Acceleration / 45
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Astheir computers got fasterand faster, the quants
wereabletodomoreandmore.Theycreatednewexotic
derivativesbasedonstrangecombinationsofthings.They
could magnify the reward (and risk) of a security. They
could invertit,soyougainedifthesecurityfellinvalue.
Theycouldeventrytocapturetherewardifaninvestment
increasedinvalue,buteliminatetheriskifitwentdown—
oratleasttheythoughttheycould.
Ashousingpricescontinuedtoclimbduringthebub-
ble, the subprime loans were packaged into mortgage-
backedsecuritiessothattheycouldbetradedlikebonds.
This had become standardpracticeformortgages. How-
ever, in addition to that, new types of derivatives were
createdbasedonthepackagedsubprimeloans.Mostnot-
able were “collateralized debt obligations” (or CDOs),
which attempted tosiphonoffthelowest risk loans and
repackagethemintoasecuritythatcouldbemarketedasa
high quality investment. These new derivative securities
werethensoldtobanksandfinancialinstitutionsallover
theworld,withtheunderstandingthattheywereverylow
riskinvestments.
Whenthesubprimeborrowersstarteddefaulting,the
valueofthemortgage-backed securitiesplunged,andthe
derivativesdidnotworkasexpected.Inmanycasesitwas
difficultorimpossibletocalculatetheirvalue.Inaddition,
financialinstitutionshadengagedinmanyothercomplex
interrelationships based on exotic derivatives that were
intendedtohelpmanagevariousrisks.Allthisledtoun-
certaintythatcausedvaluestofallevenmore. Theresult
THE LIGHTS IN THE TUNNEL / 46
CopyrightedMaterial–Paperback/Kindle available @ Amazon
wasthedownfallofBearStearnsinMarch2008,andthe
globalcrisisthatfollowed.
The point of this, of course, is that it would have
beenimpossibletocreatetheseweirdderivativeswithout
accesstoverypowerfulcomputers.Ifthesubprimecrisis
hadoccurredinearlieryears,itwouldcertainlyhavebeen
a far smaller event. It’s worth noting that the meltdown
startedin2007.Aswearenowin2009,weknowthatthe
powerofthecomputersonWallStreetdeskshasroughly
doubled,evenasthecrisishascontinued.
Exoticderivativesare,ofcourse,nottheonlyexam-
ple of the dramatic impact of advancing computer tech-
nology on financial markets. On October 19, 1987, the
stock market fell a staggering twenty percent in a single
day.Therewasreallynospecificnewseventorotherfac-
tor thatmighthaveexplainedthesuddendrop. Manyof
the people involved in quantitative technologies on Wall
Street at the time believe that the crash may have been
precipitated by computer programs that traded autono-
mouslyinthehopeofproviding“portfolioinsurance”for
biginvestors.
Asthisis beingwritten,articlesare appearinginthe
pressregardingtheuseofextremelyfastWallStreetcom-
putersthatallowtransactionsto beexecutedinfractions
ofasecond. Thispractice,known as“flashtrading,” has
quicklyattractedthenoticeoftheSecuritiesandExchange
Commissionandmayresultinnewregulation.
Astheseexamplesshow,wecanexpectthattherate
ofchangeandthevolatilityofnearlyeverythingaroundus
will be somehow amplified by the incredible increase in
Acceleration / 47
CopyrightedMaterial–Paperback/Kindle available @ Amazon
ourabilitytocompute.Wecanalso certainlyexpectthat
thisdramaticallyexpandedcomputationalcapacitywillbe
focusedincreasinglyonautomatingourjobs.
Laterinthischapter,we’lllookinmoredetailatsev-
eral specific advancing technologies and how theymight
impact the job market and the economy in general. But
first,let’snowturnfrommachinestohumanbeings.Isit
possiblethatwecansomehow“outrun”computerssowe
canallkeepourjobs?
Diminishing Returns
In1811,EnglandwasinthemidstoftheIndustrialRevo-
lution. That year, a group called the Luddites formed in
Nottingham. The Luddites consisted of skilled textile
workers who felt threatened by the introduction of me-
chanical looms that could be operated by low-paid, un-
skilledworkers.Theytooktheirnamefromamannamed
NedLudwhohadreportedlydestroyedoneof thesead-
vanced looms. The Luddites’protests grew into outright
riotsanddestructionofmachines.TheBritishgovernment
finallyenactedharshmeasuresandthemovementcameto
an end in 1812. Since then, the word “luddite” has, of
course,evolvedintoasomewhatderogatorytermforany-
oneopposedtotechnological progressorillequippedto
dealwithnewtechnologies.
