that canbe used to scan for internal injuries, suchas �uid in the abdomen or internal bleed­ing, and a video laryngoscope, which canbe used to view and open a patient’sairway by inserting a tube�a tricky andpotentially dangerous procedure. A minia­ture video camera on the laryngoscope’stip relays a picture to the screen on thedevice to help its correct insertion. Themonitor can also be preprogrammed withmedical details of individual members ofa unit, or automatically obtain them from�smart� dog tags, which are being consid­ered by some armed forces.

Just as the crew of an aircraft or shipcan use such monitors to connect to re­mote medical centres in order to helpmake a diagnosis, obtain treatment advice

The Economist Technology Quarterly September 4th 2010 Monitor 3

4 Monitor The Economist Technology Quarterly September 4th 2010

2

IN A world worried about global warm-ing, improving the cleanliness and

e�ciency of jet engines is a priority forairlines and aircraft manufacturers. It isnot just altruism: greener engines also useless fuel, and so cut costs. Incrementalimprovements over the years have made adi�erence. Modern jets burn only half asmuch fuel per unit of thrust as their 1960scounterparts. But some people think it istime for a radical redesign. One of thosepeople is David Lior, the boss of a smallIsraeli �rm called R-Jet Engineering.

Jet engines rely on Isaac Newton’s thirdlaw of motion: for every action there is anequal and opposite reaction. When a jet isrunning, a compressor at the front drawsin air and compresses it (see illustration).This air is guided and di�used by staticblades to allow for easier ignition when itis mixed with fuel and ignited in a com-bustion chamber. The reaction comes inthe form of rapidly expanding hot gases,which blast out of the rear of the jet andthus drive the aircraft forward. As they doso, they pass through another set of staticblades which direct and accelerate the hotgases to turn a turbine. The turbine isconnected by a shaft to the compressor atthe front, thus turning it and keeping thewhole process running.

The approach taken by R-Jet involveshaving the air and hot gases in the com-bustor rotate with the compressor andturbine. To achieve this, the company useswhat it calls an orbiting combustion noz-zle (OCN), which turns with the compres-sor to inject the air into the combustionchamber as a vortex. The vortex is main-tained by blades that rotate on the innercasing of the combustor. This swirlingaction helps mix the air and fuel for amore complete and much quicker com-bustion. The hot gases then exit, also in a

vortex, to drive the turbine.This, says Dr Lior, eliminates the need

for the two sets of static blades. Thatmeans an OCN engine can be built morecheaply with fewer components. It wouldalso need to be only half the size of aconventional jet of similar power, says DrLior. The engine would use at least 25%less fuel and, he claims, its emissions ofcarbon dioxide and nitrogen oxide wouldbe cut by three-quarters because of itsunique ignition properties.

So why are airlines not beating theirway to R-Jet’s door? The company, found-ed by a group of Israeli military o�cialsand jet-turbine experts from the formerSoviet Union, has built a technologydemonstrator but needs a bigger partnerto take the concept further. As with anyradical change to an existing technology,

especially a jet engine, a large installedbase of expertise together with lots ofregulation mean it can be hard for a new-comer to make headway. To ease its entryinto the market, R-Jet reckons that OCN

engines could be used �rst as generators toproduce electricity, or to power un-manned drone aircraft. Having estab-lished a track record for themselves in thisway, the engines could then migrate totheir intended use on airliners.

The big aircraft-engine-makers areexploring new ideas for jet engines. Themost obvious change has been to theirshape. Whereas jet engines were long andslim in the 1960s, today they have a vastopening at the front containing a giant fan.This fan is used to blow up to 90% of theair around the outside of the compressorand combustion chamber. This air isslower-moving than that going throughthe jet, but its greater mass also means itprovides plenty of thrust�and moree�ciently too, which is why modern�high-bypass� jets burn only half as muchfuel as their 1960s counterparts. They arealso some 80% quieter because the massof slower-moving air shrouds the noisyhot gases coming from the rear of the jet.

One way to increase the e�ciency of ahigh-bypass jet engine is to turn the faninto an open rotor, a bit like returning topropellers, but using two rings of stubbycounter-rotating blades. Rolls-Royce andGeneral Electric are looking at this ap-proach. Pratt & Whitney is exploring put-ting a gearbox between the fan and theturbine, because fans run more e�cientlyat low speeds and turbines operate betterat high speeds.

But however high-bypass engines arebuilt in the future, at their core they willstill have a basic jet engine. So if R-Jet’stechnology can prove itself, it might pro-vide another leap forward. 7

Powering up

Jet engines: A nifty new engine design promises to improve combustione�ciency, thus cutting fuel consumption and reducing emissions

and determine whether to divert for helpor not, the idea is that a commander in the�eld will be better informed about theneed to evacuate a casualty�which mightmean calling o� an operation. And if thetrials with the special forces go well, saysGraham Murphy, RDT’s chief executive,versions of the new device could be pro-vided for civilian paramedic use.

Future capabilities, Mr Murphy hopes,might include a digital stethoscope toanalyse breathing sounds and a haema-

toma scanner to look for a brain haemor-rhage. And in both civilian and militaryuse, the vital signs recorded by the mon-itor could provide doctors attending acasualty who arrives at a hospital with ahistory of the patient’s condition and anytreatment administered�somethingwhich is not always readily available. Thiscould be sent remotely, or carried with thepatient and transferred via a memorystick. Medical emergency-room dramasmay never be the same again. 7

User: suzannebawden Jobname: TQ3 Zone: UKPB ProofState: 01-09-2010----13:31

ART CART EDITORIAL READ BY: SPELL CHK:

The Economist Technology Quarterly September 4th 2010 Monitor 5

1

�GUITAR HERO�, a hugely popularvideo game, has done wonders to

transform the �amboyant strumming ofcloset air guitarists into at least someapproximation of music. But soon eventhe feigned exertions of fantasy rock starsmay become unnecessary because re­searchers in America have developed anacoustic �bre, like a guitar string, capableof electrically plucking itself.

Electrical signals make the �bre vibrateto produce a sound (although rather quiet­ly, so you must listen to it closely). But theprocess can also be reversed, which ispotentially more useful. When acousticwaves cause the �bre to vibrate, it pro­duces a corresponding electrical signalthat can be detected. This means the �brescan also work much like a microphone. Inshort, the �bres can both sing and hear.

The work, by Yoel Fink of the Mas­sachusetts Institute of Technology and hiscolleagues, is at an early stage. But Dr Finkimagines that fabrics woven from these�bres would be capable of continuouslymonitoring the sounds of the body forhealth screening. Obstructions in blood�ow carry very speci�c acoustic signa­tures that could be picked up by, say, astocking laced with these �bres.

The �bres are based on a type of piezo­electric polymer. This is a material whichcan translate electrical �elds into mechan­ical stress, or vice versa. Although piezo­electric materials are well known for theirtendency to vibrate when an electric �eld

is applied, drawing them out into thin�bres would require a huge amount ofenergy and would merely result in the�bre stretching and contracting along itslength, says Dr Fink.

To make a �bre vibrate in a transversefashion, like the strings on a musical in­strument, a more complex architecture isrequired. Dr Fink’s �bre consists of a hol­low rod of an insulating polycarbonatematerial with a 30­micrometre ring of apiezoelectric polymer running along itslength. This is sandwiched between innerand outer rings of conducting materialthat act like electrodes, and the wholething is surrounded by yet another layerof polycarbonate as cladding.

What is remarkable about these �bresis that they are only 1.1millimetres indiameter. They are drawn from a muchlarger �preform� block of material, inmuch the same way that conventionaloptical �bres are created. This allows thevarious layers to be assembled at a moremanageable scale and then heated andstretched out, like seaside rock. It is trickyto do this without changing the molecularstructure of the active material, says DrFink. �A key issue for piezoelectric materi­als is order,� he says. So when the �bre isdrawn the researchers apply a powerfulelectric �eld to ensure that the moleculesin the plastic all line up in the same direc­tion as the �bre cools.

Adding electrodes posed a furtherchallenge. Simply �anking the piezoelec­tric layer with thin concentric layers ofmetal would not work because metalstend to have a low viscosity. �When youdraw a material of low viscosity it tends tobreak up,� says Dr Fink. Instead the re­searchers use two layers of graphite­richconductive plastic, each �anked by asingle metal strip in the inner and outercladding. They recently described theirwork in the journal Nature Materials.

By creating complex material struc­tures within the �bres, new things becomepossible. To demonstrate this Dr Fink andhis colleagues have also shown how the�bres can be used for communications.This involves covering the �bres with are�ective coating, and then controllingtheir vibration by applying electricalsignals. When laser light is bounced o�the �bre, the re�ected light is modulated inaccordance with the electrical signals. Thiscould be used like an optoacoustic versionof the radio­frequency bar codes that areused by cars passing through the paybooths of automated toll roads or to tagsome goods in shops.

Other proposed uses for the self­pluck­ing acoustic �bres include nets that mon­itor the �ow of water in the ocean andlarge­area sonar imaging systems. Andonce the �bres can be made small enoughto be woven into clothing, they could alsofunction as a wearable microphone. 7

A suit that cansing and hearMaterials: Optical �bres made ofpiezoelectric materials can turnsound into subtle electrical signals,and vice versa

SOME people always take things toextremes. For those trying to save fuel

there is hypermiling, in which the reallydedicated try to use less than 4.5 li­tres/100km (ie, travel more than 80 mileson a gallon) in a car that under normal usemight do only half as well. Apart fromdriving very slowly and trying not to usethe brakes (which dissipates energy),hypermilers employ other tricks, such aswiring the fuel injectors up to lightsmounted on the dashboard so they cansee whether or not they are squirting fuelinto the cylinders. Although this is all toomuch trouble for most motorists, thehypermilers do have a point: drivingtechnique plays a big part in how muchfuel a car consumes. Now various devicesare being used to help teach more moder­ate ways of driving economically.

Not surprisingly, companies that oper­ate �eets of cars and trucks are among the�rst users of fuel­saving �eco­assist� sys­tems. The most popular of these are glo­bal­positioning system (GPS) units thatuse live tra�c information and other data,such as weather and past trends, to plotnot the fastest but the most economicalroute to a destination at a particular time.According to iSuppli, a Californian re­search �rm, fewer than 1% of new carshave such �eco­routing� systems �tted, butit expects that by 2020 a third will.

Interest is also growing in other devicesthat go beyond a simple fuel­economymeter and provide more informationabout drivers’ behaviour. One such isused by Masternaut Three X, a Britishcompany that specialises in vehicle track­ing. It taps into the engine­managementsystem which, because of the increasingamount of electronics used in cars, con­tains data that can be analysed to monitorsuch things as excessive revving and harshbraking. This information can be shownas a series of warnings on the dashboardand is monitored by �eet managers. Firmsusing such systems say they can yield fuelsavings of around 10% a year.