Economistsgenerallydismisstheideathatadvancing
technology will ever permanently displace humans and
thereby continuouslyincreasethe unemployment rate.In
otherwords,mostmainstreameconomistsfullyacceptour
assumption at the beginning of this chapter. (Not the
THE LIGHTS IN THE TUNNEL / 48
CopyrightedMaterial–Paperback/Kindle available @ Amazon
“2089” version; the never one.) Those who have raised
concerns in more recent times are dismissed as “neo-
Luddites.” Economists have also formulated something
calledtheLudditefallacytohelpexplainwhytheconcerns
of neo-Luddites are wrong. We’ll look at this in a little
moredetaillater.
Obviously, England is now a modern country, and
the vast majority of workers still have jobs. The British
peoplearenowfarbetteroffthantheywerein1812.So
weretheLudditeswrong?Orjusttwohundredorsoyears
tooearly?
Weknowthattechnologyhasadvancedtremendously
since1812.Whatabouthumanbeings?Haveweadvanced
aswell? In termsof basicbiology, we areessentially un-
changed. Little if any biological evolution takes place in
onlytwohundredyears.Still,doesn’titseemlikelythatthe
average British worker today is far more capable than a
typicalworkernearlytwohundredyearsago?
Let’simaginewhatlifewaslikeforanaverageEnglish
personin1812.Asitturns out,it’seasytoget somein-
sightinto this becauseCharlesDickens was bornin that
exactyear.Dickensdrewonhisownexperiencesandob-
servationsasachildwhenhelaterwrotehisfamousno-
vels.Hisdescriptionsofaharsh,poverty-strickensociety
and an environment made filthy by the soot from coal-
burningindustryarewellknown.
InOliverTwist,Dickensdescribesthemiserablelifeof
anorphanboyduringtheIndustrialRevolution.Herehe
expresseshisfeelingsasthestarvingOliverisgivenscraps
ofmeatthathadfirstbeenofferedtoadog:“Iwishsome
Acceleration / 49
CopyrightedMaterial–Paperback/Kindle available @ Amazon
well-fed philosopher, whose blood is ice, whose heart is
iron;couldhaveseenOliverTwistclutchingatthedainty
viands the dog had neglected.I wish he could have wit-
nessedthehorribleaviditywithwhichOlivertorethebits
asunderwithalltheferocityoffamine.”
15
Clearly, the average British worker is far better fed
today.Weknowtheenvironmentisalsomuchcleanerand
more healthy. The literacy rate in Britain today is pur-
portedtobeashighas99percent.It’shardtoknowwhat
itwasin1812,butaround50percentmightbea decent
guess—andofcourse,theabilitytoreadandwritewould
havebeenhighlyconcentratedintheupperclasses.
In 1812, there was essentially no public education
available in England. The government did not begin to
investsignificantlyineducationuntil1870,andattendance
was not compulsory until 1880. Obviously, the average
workertodayisfarbettereducatedthanheorshewould
havebeenin1812.
Givenallofthis,wecansaythat,duetodramaticim-
provementsinlivingconditionsandeducation,anaverage
worker today is certainly more capable and able to per-
form more complex, high-level tasks than a worker in
1812.Buttherealquestionis:canweexpectthatkindof
improvementtocontinueinthefuture?
Thefollowinggraphshowswhatanaverageworker’s
abilitytoperformcomplextasksmightlooklikeoverthe
pasttwohundredorsoyears.Thegraphicisjustanintui-
tiveestimate.Itisnotbasedonanyrealdata.However,I
suspect that most people would agree with the general
shapeofthegraph,andthatisallthatreallymatters.
THE LIGHTS IN THE TUNNEL / 50
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Average Worker’s Ability to Perform Complex Tasks
I’vechosenanarbitrarypointonthegraphtoindicate
theyear1812.After thatyear,wecanreasonablyassume
thathumancapabilitycontinuedtorisequitesteeplyuntil
we reach modern times.The steep part of thegraph re-
flectsdramaticimprovementstoouroveralllivingcondi-
tionsintheworld’smoreadvancedcountries:
Vastlyimprovednutrition,publichealth,andenvi-
ronmental regulations have allowed us to remain
relativelyfreefromdiseaseandreachourfullbio-
logicalpotential.