As well as discovering those with theheaviest feet on the accelerator pedal whocould do with retraining, this can also helpreduce accidents. �There is a correlationbetween driving e�ciently and safely,�says Dan Steere, chief executive of Green­Road, a company based in California. Forinstance, driving more smoothly by antici­

Gently does it

Motoring: Spies on the dashboardcan teach people to drive moreeconomically�and tick them o� ifthey fail to do so

THE dream of the paperless o�ce hasbeen around for years, but it has re­

mained just that, despite the rise of e­mailand the web. True, paper consumption inAmerican o�ces peaked in 2001, but sincethen it has declined only slightly from itshigh of around 150 pounds (68kg) of paperper worker per year. In Europe, mean­while, each worker prints an average of 31pages a day, seven of which were not evenwanted, according to recent research byLexmark, a printer manufacturer.

The cost of all that paper, toner and inkquickly adds up. Which is why, earlier thisyear, the University of Wisconsin­GreenBay adopted a novel strategy to save mon­ey on print supplies: it changed its fonts.Programs like Microsoft Outlook default toArial, but a thinner­lined typeface such asCentury Gothic requires less toner or inkto form its characters. A study in 2009showed that switching to Century Gothiccould save businesses as much as $80 per

printer per year. The university predictsthat this year it will reduce its $100,000print­supplies bill by around 10% by mak­ing this simple change.

�No one had put the facts together,namely, that ink or toner is very expen­sive, that people still print a lot, and thatthere is a lot of variance between fonts inhow much ink they use,� says DianeBlohowiak, the university’s director ofcomputing. So far no students or sta�members have complained. And the FBI,local governments and even America’sspace agency, NASA, have all come calling,hoping to make similar savings too.

Thrifty though a font like CenturyGothic may be, a Dutch company, Ecofont,has found a way to make a font that iseven thriftier: by punching holes in theletters. The �rm did this by creating a newversion of a popular font called Vera Sanswhich is shot through with tiny holes (seebelow). The �rm says this can reduce theamount of ink or toner needed by 25%,with no e�ect on legibility. (Ms Blohowiakand her colleagues disagreed, deciding notto use the new font because it was some­times di�cult to read on screen.)

Ecofont’s new software goes one better,letting people use �normal� fonts onscreen and inserting the holes only at printtime�and then only for small type sizes,where they are less apparent. Ecofontrecently picked up a European Environ­mental Design Award for its work.

Another money­saving trick involves amore low­tech approach. Given the in­dustry’s model of selling printers atknock­down prices and then charginghigh prices for re�ll cartridges�one in­vestigation famously found that inkjet inkcosts more than seven times as much,millilitre for millilitre, as 1985 Dom Perig­non champagne�there is understandablescepticism when printers claim to berunning low even while turning out beau­tiful, streak­free pages. Surely there mustbe more ink in the tank?

There often is. Sensors in ink and tonercartridges report them as being emptyeven when they are not, a fact that has ledto lawsuits (manufacturers say they needto protect the non­replaceable print headfrom trying to print without ink).

Many laser printers use an opticalsensor to measure toner levels. Instead ofsimply warning users, printers refuse toproduce another page until the cartridge is

changed. But frustrated users �ght back inonline forums. Perhaps the most populartrick is to cover the tiny window in tonercartridges with a piece of tape, fooling theoptical sensor into thinking that the car­tridge is always full. This correspondentrecently squeezed an additional twomonths of life from a supposedly emptytoner cartridge before the �rst streaksappeared on printed pages.

A further way to cut costs and reduceenvironmental impact is to use less paper.These days it’s usually a matter of a fewclicks to print out a document on half asmany sheets, simply by rotating andsqueezing two pages onto each one. Butthis can be hard on the eye. So a betteroption is duplex printing: printing at fullsize, on both sides of each sheet. Accord­ing to a recent European Union �greenpurchasing� guide, duplex printing can cutcosts by 38% over the life of a printer.Citigroup once estimated that it could save$700,000 a year and eliminate 76 tonnesof solid waste if every employee savedonly one sheet of paper per week by usingduplex printing or copying. Duplex print­ing is a common feature on modern print­ers and copiers. But when was the lasttime you used it? The real challenge, as theEU guide notes dryly, is persuading peopleto �actually use the duplex function�.

Using clever fonts, making toner car­tridges last longer and saving paper are allbottom­up ways to cut printing costs.There is also an increasingly populartop­down option, known in the trade as�managed print services� (MPS). Thisinvolves outsourcing the operation andmanagement of o�ce printers and copiersto an external supplier such as HP, Xeroxor Ricoh. That supplier is paid a monthlyfee, and then has an incentive to cut print­ing costs by exploiting economies of scalein procurement, replacing printers withmore e�cient models and so forth. SomeMPS suppliers even monitor individualemployees’ use of printers or copiers inorder to identify particularly wasteful orine�cient practices.

The industry’s rule of thumb is thatMPS can cut costs by around 30%. Ascompanies look for ways to save money, itis not surprising that the global MPS mar­ket grew by 27% to $25.8 billion in 2009,according to Photizo Group, a market­research �rm.

Of course, the best thing, from both acost­saving and an environmental pointof view, would be not to print at all. Inparallel with e�orts to cut costs withintheir o�ces, many �rms have also in­troduced automatic e­mail signature linesthat ask external recipients to �considerthe environment before printing thise­mail�. Does this work? Nobody knows.But until the paperless o�ce �nally comesalong, it is at least a cheap way to lookenvironmentally conscious. 7

Ruses to cutprinting costsO�ce technology: All kinds oftechnological tricks are being usedto reduce the cost and environmentalimpact of o�ce printers

How to print in green on a black­and­white printer

6 Monitor The Economist Technology Quarterly September 4th 2010

2 pating manoeuvres and then braking andaccelerating lightly not only uses less fuelbut also tends to make drivers more alertto avoiding potential accidents.

GreenRoad’s driver­monitoring devicedoes not need to tap into the car’s innards.It uses GPS to measure a vehicle’s speedand a set of accelerometers to measure theforces acting on the car as it accelerates,brakes and turns. The data are analysed todetermine how the vehicle is being drivenand the results shown to the driver asgreen, amber and red LEDs. For �eet cars,the data can also be relayed to a controlcentre, so that Big Brother can tick o�o�ending drivers.

Mr Steere claims that despite the Or­wellian overtones most drivers will soonbecome comfortable with the device andvalue the fuel savings it o�ers. He hasclearly never met Jeremy Clarkson. 7

The Economist Technology Quarterly September 4th 2010 Monitor 7

1

THERE is often something sweet, inti­mate even, about couples who �nish

each other’s sentences. But it can also be asource of irritation, especially when theyget it wrong. A similar irritation (minus thesweetness) is often felt by users of speech­recognition software, which still managesto garble and twist even the most clearlyspoken words. Might the solution lie in amore intimate relationship between theuser and the software?

Modern speech­recognition programsdo not merely try to identify individualwords as they are spoken; rather, theyattempt to match whole chunks of speechwith statistical models of phrases andsentences. The rationale is that by know­ing statistical rules of thumb for the way inwhich words are usually put together�anabstract probabilistic approximation ofgrammar, if you will�it is possible tonarrow the search when attempting toidentify individual words. For example, anoun­phrase will typically consist of anoun preceded by a modi�er, such as anarticle and possibly also an adjective. So ifpart of a speech pattern sounds like �ball�,

the odds of it actually being �ball� willincrease if the utterances preceding itsound like �the� and �bouncy�.

Although this �continuous speechrecognition� technique has indeed im­proved accuracy, it is by no means in­fallible. Moreover, when it gets thingswrong, it often does so spectacularly. Theproblem is that, as a direct consequence ofthis technique, the misidenti�cation ofeven a single word can take the programo� on the wrong path as it tries to predictwhat the rest of the phrase is likely to be.

Though such errors are inevitable,there may be a way to let speech­recogni­tion programs take the pain out of makingcorrections. Per Ola Kristensson and KeithVertanen, at the University of Cam­bridge’s Computer Laboratory, have de­veloped a method of allowing speech­recognition programs to share theirthoughts, as it were, with the user in orderto speed up the correction process. Theirsolution, called Parakeet, is a touch­screen­based interface for phones and othermobile devices, which not only displaysthe words, phrases or sentences thatscored highest in the program’s statisticalmodel, but also any close contenders. Thisallows the user to select alternatives easily,with a quick tap of the �nger. More subtly,if none of the predicted sentences is en­tirely correct, yet collectively they containthe words that were spoken, the user cansimply slide his �nger across the appropri­ate words to link them up.

In a sense, all Parakeet is doing is allow­ing the user to see which alternativewords or sentences the program wouldhave predicted. The di�erence is thatexisting programs require the user tocorrect each word individually from adrop­down list of alternatives, or else toretype or reutter the words. What is frus­trating about this, says Dr Kristensson, isthat more often than not the correct stringsof words were recognised, but rejected bythe speech­recognition program on statis­tical grounds. Parakeet makes them allavailable to the user.

The prototype uses an open­sourcespeech­recognition program called PocketSphinx, developed at Carnegie MellonUniversity, in Pittsburgh, but Dr Kristens­son reckons it would be easy to apply thesame approach to commercially availableprograms like Nuance’s Dragon. So far DrKristensson and Dr Vertanen have carriedout only limited trials on a handful ofpeople. Even so, these have achievedoperating rates of around 22 words perminute�considerably higher than the 16an average user can achieve using predic­tive texting. With the likes of Google,Nuance and Vlingo now o�ering mobilespeech­recognition services for phones,and the development of speech­drivensystems for use in vehicles, Parakeet maybe �ying into a growing market. 7

Correct me if I’mwrongð

Software: A new approach to speechrecognition gives users the chance to�x misunderstandings withouthaving to repeat themselves

TO MOST people magnetic levitation(maglev) connotes high­speed pas­

senger trains. It is what enables the Shang­hai Transrapid to glide over the tracks atspeeds of as much as 430kph (267mph).But the same technology has recentlyfound a much more pedestrian use intesting food and water.

One way to identify a substance with­out resorting to �ddly chemical methodsis to determine its density. This will notprovide a precise composition but it cangive a decent approximation. The purityof minerals is often assessed in this way, asare things like the amount of fat in milk orsalt in water. (The less fat there is in milk,the more dense it is; the less salt there is inwater, the less dense it is.)

The problem is that existing devices formeasuring density tend to be either pre­cise or portable. Those devices that areboth, such as oscillating tubes, can set youback several thousand dollars. But GeorgeWhitesides and his colleagues from Har­vard University have come up with aningenious way to square this circle usingthe principles of maglev.

Most materials, including milk andwater, are diamagnetic, which means theyare repelled by external magnetic �elds�though this phenomenon is far too subtleto be observed in normal circumstances.This means that when a drop of liquid ismixed with a suitable solution of para­magnetic ions, which become magnetic inthe presence of a permanent magnet, itwill be pushed to where the �eld is weak­est, and stay there.

The researchers twigged that this phe­nomenon could be used to measure densi­ty and, as they report in the Journal ofAgricultural and Food Chemistry, they setabout this task by erecting a stack of twoo�­the­shelf neodymium magnets sepa­rated by a vertical vial of paramagnetic�uid. The magnets’ like poles were facingone another, creating a �eld which wasweakest precisely in the centre of the vial.

When a drop of an unknown sub­stance is injected into the vial, however, itwill not settle quite in the middle. Themagnetic forces will push it towards thecentre, but gravity will pull it downwardsuntil an equilibrium point is attained. Thedrop’s height above the bottom magnetcan then be measured using a standard

Fast­tracktesting

Magnetic levitation: The sametechnology used to make trains gofast can help identify unwantedsubstances in food and water

8 Monitor The Economist Technology Quarterly September 4th 2010

2 millimetre ruler, and that measurementcan be plugged into a formula to arrive at aprecise �gure for density. (In an earlierpaper the researchers conveniently in­cluded an Excel spreadsheet that performsthis calculation automatically.)

Dr Whitesides puts the total cost of thecomponents for his device, including analuminium casing, at less than $50. Massproduction would reduce the cost even

further. It is a small price to pay for a quickand easy way to check whether, for in­stance, water is too briny for irrigation. Hethinks the device could even prove handyin some carefully chosen biomedicalapplications, especially in the developingworld where inexpensive solutions are indemand. The Bill & Melinda Gates Foun­dation, which helped �nance the project,must certainly hope so. 7

IT SOUNDS foolproof. One of the funda­mental tenets of quantum mechanics is

that measuring a physical system alwaysdisturbs it. If the system in question is amessage written as a series of digital bitsencoded in the polarisation of light, thismeans that intercepting and reading themessage can no longer be done surrepti­tiously. The receiver should be able todetect an eavesdropper and take appropri­ate countermeasures.

To a hacker, though, the word �fool­proof� is a challenge. And to prove thepoint, two groups of academic spies havenow shown that whatever the theorysays, practical attempts to hide messagesthis way can still be vulnerable.

In order to encrypt a message, thesender, known conventionally as Alice,scrambles it using a secret key beforesending it to the receiver, known as Bob.

Even if Eve, the eavesdropper, interceptsthe message, she cannot make sense of itwithout the key. The problem, then, is howto pass the key from Alice to Bob withoutEve getting hold of it as well.

Quantum key distribution does this byencoding the information in the polar­isation states of individual photons, theparticles of light, which are sent from Aliceto Bob over an optical �bre. If Eve taps intothe line and intercepts the key, she disturbsthe photons when she measures theirpolarisation. By comparing a subset of thephotons that Alice sends with what Bobmeasures, the pair can check for the pres­ence of errors introduced by Eve. If errorsare detected, Bob can throw away the keyand ask for another.

In practice, quantum­key­distributionsystems rely on sophisticated opticalequipment to prepare, transmit and detect

the individual polarised photons thatmake up the key. And when these real­world components meet the clever aca­demic theorems that are supposed toguarantee security, holes emerge.

In the �rst piece of research, a teamfrom the Norwegian University of Scienceand Technology and the National Univer­sity of Singapore, led by Vadim Makarovand Ilja Gerhardt, hacked into a systemthat connects several buildings on theNational University of Singapore’s cam­pus. Their eavesdropping apparatus(which is small enough to �t in a suitcase)was designed to take advantage of a weak­ness in a particular sort of photon detectorin Bob’s receiving equipment. If hit with abright enough �ash of light, such detectorsare blinded. And if, on top of the brightpulse, a smaller pulse of just the right typeis sent, the detector can be forced to recorda one or a zero.

In essence, Eve now has control ofBob’s detector. After intercepting the key,she can make it record just the right pat­tern of bits without any of the telltaleerrors her eavesdropping was supposed tointroduce. Using this technique, Dr Mak­arov and his team were able to steal theentire key without leaving any trace oftheir activities.

The second hack was carried out by ateam from the University of Toronto, ledby Hoi­Kwong Lo. They stole informationfrom a research version of a system madeby ID Quantique, a Swiss �rm that istrying to commercialise quantum cryptog­raphy, by �ddling with synchronisationsignals that pass between Alice and Bob.

To start the key exchange, Bob sendsAlice two strong laser pulses separated bya precise interval. She uses these pulses todetermine how to polarise the photonsshe sends back. If Eve intercepts Bob’spulses and changes the separation be­tween them, she can trick Alice into send­ing photons that have slightly di�erentpolarisations from those that were intend­ed. When Eve intercepts these tweakedphotons, she can gather information onthem while keeping the error rate sheintroduces at just below a tolerable level.Her slight tweaks have compensated forthe disturbances she creates.

Neither of these techniques actuallybreaks the fundamental principles onwhich quantum cryptography is based.They simply exploit loopholes introducedwhen it is implemented to practice. Asquantum hackers continue to put systemsthrough their paces, such loopholes willbe closed�as these now have been�andthe systems become more secure. Peoplelike Gregoire Ribordy, the chief executiveof ID Quantique, are therefore encourag­ing the hackers in their activities. Suchhacking is, nevertheless, a useful reminderof an old adage: if something looks toogood to be true, it probably is. 7

Schrödinger’s cat and mouse

Computing: Quantum cryptography is unbreakable in theory. But like anysecurity system, in practice it is only as safe as its weakest link

The Economist Technology Quarterly September 4th 2010 Monitor 9

Jon Nguyen, one of the producers in­volved, says supporters who donate $50receive a tote bag or a print of a self­por­trait by Mr Lynch. They can then followthe project’s progress online. �We’re goingto write about it, blog about it, talk topeople on Facebook,� says Mr Nguyen.Crowdfunding will be used to raise only15­20% of the funding, however; the rest isexpected to come from European grants.

Ted Rall, a syndicated editorial car­toonist and opinion columnist, raised$26,000 from over 200 contributors viaKickstarter for a four­week trip to Afghani­stan. When he �rst visited Afghanistan in2001, he says, there were plenty of mediaoutlets that were willing to fund his trip;indeed, they were having trouble �ndingpeople prepared to go. But this timearound Mr Rall was unable to secure anyfunding in advance. The crowdfundinge�ort, however, brought him to the atten­tion of a publisher, who has contractedhim to write a book about his trip. Donorswho give $100 will receive a signed copy.

Crowdfunding may turn out to be afad, says Cory Doctorow, a bestsellingnovelist and blogger who is experiment­ing with various forms of micropatronage,including selling a bespoke short story for$10,000 to one of his fans. �There will besome people for whom the fact that theyraise money for themselves will be amarketing story,� he says. But crowdfund­ing’s early success at raising sums largeenough to be useful, though not largeenough to replace other sources of fund­ing for creative works, �ts in with a broad­er trend of using technology to bringartists and their audiences closer together.As Mr Chen notes, artists can now asktheir audiences directly for support, andwill often get it. �People are thrilled to beinvolved in the creative process and seesomething come to life,� he says. 7

Putting your money where your mouse is

Crowdfunding: Artists, musicians and writers are using the internet toaggregate lots of small donations to fund their work

WIKIPEDIA, a giant online encyclope­dia compiled by volunteers, is the

product of the aggregation of lots of peo­ple’s spare time. An example of �crowd­sourcing�, it demonstrates that on theinternet, as in the real world, many handsmake light work. Can the same approachbe applied to money as well as time? Thatis the idea behind �crowdfunding�, inwhich lots of small contributions areaggregated online to support artistic orcreative ventures.

As crowdfunding has matured from aseries of one­o� e�orts into somethingreproducible, the money has followed.Millions of dollars, in increments as smallas $5, have poured into e�orts that connectartists, musicians, writers and others withpeople willing to fund their projects.Venture capitalists have also shown aninterest by investing in start­ups thatfacilitate crowdfunding.

There have of course been �tip jars� onweb pages for years, and even big sites likeWikipedia ask for donations. But thisapproach works for a vanishingly smallnumber of sites, and then only in conjunc­tion with other sources of revenue.Crowdfunding is di�erent, say its ad­vocates. �It’s not a tip jar, and that’s whatmakes it sustainable,� says Perry Chen, theboss of Kickstarter, the largest of severalstart­ups that act as matchmakers be­tween donors and projects.

Instead, crowdfunding works by rais­ing money for a well­de�ned projectwithin a speci�ed time limit and with agoal of raising a particular minimum sum(typically around $2,500). If the goal is notmet, no funds are collected. Donors usual­ly get some kind of reward or recognition(a mention in a �lm’s credits, for example),but they do not have any rights in theresulting work. The term crowdfundinghas also been applied to raising money forcompanies or charity, says Kevin Lawton,the author of �CrowdFunding: The Mi­cro­VC Revolution!�. But there are regu­latory limits on micro­investing, so theterm is most widely used to refer tofund­raising for creative work.

Kickstarter lets donors fundart shows, movies, short �lms,dance, graphic novels and the­atre productions. It helped Dias­pora, an open­source social­networking project, raise $200,000during the recent controversy over Face­

book’s privacy policies. IndieGogo sup­ports �lmmakers, writers and game de­signers. Some sites specialise: Sellabandhelps bands raise money to fund profes­sional recording of albums, and Spot.usraises money for journalistic projects.

Yancey Strickler, Kickstarter’s chiefcommunity o�cer, says the �rm acceptsabout half the projects submitted to it.�We turn down projects that are charity,that are just straight business expenses, or‘my dog has cancer’,� he says. Of thosethat are accepted, about half meet theirfunding goals: around 1,600 projects hadbeen funded by July 2010. (Anothercrowdfunding �rm, RocketHub, screensout only projects it deems illegal or in badtaste.) Kickstarter says it has raised over$15m for its users since its launch in 2009.Sellaband says it has raised over $3m andhas contributed to the funding of 50 al­bums since 2006. Crowdfunding �rmstypically take a 5% commission and chargea 3­4% payment­processing fee.