Investmentinliteracyandinprimaryandsecond-
aryeducation,aswellasaccesstocollegeandad-
vancededucationforsomeworkers,hasgreatlyin-
creasedoverallcapability.
A generally richer and more varied existence, in-
cludingeasyaccesstobooks,media,newtechnol-
ogies and the ability to travel long distances, has
probably had a positive impact on our ability to
comprehendanddealwithcomplexissues.
Luddites riot - 1812
Time
Average
Human
Capability
Acceleration / 51
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Thedegreeofimprovementthatwehaveseen,how-
ever,islargelyrelatedtothelowlevelatwhichthingsgot
started.In educationinparticular,weseemtohavehit a
ceiling—andmayactuallybeseeingsomeevidenceofde-
cline.IntheUnitedStates,themediaisrepletewithacon-
tinuingparadeofstoriesabouttheongoingcrisisinboth
primaryandsecondaryeducation.
IntheU.S.,wearenotevensurewhattheactualhigh
schoolgraduationrateis.Apaperpublishedin2008bythe
NationalBureauofEconomicResearch
16
pointsout that
“Dependingonthedatasources,definitions,andmethods
used, the U.S. graduation rate has been estimated to be
anywherefrom66to88percentinrecentyears—anasto-
nishinglywiderangeforsuchabasicstatistic.Therangeof
estimated minority rates is even greater—from 50 to 85
percent.”ArecentlypublishedstudybytheNationalCen-
terforEducationStatistics
17
showedthatover14percent
ofadultsintheUnitedStatesmaylackbasicreadingskills.
Itseemsselfevidentthatifasmanyasathirdofourchild-
renareunabletograduatefromhighschoolandupto1/7
of ourpopulation fails to achieve basic literacy, then we
arenotsucceedinginsignificantlyadvancingthecapability
oftheaverageworker.
Even the earlier trends toward improved nutrition
andpublichealthhave,inmanyways,turnedagainstus.In
mostWesterncountrieswenowhavearagingobesityepi-
demicamongtheadultpopulation,and—mostdisturbing-
ly—alsoamongchildren.Whileadvancesinmedicinecon-
tinue,manyofthesebreakthroughsseemtoprimarilyim-
pact the health of retirement-age people. The overall
THE LIGHTS IN THE TUNNEL / 52
CopyrightedMaterial–Paperback/Kindle available @ Amazon
healthofouryoungerpopulationisstagnantor, insome
cases,perhapsevendeclining.Inrecentyears,oneofthe
fewpositivestoriesinthepublichealthandnutritionarena
hasbeenthedeclineinthesmokingrate.
While the last graph was just an estimate, here is
anothergraph
18
thatisbasedonactualdata:
TheaveragemathscoreonSATtestsadministeredby
theCollegeBoardhasremainedessentiallyflatforthepast
35years.Thegraphforaverageverbalscoreslooksvirtual-
lyidentical.College-boundstudentsthattaketheSATare,
ofcourse,probablyaboveaverageinturnsofworkcapa-
bility.Itseemsprettyclearthat,intermsofincreasingthe
capabilityofouraverageworkers,wehavealreadypicked
thelow-hangingfruit,andwearestrugglingjusttomain-
tainthingsattheircurrentlevel.
Atthispoint,weshouldhaveaprettygoodsensethat
if computer technology continues to progress at the ex-
Average SAT Math Scores 1972-2007
0
100
200
300
400
500
600
700
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
Year
Average Math Score
Acceleration / 53
CopyrightedMaterial–Paperback/Kindle available @ Amazon
traordinaryratewehaveseenintherecentpast,thenhu-
man workers will not be able outrun machine capability.
Youcanseethisvisuallywiththetwographsbelow:
Human Capability v. Computer Technology
Whilethesetwographsarenotbasedonanyspecific
data,wehaveshownprettyconvincinglythattheirshapeis
moreor lesscorrect.Weknowthatthelower(computer
technology) graph currently lies somewhere below the
human average capability graph. And we know that the
technology graph is increasing at an exceptionally fast
geometric pace.Whatelsedo weneed toknow? Clearly,
thelinesseemverylikelytointersectatsomepointinthe
future.