With a little cash from my friendsCrowdfunding has bene�ted from the riseof social networking, which allows evennon­celebrities to accumulate large num­bers of fans or followers online, to whomthey can reach out when a project needsfunding. Successful projects, says MrChen, usually require an �anchor audi­ence� of friends or fans who engage in�micropatronage�, enjoying the associa­tion with a successful project and a perso­nal link with an artist or writer.

The makers of a series of docu­mentaries about David Lynch, a�lmmaker, are using crowd­funding to raise the seedmoney to start workon the project’sthird �lm.

TELECOMS operators naturally prizemobile­phone subscribers who spend

a lot, but some thriftier customers, it turnsout, are actually more valuable. Known as�in�uencers�, these subscribers frequentlypersuade their friends, family and col­leagues to follow them when they switchto a rival operator. The trick, then, is toidentify such trendsetting subscribers andkeep them on board with special discountsand promotions. People at the top of theo�ce or social pecking order often receivequick callbacks, do not worry about call­ing other people late at night and tend toget more calls at times when social eventsare most often organised, such as Friday af­ternoons. In�uential customers also revealtheir clout by making long calls, while thecalls they receive are generally short.

Companies can spot these in�uencers,and work out all sorts of other thingsabout their customers, by crunching vastquantities of calling data with sophisticat­ed �network analysis� software. Instead oflooking at the call records of a single cus­tomer at a time, it looks at customers with­in the context of their social network. Theability to retain customers is particularlyimportant in hyper­competitive markets,such as India. Bharti Airtel, India’s biggestmobile operator, which handles over 3 bil­lion calls a day, has greatly reduced cus­tomer defections by deploying the soft­ware, says Amrita Gangotra, the �rm’sdirector for information technology.

The market for such software is boom­ing. By one estimate there are more than100 programs for network analysis, alsoknown as link analysis or predictive analy­

sis. The raw data used may extend far be­yond phone records to encompass infor­mation available from private andgovernmental entities, and internetsources such as Facebook. IBM, the suppli­er of the system used by Bharti Airtel, saysits annual sales of such software, nowgrowing at double­digit rates, will exceed$15 billion by 2015. In the past �ve yearsIBM has spent more than $11billion buyingmakers of network­analysis software.Gartner, a market­research �rm, ranks thetechnology at number two in its list of stra­tegic business operations meriting signi�­cant investment this year.

Adoption is being driven by the avail­ability of more sources of information,and by the fact that network­analysis soft­ware is becoming easier to use. A decadeago IBM employed experts with PhDs inmathematics to study social networks, ac­cording to Mark Ramsey, the �rm’s head ofbusiness analytics for eastern Europe, theMiddle East and Africa. Today, collegegraduates can operate analysis softwarehandling enormous quantities of data.Bharti Airtel employs only about 100 an­alysts to keep tabs on its 135m subscribers.

Take me to your leadersOf course, companies have long minedtheir data to improve sales and productivi­ty. But broadening data mining to includeanalysis of social networks makes newthings possible. Modelling social relation­ships is akin to creating an �index of pow­er�, says Stephen Borgatti, a network­anal­ysis expert at the University of Kentucky inLexington. In some companies, e­mails are

analysed automatically to help bossesmanage their workers. Employees who areoften asked for advice may be good candi­dates for promotion, for example.

Ellen Joyner of SAS, an analytics �rmbased in Cary, North Carolina, notes thatmore and more �nancial �rms are usingthe software to uncover fraud. The latestversion of SAS’s software identi�es riskyborrowers by examining their social net­works and Internal Revenue Service re­cords, she says. For example, an applicantmay be a bad risk, or even a fraudster, if heplans to launch a type of business whichhas no links to his social network, educa­tion, previous business dealings or travelhistory, which can be pieced together withcredit­card records. Ms Joyner says the soft­ware can also determine if an applicanthas associated with known criminals�perhaps his �ancée has shared an addresswith a parolee. Some insurers reduce pre­miums for banks that protect themselveswith such software.

Last year an American governmentbody called the Recovery Accountabilityand Transparency Board (RATB) began us­ing network­analysis software to look forfraud within the $780 billion �nancial­stimulus programme. In addition to the in­ternet, RATB combs Treasury and law­en­forcement databases to uncover �non­ob­vious relationships�, says Earl Devaney, itschairman. The software works very well,he says. It has triggered about 250 ongoingcriminal investigations and 400 audits.

Joe Biden, America’s vice­president,said in June that such software would beused to prevent fraud within the govern­ment’s Medicaid and Medicare health­careschemes. The Army Criminal Investiga­tion Command already sni�s out procure­ment fraud by scanning text in e­mails. Thesoftware, developed by SRA, an American�rm, can correlate numbers and phraseswritten in nine languages with �nancialdatabases. If a person discusses a particu­lar Department of Defence payment withan individual not o�cially linked to thedeal, SRA’s software may notice it.

Untangling thesocial web

Software: From retailing to counterterrorism, the ability to analyse socialconnections is proving increasingly useful

10 Mining social networks The Economist Technology Quarterly September 4th 2010

1

The Economist Technology Quarterly September 4th 2010 Mining social networks 11

2 The police department of Richmond,Virginia, has pioneered the use of net­work­analysis software to predict crimes.Police o�cers know that crime increases atcertain times, such as on paydays andwhen there is a full moon. But the softwarelets them analyse the social networksaround suspects, such as dealings withemployers, collection agencies and the De­partment of Motor Vehicles. The goal, ac­cording to Stephen Holli�eld, the depart­ment’s technology chief, is to �pulltogether a complete picture� of suspectsand their social circle.

Party plans turn out to be a particularlyuseful part of this picture. Richmond’s po­lice have started monitoring Facebook,MySpace and Twitter messages to deter­mine where the rowdiest festivities willbe. On big party nights, the departmentnow saves about $15,000 on overtime pay,because o�cers are deployed to areas thatthe software deems ripe for criminal activi­ty. Crime has �dramatically� declined as aresult, says Mr Holli�eld. Colin Shearer,vice­president of predictive analytics atSPSS, a division of IBM that makes the soft­ware in question, says it can largely replacepolice o�cers’ reliance on �gut feel�.

Network analysis also has a useful roleto play in counterterrorism. Terror groupsare often decentralised, so mapping theirsocial networks is akin to deciphering �abig spaghetti picture�, says Roy Lindelaufof the Royal Dutch Defence Academy, whodevelops software for intelligence agen­cies in the Netherlands. It turns out that thekey terrorists in a group are often not theleaders, but rather seemingly low­levelpeople, such as drivers and guides, whokeep addresses and phone numbers mem­orised. Such people tend to stand out innetwork models because of their high lev­el of connectedness. To �nd them, analystsmap �structural signatures� such as shortphone calls placed to the same numberjust before and after an attack, which mayindicate that the beginning and end of anoperation has been reported.

The capture of Saddam Hussein in

2003 was due in large part to the mappingof the social networks of his former chauf­feurs, according to Bob Gri�n, the chief ex­ecutive of i2, a British �rm which devel­oped the software used in the manhunt.Senior members of the Iraqi regime weremostly clueless about the whereabouts ofthe former president, says Mr Gri�n, butmodelling the social networks of hischau�eurs who had links to rural propertyeventually led to the discovery of his hide­out, on a farm near his hometown of Tikrit.

From social to societal networksWhere is network analysis headed? Thenext step beyond mapping in�uence be­tween individuals is to map the in�uencesbetween larger segments of society. A fore­casting model developed by Venkatra­mana Subrahmanian of the University ofMaryland does just that. Called SOMA Ter­ror Organization Portal, it analyses a widerange of information about politics, busi­ness and society in Lebanon to predict,with surprising accuracy, rocket attacks bythe country’s Hizbullah militia on Israel.

Attacks tend to increase, for example, asmore money from Islamic charities �owsinto Lebanon. Attacks decrease duringelection years, particularly as more Hizbul­lah members run for o�ce and campaignenergetically. By the middle of 2010 SOMA

was sucking up data from more than 200sources, many of them newspaper web­sites. The number of sources will havemore than doubled by the end of the year.

Once these societal networks of in�u­ence can be accurately mapped, they canbe used to promote the spread of particu­lar ideas�those that support stability anddemocracy, for example. Last year Ameri­ca’s army, which jointly funds SOMA withthe air force, began disbursing about $80min �ve­year research grants for networkanalysis to promote democracy and na­tional security. An authoritarian govern­ment, for instance, may have di�cultiesslowing the spread of a new idea in a cer­tain medium�say, internet chatter about abook that explains how corruption under­

mines job creation. Diplomatic servicescan use this information to help ideasspread. Brian Uzzi of Northwestern Uni­versity in Evanston, Illinois, who advisesintelligence agencies on democracy­pro­motion analytics, says diplomatic servicesare mapping the �tipping point� whenideas go mainstream in spite of govern­ment repression.

SPADAC, a �rm based in McLean, Vir­ginia, performs such analyses on Egyptand other countries in Africa, the MiddleEast and South­East Asia. Clients includethe United States, Mexico and various dip­lomatic services. Riots, bloody electionsand crackdowns, among other things, canbe forecast with improving accuracy bycrunching data on food production, unem­ployment, drug busts, home evictions andslum growth detected in satellite images.Mark Dumas, the head of SPADAC, notesthat societies with longstanding andstrong social and business ties abroadweather change well. In relatively closedcountries, like Egypt, rapid shifts in socialnetworks can trigger upheaval, he says.Last year SPADAC’s revenue reached$19m; this year it will exceed $27m.

Country analyses have great potentialin peacekeeping and counterinsurgencyoperations, according to Kathleen Carleyof Carnegie Mellon University in Pitts­burgh. She is developing a societal modelof Sudan with a team of about 40 research­ers. Foreign aid workers and diplomats fre­quently stumble in Sudan because theyfail to work out which tribal and politicalleaders they should work with, and how.

Ms Carley’s model, known as ORA, an­alyses a decade of data on such things asweather, land and water disputes, cabinetreshu�es, reactions to corruption, courtcases, economic activity and changes intribal geographic maps. Within the infor­mation that emerges are lists of the localsmost likely to co­operate with Westerners,with details of the role each would bestplay. This depth of insight, a demonstra­tion of the power of network analysis to­day, will only grow. 7

12 Inside story The Economist Technology Quarterly September 4th 2010

OVER the course of the next ten years acompany called Geodynamics,

based in Queensland, Australia, is plan­ning to drill as many as 90 wells, each4,500­5,000 metres deep, in the CooperBasin, a desert region in South Australiawith large energy reserves. But the com­pany is not drilling for oil or gas. It is look­ing for an energy source that is far cleanerand more abundant than any fossil fuel:heat emanating from hot rocks deep be­neath the Earth’s surface, a promisingemerging form of geothermal energy.

Conventional geothermal powerexploits naturally occurring pockets ofsteam or hot water, close to the Earth’ssurface, to generate electricity. (Heat fromthe water is used to boil a �uid and drive asteam turbine connected to a generator.)Because such conditions are rare, the

majority of today’s geothermal powerplants are located in rift zones or volcani­cally active parts of the world. In Iceland,around one­quarter of the country’selectricity is produced by geothermalpower stations; at the Svartsengi powerstation, the naturally occurring hot wateralso �ows into a lagoon, which is a pop­ular (and photogenic) bathing spot.

Geothermal power stations can also befound along the �Ring of Fire� around thePaci�c, in Indonesia, the Philippines andon America’s west coast. Conventionalgeothermal power stations worldwidehave a total capacity of 10.7 gigawatts(GW) and will generate 67.2 gigawatthours (GWh) of energy this year�enoughto supply power to more than 52.5m peo­ple in 24 countries, according to America’sGeothermal Energy Association.

Engineered geothermal systems (EGS)are based on a related principle, but theywork even in parts of the world that arenot volcanically active, by drilling thou­sands of metres underground to mimicthe design of natural steam or hot­waterreservoirs. Wells are bored and pathwaysare created inside hot rocks, into whichcold water is injected. The water heats upas it circulates and is then brought back tothe surface, where the heat is extracted togenerate electricity. Because the Earth getshotter the deeper you drill, EGS couldexpand the reach of geothermal powerenormously and provide access to a virtu­ally inexhaustible energy resource.

�The beauty of the concept is that if itworks, it can work anywhere in theworld,� says Subir Sanyal, president ofGeothermEx, a consultancy based inCalifornia. According to �The Future ofGeothermal Energy�, a report issued bythe Massachusetts Institute of Technology(MIT) in 2007, the thermal energy avail­able in America in rocks 3­10km (1.9­6.2miles) beneath the Earth’s surface is nearly140,000 times greater than its annualenergy consumption. Conservative esti­mates suggest just 2% of that energy couldbe tapped by EGS in practice, but even thatwould be far more than is needed to sup­ply all of America’s electricity. Tapping itwill, however, require both technical and

Hot rocks and high hopes

Geothermal power: Deriving energy from subterranean heat is no longerlimited to volcanic regions. By drilling deep wells into the ground, it can bemade to work almost anywhere. Just watch out for the earthquakes

1

The Economist Technology Quarterly September 4th 2010 Inside story 13

economic hurdles to be overcome.At the moment only a few EGS plants

exist worldwide, including a pilot plant inSoultz, France, and a small commercialplant in Landau, Germany. But Geody­namics and other companies around theworld are hoping to change that. Over thenext decade Geodynamics plans to buildten 50 megawatt (MW) power stations inCooper Basin, and that may just be thebeginning. According to Doone Wyborn,the company’s chief scientist, the area’sresources could support hundreds ofpower stations with a total generatingcapacity of up to 12.5GW�more than allthe geothermal power stations now oper­ating worldwide. There are also plans fornew EGS projects in America, Britain,France and Germany. Those in the �eldhave high hopes for future expansion: theInternational Geothermal Associationpredicts that there will be 160GW of geo­thermal capacity installed worldwide by2050, about half of which will be EGS.

Like other forms of renewable energy,geothermal power produces little or nocarbon dioxide. But unlike other forms ofrenewable energy, such as solar or windpower, it has the further advantage that itis not intermittent, but can provide steady,predictable baseload electricity, all dayand all night. This makes it particularlyappealing to utilities.

These bene�ts, in combination withgrowing electricity use worldwide, con­cerns about limited supplies of fossil fuels,and e�orts to reduce carbon­dioxideemissions and prevent climate change,have prompted governments and in­vestors to pour money into this emergingtechnology. Google, for example, hasinvested more than $10m in two EGS

companies in California, Potter Drillingand AltaRock Energy. Meanwhile Ameri­ca’s Department of Energy has an­nounced up to $338m in stimulus fundsfor 123 geothermal projects, with nearly$133m earmarked for EGS research.

Australia’s e�orts are probably themost ambitious. Primary Industries andResources SA (PIRSA), an Australian gov­ernment agency, projects that between2002 and 2014, investments in Australiangeothermal projects (including more than$250m in government grants) could reach$2.7 billion�with roughly 72% of that�gure going toward EGS projects. Morethan 50 companies exploring geothermalprojects in Australia have taken out over400 licences for areas covering nearly500,000 square kilometres�a combinedarea roughly the size of Spain.

The �rst conventional geothermalpower station, powered by hot springsnear Larderello, Italy, began generatingelectricity in the early 1900s. It was severaldecades before scientists thought of de­signing systems that could work any­where. In the early 1970s the concept of�hot dry rock� (HDR) geothermal poweremerged at Los Alamos National Labora­tory, New Mexico, and researchers fromthe lab conducted the �rst tests at nearbyFenton Hill. This led to similar projects inBritain, Japan, France and elsewhere.

Drilling into the pastHDR was based on the idea that by drillinginto hot, dry rocks and fracturing them itwould be possible to mimic a naturalwater­based geothermal system. Coldwater is injected into one well and �owsthrough the reservoir’s cracks and path­ways, absorbing heat. The hot water isthen brought back to the surface through aproduction well, where it heats up a sec­ondary working �uid with a lower boilingpoint. The vapour from that �uid thenspins a turbine to generate electricity,while the water is reinjected into the well.

These early experiments led to animportant lesson: that creating a produc­tive, permeable reservoir is best done byworking with the site’s existing geology,and cracking open or widening existingfractures, rather than trying to createentirely new ones. Barry Goldstein, direc­tor of petroleum and geothermal at PIRSA,says it is important to choose an area withthe potential to support a su�cient rate ofwater �ow to make the project economic.

In fact, EGS projects can be operatedunder a range of geological conditions,from HDR to hot fractured wet rocks, all ofwhich need di�erent levels of enhance­ment (ie, di�erent amounts of drilling,fracturing and water injection). �There’s acontinuum,� says Karl Gawell, executivedirector of America’s Geothermal EnergyAssociation. It extends all the way toconventional geothermal systems, someof which are bene�ting from EGS research.The power stations at The Geysers inNorthern California, the world’s largestdeveloped geothermal �eld, reinject waterinto their reservoirs to restore steam re­serves and boost output�a techniqueborrowed from EGS, says Mr Gawell.

Costs for conventional geothermalprojects vary widely depending on loca­tion, temperature and drilling depth,among other things. Geothermal plantshave no fuel costs, but upfront costs arehigh, and verifying the potential of a site,as with an oil well, can take a long time,making �nancing di�cult in the currenteconomic climate. The most economicallyviable projects, as you might expect, arethose that exploit high temperatures atshallow depths. A typical American geo­thermal power station produces electric­ity at a cost of around $0.10/kWh. Thatmakes geothermal power competitivewith many other technologies, especiallywith added �nancial incentives such asAmerica’s production tax­credit for re­newable­energy projects, currently about$0.02/kWh. (Producing electricity fromcoal or gas also costs around $0.10/kWh.)

The same cannot be said for EGS, at

�The beauty of the concept is that if it works, it canwork anywhere in the world.�

2

1

14 Inside story The Economist Technology Quarterly September 4th 2010

least for the foreseeable future. As you godeeper, temperatures go up�but so docosts. The equipment on the surface costsabout the same for EGS as it does for con­ventional geothermal power, but thedrilling costs can be twice as much ormore for EGS. Dr Wyborn estimates thatelectricity from EGS could initially cost anadditional $0.09/kWh over conventionalgeothermal, or about $0.19/kWh. Thatwould make EGS economic only in placeswith strong �nancial incentives, such asGermany, where operators of renewable­energy projects receive generous subsidiesin the form of feed­in tari�s�currently$0.31/kWh for power from EGS.

Technological improvements, such ascheaper and better methods for drilling,creating reservoirs and improving water­�ow rates, could cut the cost of EGS. Wellproductivity is especially important. �Youwant to get as much extracted energy aspossible for that set of wells you’vedrilled, to maximise the return on yourinvestment,� explains Je�erson Tester,associate director of the Cornell Centre fora Sustainable Future and lead author ofthe MIT report. So far most EGS projectshave achieved �ow rates of only around25 litres per second, far short of the 50­100litres per second that is required to operategeothermal projects pro�tably. Both Alta­Rock and Geodynamics are working oncreating more fractures per well, whichthey hope will increase both the �ow rateand heat absorption from the rocks.

But unexpected problems can pop up.In April 2009 Geodynamics was ready tocommission a pilot plant when the steelcasing of a well cracked, causing uncon­trolled �ow of water out of the well. Anindependent investigation determinedthat the problem could be avoided in thefuture by choosing a di�erent type of wellcasing. Geodynamics has announced thatit will drill two new wells. Its 1MW pilotplant is now scheduled to come online inearly 2012, followed by a 25MW commer­cial demonstration plant three years later.

AltaRock also encountered drillingproblems in 2009, when it made threeattempts to redrill a well for a demonstra­tion project at The Geysers. It eventuallyabandoned that e�ort after the drillingassembly repeatedly got stuck due to thehole collapsing. Along with its partner,Davenport Newberry, it now plans todemonstrate its technology at another sitenear Bend, Oregon, a project for which itwas awarded $21.5m in stimulus funds byAmerica’s Department of Energy.

Much of the drilling equipment for

geothermal wells comes from the oil­and­gas industry, but Bob Potter, a member ofthe original HDR team at Los Alamos, istrying something di�erent. His �rm, PotterDrilling, is pursuing a process called spall­ation, which uses superheated steam.When it comes into contact with rock,crystalline grains in the rock expand andcreate tiny fractures, causing small parti­cles, called spalls, to break o�. In e�ect, it isa drill that melts rock, says Mr Potter’s son,Jared, who is the chief executive of PotterDrilling. Spallation can get through rockmore quickly than conventional drilling,and the use of steam means there are nocostly drill bits that need replacing.

Little earthquakesPerhaps the biggest hurdle that will pre­vent EGS from spreading is its propensityto cause noticeable earthquakes thatfrighten people. Earthquakes are in fact arequirement for the technology to work.In order to prop open or enlarge existingcracks and fractures, water is injected intoboreholes at high pressure, causing smalltremors. �There’s no doubt that what youdo when you fracture rock causes seismic­ity,� says Susan Petty, president and chieftechnology o�cer of AltaRock. �But the

goal is to have those events be so tiny thatpeople can’t feel them.� Most earthquakescreated by EGS are indeed too small to befelt, but a few have caused damage toproperty. One project in Basel, Switzer­land, was shut down because of a 3.4­magnitude earthquake in December 2006that scared residents and cracked build­ings. Earthquakes of a similar magnitudehave also been reported from projects inAustralia, Germany and France.