*
*
IfyouarefamiliarwiththewritingsofThomasRobertMalthus,this
graphmaylookfamiliartoyou.In1798,MalthuspublishedhisEssays
on thePrinciple of Population in which he argued thatgeometrically in-
creasinghumanpopulationwouldoutstripsociety’sabilitytoproduce
food. In Malthus’ version of the graph above, the top (diminishing
returns)linerepresentsfoodproduction,whilethebottom(geometric)
Capability
to Perform
Routine Jobs
Computer Technology
Human Workers
Time
THE LIGHTS IN THE TUNNEL / 54
CopyrightedMaterial–Paperback/Kindle available @ Amazon
The continuing advance of computer technology
alongageometricallyincreasingpathandthediminishing
returnsfrominvestmentineducationseemtomakeavery
strong case that the average worker—and perhaps many
above-average workers—are in clear danger of having
theirjobsautomated.Next,let’slookatsometrendsand
specifictechnologiesthatshowexactlyhowthisislikelyto
happen.
Offshoring and Drive-Through Banking
Automationandoffshoreoutsourcinghaveoneimportant
thing in common: they are both driven by technology.
Obviously,itisthevastimprovementinourcommunica-
tionandinformationtechnologiesthathasenabledmany
service-oriented jobs to be relocated to low-wage coun-
tries.
WhenIwasgrowingupinthe1970s,Ioftenhadthe
opportunitytoseedrive-throughbankinginaction.This,
ofcourse,wasbeforetheintroductionofATMmachines.
Atypicalbankdrive-throughwassetupwithtwoorthree
lanessothatmultiplecustomerscouldbehandledatone
time. If you used the lane closest to the building, you
linerepresentspopulation.Hebelievedthatthetwolineswouldinter-
sectandresultinwidespreadfamine.Malthus,ofcourse,turnedoutto
be wrong largely because he failed to anticipate the technological
progressthatwouldoccurinfoodproductionandprocessing.Sodoes
that mean the graph above is just another “Malthusian” prediction
whichisalsodestinedtobewrong?Onethingtokeepinmindisthat
Malthusinessenceplacedhisbetagainsttechnology;thegraphabove
assumes exactly the opposite. We should also acknowledge the un-
happypossibilitythatMalthusmightstillbevindicatedinthefuture,
especiallyifclimatechangehasahighlynegativeimpactonagriculture.
Acceleration / 55
CopyrightedMaterial–Paperback/Kindle available @ Amazon
communicated with the teller through a standard cash
drawer.
Ifyouwereinoneofthelanesfurtherout,however,
thingswerefarmoreinteresting.Yousealedyourmoney,
paperwork,checkbook,etc.intoaplasticcylinderandthen
droppedthe cylinder into theprovidedopening.Thecy-
linder then traveled through an underground tube—
propelled I think by some sort of vacuum mechanism—
untilitreachedtheteller.Shethencompletedthetransac-
tion and sent the cylinder back to you the same way. It
arrived somewhat like a ball being returned at a bowling
alley.
Atthetime,thisseemedveryhightech.Thesystem
haditsflaws,however.Iclearlyrememberwaitinginline
behind one unluckybank customerwho,failingto insert
the cylinder properly, watched it fall to the ground and
then rollunder his car. He then found that when he at-
temptedtogetoutandretrieveit,hewasunabletoopen
hisdoor.This,ofcourse,wasanuproarioussightforan
elevenor twelve year old. I would bewilling to bet that
another potential problem was customersforgetting they
stillhadthecylinderandsimplydrivingawaywithit.
Thistypeofdrive-throughbankhasnow,ofcourse,
followedinthepathofthedinosaurs.Todaythetechnolo-
gyseemsclunky.Atthetime,however,itrepresentedthe
leading edge of what was technically possible. Drive-
throughbanksprovidedausefulconveniencetocustom-
ersandalsooftenofferedextendedhoursofoperation.
ThepointIammakinghereisthatoffshoringisreally
a precursor of automation. Offshoring is what you do
THE LIGHTS IN THE TUNNEL / 56
CopyrightedMaterial–Paperback/Kindle available @ Amazon
when you have some technology, but not enough to fully
automate a job. Just as clunky drive-throughbankswere
eventuallymadeobsoletebyATMs,somanyjobsthatare
currentlybeingoffshoredwill,inthefuture,endupbeing
fully automated. This trend was already discernable in
2004, whenanarticleinInformationWeek pointed out that
“low-wage foreign labor may pose a threat to American
call-center workers, but their counterparts in countries
such as India and the Philippines themselves face being
replaced by increasingly sophisticated voice-automation
technology.”