But man­made earthquakes are notunique to EGS; they also occur as a resultof oil­and­gas drilling, and damming andmining operations. The question is wheth­er they can be controlled. Ernie Majer, aseismologist and deputy director of theEarth Sciences Division at LawrenceBerkeley National Laboratory, who isworking on re�ning EGS seismicity guide­lines for America’s Department of Energy,believes they can. �With proper study andimplementation, you can guarantee thatthere won’t be big ones,� says Dr Majer,who sees small quakes as a nuisancerather than a danger. Still, many in theindustry agree that EGS should be devel­oped in remote areas �rst, rather than indensely populated cities such as Basel.

And the risks associated with EGS mustbe balanced against the drawbacks ofother energy technologies, such as fossilfuels, which produce carbon­dioxideemissions and occasional oil spills, andnuclear power, which produces radioac­tive waste. Wind power, meanwhile, iscriticised for causing noise pollution,killing birds and despoiling landscapes.The real question, in the end, is whatpeople are ready to put up with in returnfor a secure energy supply. �It’s a trade­o�,� says Dr Majer. �You have bene�ts andhazards. There’s no perfect technology.�

Whether EGS can overcome the obsta­cles it currently faces, and go on to play animportant role in the world’s renewable­energy portfolio, should become clear inthe next decade. �The well failure has setus back,� acknowledges Dr Wyborn ofGeodynamics. But he is certainly notgiving up. According to the MIT report, the�rst 100MW of installed EGS capacityshould be the most di�cult and costly toachieve, but after that it should get easierand cheaper. Scarcer and more expensiveoil would certainly help. �There are thou­sands of wells being drilled for oil acrossthe world every year,� says Dr Wyborn. �Iimagine that in a couple of decades all ofthose drilling rigs that are now redundant,because we’ve run out of oil, will be drill­ing geothermal wells instead.� 7

�Most earthquakes created by EGS are too small tobe felt, but a few have caused damage to property.�

Showing the way in Landau

2

IT IS known as �phantom limb syn­drome� or �phantom pain�. But this

strange phenomenon feels all too real tothe people it a�ects, and can be agonis­ingly painful. Amputees and people whohave become paralysed may still �feel� amissing limb or a part of their body, eventhough it is no longer connected to theirnervous system. Yet such sensations o�ercon�rmation that even when a limb hasbeen severed or cut o� from the nervoussystem, the nerves that once serviced it re­main alive and well. Doctors are now �nd­ing ways to put these nerves to good use,by rewiring them to control prostheticlimbs or reanimate paralysed limbs.

Moreover, rewiring the nervous systemshould allow amputees to gain a sense of�embodiment� of a prosthetic. That is, bycontrolling and sensing the prosthetic us­ing the same neural pathways and parts ofthe brain that once governed the real limb,the prosthetic can be made to feel and actlike a genuine extension of the user’s body.And by stimulating the nerves in the legsor arms of paralysed patients�nerves thathave been cut o� from the central ner­vous system�it is possible to createco­ordinated movement of greatsubtlety. For example, the handsof paralysed patients have beenstimulated to enable them tograsp and turn door knobs.And with careful control andco­ordination of the musclegroups in their legs, patients caneven rise from their wheelchairs andtake steps.

Prosthetic limbs are becoming increas­ingly sophisticated, but it can be very di�­cult to control them in a natural way, saysPaul Marasco, a biomedical engineer at theLouis Stokes Cleveland Department of Vet­eran A�airs Medical Center, in Ohio. Forexample, patients control some motoriseddevices by �exing muscles in their remain­ing stump, shoulder or chest. These musclemovements are detected by electromyog­raphy (EMG) sensors on the skin, and the

signals are translated into movements bythe prosthetic.

This approach can provide incrediblecontrol over an arti�cial arm, but patientsoften prefer to use simpler, mechanicalprosthetics. For one thing, such devices al­low the patient to sense the movement ofthe arm through a system of cables whichare used to control the device, usually byattaching them to the opposite shoulder.So even when their eyes are closed theycan get a sense of whether an arti�cial armis extended, or if there is resistance to agrasping motion, making the limb feelless detached and unnaturalthan an EMG device. Anoth­er problem with EMG

prosthetics is that patientsliterally have to retraintheir brains to makenew associations be­tween muscle move­ments and their out­

comes�a shoulder �ex could become agrasp motion, for example, while a twitchof pectoral muscle in the chest may extendthe arti�cial arm.

But there is a way to overcome boththese di�culties with motorised limbs, us­ing a technique called �targeted reinnerva­tion�. Pioneered by Todd Kuiken, directorof the Neural Engineering Centre for Arti�­cial Limbs at the Rehabilitation Institute ofChicago, along with his colleagues AimeeSchultz, Blair Lock and Dr Marasco (whowas formerly at NECAL), this involves re­routing the nerves that would have origi­nally controlled and sensed the missinglimb and connecting them instead to otherparts of the body. By rewiring a missingarm’s motor nerves to muscles in the re­maining stump, shoulder or chest, for ex­ample, and rewiring the arm’s sensorynerves to the skin in these regions, a chan­nel is opened to the part of the brain thatonce controlled the missing limb.

It is a strange and slightly ghoulish idea,because it means that if a patient tries to�ex his missing �nger, for example, a mus­cle in another part of his body (which is

now connected to the nerves thatused to control the �nger) con­

tracts instead. �When the am­putee wants to open or closetheir hand, these muscles

twitch,� says Dr Marasco.EMG sensors detectthese signals and trans­late them into controlsignals that cause the

mechanical hand to openand close. The patient can

then open and close hisprosthetic hand simply by

trying to move the �ngers thatare no longer there.

Biomedicine: Doctors are rerouting nerves to give patients more naturalcontrol of prosthetic arms and bring paralysed limbs back to life

How to rewire the nervous system

Natural control of an arti�cial arm

The Economist Technology Quarterly September 4th 2010 Rewiring nerves 15

1

16 Rewiring nerves The Economist Technology Quarterly September 4th 2010

2 The sensory side of things works in asimilar fashion, but instead of reconnect­ing nerves to di�erent muscles, they are in­stead rewired to the skin’s underlying sen­sory systems. �They hook themselves upto the receptors in the skin,� says Dr Ma­rasco. So when the reinnervated skin (onthe chest or shoulder, perhaps) is touched,it registers to the patient as a sensation inthe missing limb. �They have very distinctsensations that they can feel: vibrations,temperature and pressure,� he says. Thismeans sensors in a prosthetic limb could,in theory, stimulate the reinnervated skinto cause realistic sensations.

The �rst example of sensory reinnerva­tion actually happened by accident back in2003, when the surgery was �rst per­formed, says Ms Schultz. Several patientsunderwent surgery to reroute their motornerves to di�erent muscles, but in the pro­cess the sensory nerves reinnervatedthemselves too, attaching themselves tothe skin receptors. This was discoveredwhen one of the patients was having alco­hol dabbed on his chest, and remarkedthat he could feel it in his missing limb.�Everyone was stunned,� says Ms Schultz.

This opened the door to the possibilityof providing feedback to patients. But en­abling patients to sense things will meandeveloping sensors to go on prostheticlimbs, and corresponding devices to stim­ulate the reinnervated skin. To this end DrMarasco and his colleagues have beenworking with Ed Colgate, a haptics expertat Northwestern University in Evanston, Il­linois, and Kinea Design, a biomechatron­ics �rm that is also based in Evanston, todevelop tactile devices that would attachto the amputee’s skin.

Multisensory devices have alreadybeen developed and tested on reinnervat­ed amputees, relaying sensations of con­tact, pressure, vibration, shear force andtemperature. These prototypes use me­chanical stimulation to create these sensa­tions, and electrical Peltier devices to gen­erate di�erent temperatures. Patients feelthe sensations in their missing limbs, andone patient was even able to discriminateby touch between sandpaper, Te�on andthe bumpy texture of a computer ribboncable using arti�cial touch sensors.

Although there are far fewer sensory re­ceptors in the skin of the chest, shoulderand upper arm than there are in the hand,the tactile acuity of these regions’ skin ap­pears to increase when reinnervated, saysDr Marasco. It’s not entirely clear why, butit probably comes down to how the brainperceives these signals. �It appears that

there is much more processing power de­voted to these nerves,� he says. Just howmuch tactile acuity can be achieved re­mains to be seen. �This is unchartered ter­ritory,� says Dr Marasco.

When it comes to rewiring the limbs ofparalysed patients in order to reanimatethem, however, a very di�erent approachis required. Researchers must instead targetthe deeper nerves that control movement.The idea of electrically stimulating themuscles of paralysed people is not new.Many di�erent techniques have emergedover the years, most of which involve ap­plying small jolts of electricity externallyto muscles, via electrodes on the skin, tomake them contract. Besides o�ering ameans of exercising the muscles, and sopreventing atrophy, the hope is that this�functional electrical stimulation� mightalso help restore mobility to patients.

Reanimate thisBut external stimulation is less than ideal,says Dustin Tyler, a biomedical engineer atCase Western Reserve University in Cleve­land, Ohio. The muscles in the legs are solarge that the whole muscle does not con­tract, he says. So he and his colleagues havebeen looking at ways to activate thesemuscles by tapping into the femoral nerve,in the groin. �By moving back to the nerveyou get the whole muscle,� he says. Thefemoral nerve is divided into several doz­en separate bundles of nerves, called fasci­cles, each of which contains hundreds ifnot thousands of individual nerves. Di�er­ent fascicles lead to di�erent muscles, sostimulating groups of fascicles at di�erenttimes and by di�erent degrees should en­able co­ordinated leg movements.

Stimulating fascicles can be done bywrapping so­called cu� electrodes around

them. The problem is that within the femo­ral nerve all the fascicles are themselvesbundled together, making it di�cult tostimulate them individually. But Dr Tylerhas developed a cu� electrode that allowsdiscrimination by gently �attening thefemoral nerve without damaging it. Mostelectrodes are designed around the as­sumption that nerves are round, he says.But �attening them out makes the individ­ual fascicles accessible, like the parallelwires in old computer cables, says Dr Tyler.

And by controlling the intensity andduration of the electrical pulses applied, itis possible to specify just how much of thetissue is contracted. Besides getting pa­tients to stand and take steps, albeit tremu­lous ones, Dr Tyler and his team have usedan early version of this technique in a pa­tient’s arms to enable grasping move­ments. They hope to carry out the �rst per­manent implantation of their newelectrodes in November.

But to create full mobility, what is notclear is how a patient might control such asystem. Some sort of joystick interfacecould be used to allow paraplegic people,who still have control of their upper body,to activate certain patterns of co­ordinatedmovement, such as standing, walking andsitting. But in the long term the ultimategoal would be to place electrodes in themotor cortex of the brain.

That is some way o�. But targeted rein­nervation is now available as a treatment.More than 40 people around the worldhave undergone the procedure so far. Eventhough patients can currently use it only tocontrol their arti�cial limbs, sensory feed­back is coming. And these are merely the�rst examples of what can be done by re­wiring the nervous system, and linkingnerves to electronic and robotic devices. 7

Getting touchy­feely with haptic sensors

�Targeted reinnervation involves rerouting nervesthat originally controlled the missing limb.�

AROUND 1.5 billion people, or more thana �fth of the world’s population, have

no access to electricity, and a billion morehave only an unreliable and intermittentsupply. Of the people without electricity,85% live in rural areas or on the fringes ofcities. Extending energy grids into these ar­eas is expensive: the United Nations esti­mates that an average of $35 billion­40 bil­lion a year needs to be invested until 2030so everyone on the planet can cook, heatand light their premises, and have energyfor productive uses such as schooling. Oncurrent trends, however, the number of�energy poor� people will barely budge,and 16% of the world’s population will stillhave no electricity by 2030, according tothe International Energy Agency.

But why wait for top­down solutions?Providing energy in a bottom­up way in­stead has a lot to recommend it. There is noneed to wait for politicians or utilities toact. The technology in question, from solarpanels to low­energy light­emitting diodes(LEDs), is rapidly falling in price. Local, bot­tom­up systems may be more sustainableand produce fewer carbon emissions thancentralised schemes. In the rich world, infact, the trend is towards a more �exible

system of distributed, sustainable powersources. The developing world has an op­portunity to leapfrog the centralised mod­el, just as it leapfrogged �xed­line telecomsand went straight to mobile phones.

But just as the spread of mobile phoneswas helped along by new business mod­els, such as pre­paid airtime cards and vil­lage �telephone ladies�, new approachesare now needed. �We need to reinventhow energy is delivered,� says Simon Des­jardins, who manages a programme at theShell Foundation that invests in for­pro�tways to deliver energy to the poor. �Com­panies need to come up with innovativebusiness models and technology.� Fortu­nately, lots of people are doing just that.

Let there be lightStart with lighting, which prompted the es­tablishment of the �rst electrical utilities inthe rich world. At the �Lighting Africa� con­ference in Nairobi in May, a World Bankproject to encourage private­sector sol­utions for the poor, 50 lighting �rms dis­played their wares, up from just a handfullast year. This illustrates both the growinginterest in bottom­up solutions and fallingprices. Prices of solar cells have also fallen,

so that the cost per kilowatt is half what itwas a decade ago. Solar cells can be used topower low­energy LEDs, which are bothenergy­e�cient and cheap: the cost of a setof LEDs to light a home has fallen by half inthe past decade, and is now below $25.

�This could eliminate kerosene lightingin the next ten years, the way cellphonestook o� in about 13 years,� says RichendaVan Leeuwen of the Energy Access Initia­tive at the UN Foundation in Washington,DC. That would have a number of bene�ts:families in the developing world mayspend as much as 30% of their income onkerosene, and kerosene lighting causes in­door air pollution and �res.

But such systems are still beyond thereach of the very poorest. �There are hun­dreds of millions who can a�ord clean en­ergy, but there is still a barrier for the bil­lions who cannot,� says Sam Goldman,the chief executive of D.light. His �rm hasdeveloped a range of solar­powered sys­tems that can provide up to 12 hours oflight after charging in sunlight for one day.D.light’s most basic solar lantern costs $10.But the price would have to fall below $5 tomake it universally a�ordable, accordingto a study by the International FinanceCorporation, an arm of the World Bank. Sothere is scope for further improvement.

It is not just new technology that isneeded, but new models. Much of the fer­ment in bottom­up energy entrepreneur­ialism is focusing on South Asia, where570m people in India, Pakistan and Bangla­desh, mostly in rural areas, have no accessto electricity, according to the InternationalEnergy Agency. One idea is to use locallyavailable biomass as a feedstock to gener­ate power for a village­level �micro­grid�.Husk Power Systems, an Indian �rm, usessecond­world­war­era diesel generators�tted with biomass gasi�ers that can userice husks, which are otherwise left to rot,as a feedstock. Wires are strung on cheap,easy­to­repair bamboo poles to providepower to around 600 families for each gen­erator. Co­founded three years ago by a lo­cal electrical engineer, Gyanesh Pandey,Husk has established �ve mini­grids in Bi­har, India’s poorest state, where rice is a sta­ple crop. It hopes to extend its coverage to50 mini­grids during 2010. Consumers paydoor­to­door collectors upfront for power,and Husk collects a 30% government subsi­dy for construction costs. Its pilot plantswere pro�table within six months, so itsmodel is sustainable.

Emergence BioEnergy takes this ap­proach a step farther. Its aim is to providemany entrepreneurial opportunities

Technology and development: A growing number of initiatives are promotingbottom­up ways to deliver energy to the world’s poor

Power to the people

The Economist Technology Quarterly September 4th 2010 Energy in the developing world 17

1

around energy production, says IqbalQuadir, the �rm’s founder, who is also di­rector of the Legatum Centre for Develop­ment & Entrepreneurship at the Massachu­setts Institute of Technology (MIT). A cattlefarmer in a small village in Bangladeshmight, for example, operate a one­kilowattgenerator in his hut, powered by methanefrom cow manure stored in his basement.He can then sell surplus electricity to hisneighbours and use the waste heat fromthe generator to run a refrigerator to chillmilk. This preserves milk that otherwisemight be spoilt, o�ers new sources of in­come to the farmer (selling power and oth­er services, such as charging mobilephones or running an internet kiosk) andprovides power to others in his village.

The farmer funds all this with a micro�­nance loan. It is no coincidence that this is asimilar model to the �telephone lady�scheme, pioneered in Bangladesh a fewyears ago, in which women use micro­loans to buy mobile phones and then sellaccess, by the call, to other villagers; MrQuadir helped establish Grameenphone,now the largest mobile operator in Bangla­desh, and hopes to repeat its success in en­ergy. After a pilot project in two villages,Emergence BioEnergy plans a broader roll­out in 2011 in conjunction with BRAC, a

giant micro�nance and devel­opment NGO.

Another project, in India,aims to convert women fromgathering wood, which de­nudes forests, to using canistersof lique�ed petroleum gas(LPG). India’s four state­ownedregional power companies, in­cluding Bharat Petroleum Cor­poration, will build a nationalnetwork of thousands of LPG­powered community kitchens.Local entrepreneurs will thenprovide the LPG and charge vil­lagers to use the kitchens in 15­minute increments.

Harish Hande, managing di­rector of Selco Solar, a socialenterprise in India that pro­motes the adoption of new en­ergy technologies, says the im­portant thing is not so much todeliver energy to the poor, butto provide new ways to gener­ate income. His �rm has de­vised a solar­powered sewingmachine, for example. Lastyear Mr Hande started an incu­bation lab in rural Karnataka,in southern India, to bring to­

gether local customers and engineering in­terns from MIT, Stanford and Imperial Col­lege, London. The lab is currently piloting ahybrid banana dryer that runs on biomassduring wet spells and sunlight on dry daysto make packets of dried banana�so thatfarmers no longer have to rely on sellingtheir crop immediately.

Making it payEven when new technology and modelsare available, the logistics of rolling themout can be daunting. The two big chal­lenges are providing the upfront invest­ment for energy schemes, and buildingand maintaining the necessary distribu­tion systems to enable them to reach su�­cient scale. At the moment, most schemesare funded by angel investors, foundationsand social venture­capital funds. There is avigorous debate about whether the privatesector on its own can make these modelswork as technology improves, or whethernon­pro�t groups are needed to �ll thegaps in funding and distribution.

Micro�nance institutions may seemthe natural �nancial partners to help thepoor pay for energy systems, since they arethe only organisations with millions ofpoor customers. But teething problems areformidable and success stories are few,

says Patrick Maloney of the LemelsonFoundation, which invests in clean­energytechnologies for the poor. A telephonelady could buy a mobile phone for a rela­tively small sum, and would immediatelyhave a source of income with which to re­pay the loan. Although a household thatbuys a solar lamp saves money on kero­sene, the investment takes several monthsto pay for itself, and there is no actual in­come from the lamp. For bigger energy pro­jects, such as micro­generators, the loan re­quired is much larger, and therefore riskier,than the loan for a mobile phone.

Moreover, micro�nance institutionsmay lack the funds to identify reliable en­ergy suppliers, educate loan o�cers aboutclean­energy technologies and build a sup­port network for energy schemes. Oneway to solve this problem, being pursuedby MicroEnergy Credits, a social enter­prise, is to plug micro�nance institutionsinto carbon markets. Projects can then befunded by selling carbon credits when amicro�nance customer switches from ker­osene to solar lighting, for example.

Distribution is also a problem, particu­larly in Africa and South Asia, where themajority of the world’s energy­poor live.Infrastructure and supply chains are pooror non­existent, particularly in rural areas.Recruiting and training a sales force, andeducating consumers of the bene�ts ofswitching away from wood or kerosene,must be paid for somehow. Social enter­prises are innovating in this area, too. SolarAid, a non­pro�t group, specialises in set­ting up microfranchises to identify andtrain entrepreneurs. The organisationworks with local authorities to identify po­tential entrepreneurs, who must gather sig­natures from their local community�pro­viding both the endorsement of theirneighbours and a future customer base.They then undergo �ve days of trainingwith an exam at the end. Solar Aid is alsotesting a kiosk­based system to help entre­preneurs distribute LED lighting in the Kib­era district of the Kenyan capital, Nairobi.

Some hurdles to bottom­up energy pro­jects are more easily addressed. In particu­lar, high import duties on clean­energyproducts in many developing countries,notably in Africa, hamper their adoptionby the poor. Ethiopia, for example, im­poses a 100% duty on imports of solar pro­ducts, while Malawi charges a 47.5% tax onLED lighting systems. Such taxes are some­times defended on the basis that only therich can a�ord fancy technology. But thesame was said about mobile phones a de­cade ago�and look at them now. 7

Generating electricity from rice husks

18 Energy in the developing world The Economist Technology Quarterly September 4th 2010

2

The Economist Technology Quarterly September 4th 2010 Brain scan 19

FROM �Wikinomics� to �CognitiveSurplus� to �Crowdsourcing�, there is

no shortage of books lauding the �Web2.0� era and celebrating the online col­laboration, interaction and sharing that itmakes possible. Today anyone can pub­lish a blog or put a video on YouTube, andthousands of online volunteers can collec­tively produce an operating system likeLinux or an encyclopedia like Wikipedia.Isn’t that great?