19
ThisisoneofthereasonsthatIdidnotincludeoff-
shoringinourtunnelsimulation.Wecouldhavesimulated
anoffshoredjobasanaveragelightflickeringoutinone
part of the tunnel and then another somewhat dimmer
light appearing elsewhere. However, our simulation was
intendedtoshowwhatwouldhappenoverthelongrunas
automation gradually increased. As technology continues
itsrelentlessadvance,manyofthejobsnowbeingtrans-
ferredoverseaswillsimplydisappearaltogether.
Currently, most of the controversy and political de-
bate is focused is on offshoring rather than automation.
Thismaywellprovetobeashortsightedview.Informa-
tion technology (IT) workers in the developed nations
havebeenoneofthegroupshardesthitbyjoblossesfrom
offshoring. A 2006 study by the Organisation for Eco-
nomic Co-operation and Development (OECD)
20
con-
cludedthatautomationhasresultedinmoreITjoblosses
thanoffshoringandpredictedthatthistrendwillcontinue.
Acceleration / 57
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Offshoringisthesmall wavethatdistractsyou.Automa-
tionisthebigonefurtheroutthatyoudon’tseecoming.
Short Lived Jobs
Theconventionalwisdomasgenerallypresentedbyecon-
omistsandotheranalystsisthattechnologycreatesjobs.While
history has shown that this is indeed true, it also shows
quiteclearlythatthenewjobtypescreatedbytechnology
areveryoftenthemselvesquickly vaporized by thesame
phenomenon. TheIT jobsthat are now being offshored
and automated are brand new jobs that were largely
createdinthetechboomofthe1990s.Forsomeonewho
choseIT as a promising career pathlittlemorethanten
yearsago,thiscanbeadishearteningreality.
Earlierinthischapter,Itoldofmyexperienceusing
computer punch cards at the University of Michigan. At
thetime,thesecardswereusedfornearlyeverything.The
utility bill you received in the mail was often a type of
punch card. As a result, there were thousands of “new”
jobs forkeypunch operators.Theselaterbecame“new”
jobs for data entry clerks sitting at computer terminals.
Now,ofcourse,technologiessuchasopticalbarcodesare
greatlyreducingtheneedforthistypeofdataentry.
Similarly,Imentionedthatmycollegefieldofstudy,
computerengineering,wasnewatthetime.Softwareen-
gineering isnowalso a highly offshored field, andmuch
progresshasbeenmadetowardautomatingsomepartsof
thesoftwaredevelopmentprocess.Acollegestudenttoday
mightwellthinktwicebeforeselectingthisrelativelynew
fieldthatwascreatedonlyaboutthirtyyearsago.
THE LIGHTS IN THE TUNNEL / 58
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Technology has always caused job transitions. Train
conductors have been largely replaced by airline flight
crews, for example. However, within the high tech and
computerfields,thepaceofchangeisunprecedentedand
continuestodriverelentlesslytowardthetotalelimination
ofjobs.Whatweareseeingisclearempiricalevidenceof
thegeometricincreaseinthepowerofcomputertechnol-
ogy.
Traditional Jobs: The “Average” Lights in the
Tunnel
Alltheattentionbeingfocusedonnewjobsbeingcreated
bytechnologytendstodistractusfromtherealitythatthe
bulkofthe job typesin oureconomyhave remained re-
markablystableovertime.Whiletechnologyhascertainly
impactedthewaypeopleinthesejobswork,orthetypes
ofbusinessesatwhichtheywork,ithasnotyetalteredthe
basicdefinitionsofthesetraditionaljobcategories.
Thetablethatfollowsisconstructedfromdatapub-
lished by the U.S. Bureau of Labor Statistics in May
2006.
21
ItlistsalltheoccupationsintheUnitedStateswith
atleastonemillionworkers.
Acceleration / 59
CopyrightedMaterial–Paperback/Kindle available @ Amazon
U.S. Occupations with at least one million workers (2006)
Occupation Number of
Workers
Percentage
of Workers
Retailsalespersons 4,374,230
3.3%
Cashiers 3,479,390
2.6%
Officeclerks 3,026,710
2.3%
Combinedfoodpreparationandserv-
ingworkers,includingfastfood
2,461,890
1.9%
Registerednurses 2,417,150
1.8%
Laborersandfreight,stock,andma-
terialmovers,hand
2,372,130
1.8%
Waitersandwaitresses 2,312,930
1.7%
Customerservicerepresentatives 2,147,770
1.6%
Janitorsandcleaners,exceptmaids
andhousekeepingcleaners
2,124,860
1.6%
Bookkeeping,accounting,andaudit-
ingclerks
1,856,890
1.4%
Secretaries,exceptlegal,medical,and
executive
1,750,600
1.3%
Stockclerksandorderfillers 1,705,450
1.3%
Truckdrivers,heavyandtractor-
trailer
1,673,950
1.3%
Generalandoperationsmanagers 1,663,280
1.3%
Elementaryschoolteachers 1,509,180
1.1%
Salesrepresentatives,wholesaleand
manufacturing,excepttechnicaland
scientificproducts
1,488,990
1.1%
THE LIGHTS IN THE TUNNEL / 60
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Executivesecretariesandadministra-
tiveassistants
1,487,310
1.1%
Nursingaides,orderlies,andatten-
dants
1,376,660
1.0%
First-linesupervisors/managersof
officeandadministrativesupport
workers
1,351,180
1.0%
Maintenanceandrepairworkers,gen-
eral
1,310,580
1.0%
Teamassemblers 1,250,120
0.9%
Teacherassistants 1,246,030
0.9%
Receptionistsandinformationclerks 1,112,350
0.8%
First-linesupervisors/managersof
retailsalesworkers
1,111,740
0.8%
Accountantsandauditors 1,092,960
0.8%
Secondaryschoolteachers,except
specialandvocationaleducation
1,030,780
0.8%
Constructionlaborers 1,016,530
0.8%
Securityguards 1,004,130
0.8%
TotalofOccupationsListedAbove 50,755,770
38.3%
AllOtherOccupations 81,849,210
61.7%
Total Employment 132,604,980
100.0%
Theseworkers makeup asignificant numberof the
averagelightsthatweautomatedaway in oursimulation.
Where are the “new” jobs created by technology? I can
findexactlyonejobmentionedinthislistwhichcouldnot
haveexistedin1930.Canyoufindit?Giveup?Fourlines
Acceleration / 61
CopyrightedMaterial–Paperback/Kindle available @ Amazon
from the top it says “including fast food.” McDonalds
didn’tintroducethefastfoodconceptuntil1948.
The job types listed in the table make up nearly 40
percentofalltheworkersintheUnitedStates.Eachofus
could probably also come up with dozens of other job
titles that have similarly remained unchanged for half a
centuryor more. Many of these are muchhigher paying
professionaljobs: doctors,dentists, CPAs,lawyers,archi-
tects,pilots,engineers,etc.Thefactisthatthevastmajori-
tyofourworkers continuetobeemployedintraditional
jobs.Thenewjobtypescreatedbytechnologyrepresenta
relatively small fraction of employment and, as noted
above,oftentendnottolastverylong.
Even within high technology industries, the bulk of
jobsaretraditionaljobs.Supposeyoufoundanewtech-
nology start-up company in Silicon Valley. You obtain
funding, and your companystarts to grow. Who do you
hire?Engineers,peopletoworkinaccounting,humanre-
sources, marketing and finance; administrative assistants
andpeopletoworkinshippingandreceiving:theseareall
traditionaljobs.ThepeopleworkingatGoogledonotall
haveweirdnew-agejobs;byandlarge,theyhavethesame
typesofjobsaspeopleworkingatGeneralMotors.What
needs to concern us is not just the number of new jobs
created by technology, butthe typesofjobs. Laterinthis
chapter, we will see that entire traditional job categories
areatriskofbeingheavilyautomatedinthenottoodis-
tantfuture.Tosuggestthattechnologyisgoingtosome-
how createcompletely newjob categoriescapableof ab-
THE LIGHTS IN THE TUNNEL / 62
CopyrightedMaterial–Paperback/Kindle available @ Amazon
sorbingmillionsofworkersdisplacedfromtraditionaljobs
ispurefantasy.