No, says Jaron Lanier, a technologist,musician and polymath who is bestknown for his pioneering work in the �eldof virtual reality. His book, �You Are Not AGadget: A Manifesto�, published earlierthis year, is a provocative attack on manyof the internet’s sacred cows. Mr Lanierlays into the Web 2.0 culture, arguing thatwhat passes for creativity today is reallyjust endlessly rehashed content and thatthe �fake friendship� of social networks�is just bait laid by the lords of the cloudsto lure hypothetical advertisers�. For MrLanier there is no wisdom of crowds, onlya cruel mob. �Anonymous blog com­ments, vapid video pranks and light­weight mash­ups may seem trivial andharmless,� he writes, �but as a whole, thiswidespread practice of fragmentary, im­personal communication has demeanedpersonal interaction.�

If this criticism of Google, Facebook,Twitter and Wikipedia had come from anoutsider�a dyed­in­the­wool techno­phobe�then nobody would have paidmuch attention. But Mr Lanier’s denuncia­tion of internet groupthink as �digitalMaoism� carries more weight because ofhis career at technology’s cutting edge.

Mr Lanier was born in New York Cityin 1960, but his bohemian parents uppedsticks and moved to Mesilla, New Mexico,when he was young. His mother died in acar accident when he was nine and hewas raised by his father in di�cult �­nancial circumstances. Their house wasrepossessed and for a long time the La­niers lived in tents. Eventually they built a�strange house� that the young Mr Lanierhelped design. It took seven years to com­plete and gave him a taste for creatingfantastical environments. He also collect­ed bits of discarded technology from thenearby White Sands missile base and built

contraptions, including a haunted �house�that he describes as �too strange andterrifying for even a single kid to visit�.

But he had few friends of his own agein any case. The young Mr Lanier was farmore comfortable in the company ofadults and had a talent for befriendingclever people who acted as mentors.Clyde Tombaugh, the head of optics at themissile range and the discoverer of thedwarf planet Pluto, took him under hiswing and helped him build telescopesfrom cast­o� technology. Mr Lanier alsohad access to the well­equipped computercentre at White Sands.

Mesilla was a rough town back then,and not an easy place for a sensitive, in­telligent young man to �t in. �I was veryintensely lonely and very disconnected,�says Mr Lanier. �It left me with a profoundawkwardness that I haven’t fully over­come.� His sense of alienation gave himan abiding fascination with what it meansto be connected to other people. �Whatdoes it mean to not be alone?� asks MrLanier. �I’ve approached that questionthrough music, technology, writing andother means.�

After dropping out of both high schooland art college, Mr Lanier tried his hand atgoat­herding and midwifery. His anti­establishment tendencies prevented himfrom doing the obvious thing and taking ajob at the White Sands missile range.Instead, he hopped into a bullet­riddledDodge Dart and drove to California inpursuit of a girl, characteristically hangingout with the Nobel prize­winning phys­icists, Richard Feynman and Murray Gell­Mann, along the way.

It all started with AtariCalifornia in the early 1980s proved to bemore accepting of bohemian inventorsthan Mesilla, New Mexico, and Mr Lanierfelt at home. Via a circuitous route, hewent from busking on the streets of SantaCruz to working for Atari, a video­games�rm, as a researcher. At his new home inSausalito, he also busied himself devel­oping a video game called Moondust,which he demonstrated to his new em­ployers. Tom Zimmerman, a co­worker,recalls the game being a �visual orgy�where the player has to move a cloudaround. Move the cloud fast enough, andyou are rewarded with a surge of light andsound. �Everyone else was writingshoot­’em­up games, so I knew this guywas di�erent,� says Mr Zimmerman.

In 1984 Atari’s fortunes changed for theworse and Mr Lanier and Mr Zimmerman

The virtual curmudgeon

Jaron Lanier, a pioneer ofvirtual­reality technology, has morerecently become an outspoken criticof online social media

1

found themselves without a job. For MrLanier, unemployment simply meantmore time to work on pet projects. Havingmade some money from selling Moon­dust, he continued to work on a �post­symbolic� visual programming language(VPL) for manipulating objects in three­dimensional space. What was missingwas an interface device to move the ob­jects about. As it happened, Mr Zimmer­man had developed a �data glove� someyears before, with the intention of using itto mould virtual pottery. But program­ming was not Mr Zimmerman’s forte.�Jaron was looking for a more interestinginterface technology for his programminglanguage,� says Mr Zimmerman. �Youcould characterise it as software lookingfor hardware. I was on the other side. I hadthis rich interface device, but no applica­tion for it.�

In 1985 Mr Zimmerman, along withChuck Blanchard, Young Harvill and SteveBryson, joined Mr Lanier’s recentlyformed company, VPL Research, thatwould pioneer virtual reality. �Jaron waspicking up these fantastical ideas of virtu­al creatures you’d inhabit through yourbody, and people would be interacting inthis virtual world, and I was doing theinterface hardware, building these glovesfor him,� says Mr Zimmerman. �We werebasically creating this industry, althoughwe didn’t know it at the time.� VPL createdthe �rst software avatars, the �rst multi­person virtual reality, the �rst commercialVR equipment and its �rst application, tosurgical simulation. VPL thrived for awhile, selling 1.3m basic data gloves toMattel, a toymaker, and a smaller numberof high­end gloves to NASA and IBM. Butthe �rm �led for bankruptcy in 1990, andall its patents were sold to Sun Microsys­tems, then a rising computer­maker.

Looking back, 20 years later, it is hard tosay whether Mr Lanier’s VR dream hasbeen realised or not. In a sense, it hasbecome commonplace in the form ofvideo games, which routinely featureavatars and have controllers that sensemovement and provide vibrational feed­back. But VR in the precise form that MrLanier originally conceived it, with head­mounted displays, data­gloves and im­mersive environments, is used only inobscure corners of industrial design. The�eld that has bene�ted most from thetechnology is probably medicine, whereVR is used for surgical training and plan­ning. VR can also be used during surgeryto enhance the surgeon’s vision, increas­ing the number of non­invasive proce­

dures that can be performed.One problem in assessing the success

of VR is that the term itself, coined by MrLanier, has come to mean many things.�The original term meant a shared experi­ence of a virtual world in which a comput­er simulation intervened in the sensory­motor loop,� says Mr Lanier. �These days,it means just about anything.�

Carry on connectingOddly, in 1996, when Allan Weis had a�crazy idea� to create distributed virtualreality, he hadn’t heard of Jaron Lanier. Atthe time, Mr Weis worked for AdvancedNetwork & Services, a non­pro�t organisa­tion that built and maintained a largesection of the internet under contract tothe American National Science Founda­tion. The NSF provided Mr Weis with agrant of several million dollars to develophis crazy idea, known as tele­immersion.His �rst move was to hold a conference atthe University of Illinois and invite every­body he knew who was involved in virtu­al reality and computer graphics to joinhim in his new venture�the NationalTele­immersion Initiative.

Each person Mr Weis recruited said MrLanier should be running the project. MrWeis eventually tracked him down ando�ered him the job of lead scientist. �Hewas able to get a lot of people with verylarge egos working together on a jointproject and accomplished an enormousamount in just a few years,� says Mr Weis.�He doesn’t let his ego get in the way ofworking with people. And I’ve never metanyone who didn’t really respect him.�

Mr Lanier’s ability to connect withpeople, in order to connect people, ispossibly his greatest asset. When describ­ing him, his friends use terms like �hug­gable�, �jovial� and �humanist�. It is easyto see why the collectivism of Web 2.0 irksMr Lanier. He regards it as impersonal,passive, anonymous and less than thesum of its parts, and calls it a �hive mind�.He has described VR, by contrast, as �the�rst medium to come along that doesn’tnarrow the human spirit�.

�You Are Not A Gadget� is a pessimisticbook, but Mr Lanier argues that it is opti­mistic about humans. �It says that hu­mans are special, they’re not computers,�he explains. He does not just rail againstWeb 2.0. He also decries the lack of ambi­tion in computer science, and worries thatthere is less experimentation than therewas in the past. To criticise computing fora lack of innovation sounds odd. Butwhen computer­science students dream

of being the next Larry Page or Mark Zuck­erberg they are all thinking within thecon�nes of narrow engineering and com­mercial strategies, Mr Lanier argues.�We’re losing track of the vastness of thepotential for computer science,� he says.�We really have to revive the beautifulintellectual joy of it, as opposed to thebusiness potential.�

So it is perhaps surprising that MrLanier has collaborated with some of thevery �rms he criticises. He has done busi­ness with Google, which has acquired acouple of the �rms where he has workedsince the collapse of VPL. He has been anadviser to Linden Lab, the makers of �Sec­ond Life�, an online virtual world. Morerecently he has worked as a �partnerarchitect� with Microsoft on Kinect, acamera that attaches to its Xbox 360games console. Kinect enables full­bodymotion capture, voice recognition and facerecognition, allowing gamers to use theirbodies, voices and faces instead of con­ventional game controllers. When askedto justify biting the hand that feeds him,Mr Lanier observes that Microsoft is �theonly institution in the world for doing thistype of work� on a large scale.

Even more surprising, perhaps, is thatJimmy Wales, the co­founder of Wikipe­dia, describes Mr Lanier as �a lovely guy�.�We may have some disagreements aboutWeb 2.0,� says Mr Wales, �but I considerhis criticism to be generally thoughtfuland always honest.� Running Wikipediameans that Mr Wales is all too aware ofthe pitfalls, as well as the advantages, ofmass online collaboration. �We may notbe as far apart as people might imagine,�he says of Mr Lanier. 7

�We have to revive the beautiful intellectual joy ofcomputing, as opposed to the business potential.�

20 Brain scan The Economist Technology Quarterly September 4th 2010

O�er to readersReprints of this special report are available fromThe Rights and Syndication Department.A minimum order of �ve copies is required.

Corporate o�erCustomisation options on corporate orders of100 or more are available. Please contact us todiscuss your requirements.For more information on how to order specialreports, reprints or any queries you may haveplease contact:

The Rights and Syndication departmentThe Economist26 Red Lion SquareLondon WC1R 4HQTel +44 (0)20 7576 8148Fax +44 (0)20 7576 8492e­mail: rights@economist.comwww.economist.com.rights

2