Whataretheimplicationsforoureconomyifalarge
fractionofthesetraditionaljobsareultimatelyautomated
away?Automatedcheckoutlanesarecurrentlyinuseata
numberofretailstores.Wecanbesurethatinthefuture,
these willbecome more reliable,easier to use, and more
popular.Whatwillwedoifsomedayasubstantialpercen-
tageofthethreeandahalfmillioncashiersintheU.S.no
longerhavejobs?Whatadditionaleducationandtraining
canweoffertheseworkers?Andwhatjobswoulditpre-
parethemfor?
And what is the impact of thatpotential unemploy-
mentonmarketdemandforgoodsandservices?Cashiers
aregenerallynothighlypaid,buttheynonethelessexistas
lightsinourmassmarkettunnel.Cashiers,justlikeother
workers,drivecars,buyclothesandconsumerelectronics,
rent DVDs, shop for Christmas gifts and perhaps drink
coffeeatStarbucks.Intermsofunitdemandformoderate-
lypricedpersonalproductslikecellphonesormp3play-
ers,acashiermaycountasmuchasacorporateCEO.
Manyofthejobslistedinthetablearealreadyinthe
processof being automated oroffshored.Otherswillbe
targetedintheverynearfuture.Millionsofotherworkers
in occupations that do not appear in the list are also at
highrisk.As wewill see,thisincludesmanyoccupations
that are not, by any means, either low-skill or low paid.
Allowing these jobs to be relentlessly eliminated by the
millions, without any concrete plan to handle the issues
thatwillresult,isaclearrecipefordisaster.
Acceleration / 63
CopyrightedMaterial–Paperback/Kindle available @ Amazon
A Tale of Two Jobs
A common misconception about automation is the idea
that it will primarily impact lowpaying jobs that require
fewskillsortraining.Toillustratethatthisisnotnecessari-
lythecase,considertwoverydifferentoccupations:ara-
diologistandahousekeeper.
Aradiologistisamedicaldoctorwhospecializesinin-
terpreting images generated by various medical scanning
technologies. Before the advent of modern computer
technology, radiologists focused exclusively on X-rays.
Thishasnowbeenexpandedtoincludealltypesofmedi-
cal imaging, including CT scans, PET scans, mammo-
grams, etc. To become a radiologist you need to attend
collegeforfouryears,andthenmedicalschoolforanother
four.Thatisfollowedbyanotherfiveyearsofinternship
and residency, and often even more specialized training
after that. Radiology is one the most popular specialties
fornewlyminteddoctorsbecauseitoffersrelativelyhigh
pay and regular work hours; radiologists generally don’t
needtoworkweekendsorhandleemergencies.
Inspiteoftheradiologist’strainingrequirementofat
least thirteen additional years beyond high school, it is
conceptually quite easy to envision this job being auto-
mated.Theprimaryfocusofthejobistoanalyzeandeva-
luatevisualimages.Furthermore, theparameters ofeach
imagearehighlydefined sincetheyareoftencoming di-
rectlyfromacomputerizedscanningdevice.
Visualpatternrecognitionsoftwareisarapidlydevel-
oping field that has already produced significant results.
Thegovernmentcurrentlyhasaccesstosoftwarethatcan
THE LIGHTS IN THE TUNNEL / 64
CopyrightedMaterial–Paperback/Kindle available @ Amazon
helpidentifyterroristsinairportsbasedonvisualanalysis
ofsecurityphotographs.
22
Realworldtaskssuchasthisare
probablytechnicallymoredifficultthananalyzingamedi-
calscanbecausetheenvironmentandobjectsintheimage
arefarmorevaried.
Radiologyis already subject to significant offshoring
toIndiaandotherplaces.Itisasimplemattertotransmit
digital scans to an overseas location for analysis. Indian
doctorsearnaslittleas10percentofwhatAmericanradi-
ologistsarepaid.
23
Aswesawearlier,automationwilloften
comerapidlyon the heelsof offshoring,especiallyifthe
jobfocusespurelyontechnicalanalysiswithlittleneedfor
humaninteraction.Currently,U.S.demandforradiologists
continuestoexpandbecauseoftheincreaseinuseofdi-
agnostic scans such as mammograms. However, this
seemslikelytoslowasautomationandoffshoringadvance
and become bigger players in the future. The graduating
medicalstudentswhoarenowrushingintoradiologyfor
itshighpayandrelativefreedomfromtheannoyancesof
dealingwithactualpatientsmayeventuallycometoques-
tionthewisdomoftheirdecision.
Nowlet’sturntoaverydifferentjob:thatofahouse-
keeper. A housekeeper, of course, doesn’t require any
formaleducationatall,butasyoumighthaveguessed,this
jobisactuallymuchhardertofullyautomatethanthera-
diologist’s.Totakeoverthehousekeepingjob,wewould
needtobuildaveryadvancedrobot—orperhapsseveral
robotstoperformvarioustasks.
Ifyouaskedthehousekeepertonamethemostdiffi-
cultpartofhisorherjob,youmightexpecttheanswerto
Acceleration / 65
CopyrightedMaterial–Paperback/Kindle available @ Amazon
becleaningthebathroomsorthewindows.Forourrobot,
however,thetrulydifficulttaskisprobablysomethingthat
isrelativelylight work forthehumanhousekeeper.Con-
sider what is involved in tidying up clutter in a typical
home. Forthehousekeeper, thisis easy. A humanbeing
can instantly recognize objects that are out of place and
canquicklyputthembackwheretheybelong.Buildinga
machinetoreliablydothesamethingisprobablyoneof
themostdifficultchallengesinrobotics.
Ahousekeepingrobotwouldneedtobeabletorec-
ognizehundredsoreventhousandsofobjectsthatbelong
intheaveragehomeandknowwheretheybelong.Inaddi-
tion,itwouldneedtofigureoutwhattodowithanalmost
infinite variety of new objects that might be brought in
fromoutside.
Designingcomputersoftwarecapableofrecognizing
objects ina verycomplexand variablefield of view and
thencontrollingarobotarmtocorrectlymanipulatethose
objects is extraordinarilydifficult. The task is madeeven
morechallengingbythefactthattheobjectscouldbein
manypossibleorientationsorconfigurations.Considerthe
simplecaseofapairofsunglassessittingonatable.The
sunglassesmightbeclosedwiththelensesfacingdown,or
withthelensesup.Orperhapstheglassesareopen with
the lenses oriented vertically. Or maybe one side of the
glassesisopenandthe other closed.And,ofcourse,the
glassescouldberotatedinanydirection.Andperhapsthey
are touching or somehow entangled with other objects.
Building and programminga robot that isableto recog-
nizethesunglassesinanypossibleconfigurationandthen
THE LIGHTS IN THE TUNNEL / 66
CopyrightedMaterial–Paperback/Kindle available @ Amazon
pickthemup,foldthemandputthembackintheircaseis
sodifficultthatwecanprobablyconcludethatthehouse-
keeper’sjobisrelativelysafeforthetimebeing.
Contrast the housekeeping robot’s complex visual
recognitionchallengewiththetaskofautomating thera-
diologist’sjob.Amedicalscanis,bydefinition,precisein
termsofitsscaleandorientation:youknowexactlywhat
youarelookingat.Youdon’tneedtoworryaboutdealing
withunknownobjectsorientedindifferentways.Infact,
theentirepointmaybesimplytolocatesomethingoutof
theordinary, suchasatumor.Itisalsomucheasier and
moreprofitabletopartiallyautomatetheradiologist’sjob.
Therewouldbelittlepointtobuildingahousekeepingro-
botthatcouldonlyclearupsomeoftheclutterinahome.
Ontheotherhand,ifyoucanautomate20percentofthe
radiologist’smoreroutinework,thenyoucanimmediately
eliminateoneoutoffiveradiologyjobs.
Noneofthisistosaythatthehousekeeper’sjobwill
neverbeautomated.Itisverylikelythatintenseresearch
anddevelopmentinroboticswilleventuallyproduceaso-
lution to even the most difficult problems. In addition,
robots already exist to automate a few of the housekee-
per’s moreroutine tasks.You can alreadypurchaseinex-
pensiverobotvacuumcleaners,andlargerindustrialfloor
cleaningrobotsarealsoavailable.AsTheEconomistpointed
out inJune 2008, “Robotsaregetting cleverer and more
dexterous.Theirtimehasalmostcome.”
24
Acceleration / 67
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Still, it seems likely that the radiologist’s job is at
higherriskofbeingautomatedinthenearfuture.
*
Abig
partofthereasonforthisisthattheradiologisthaswhatI
callasoftwarejob.
“Software” Jobs and Artificial Intelligence
WhenIspeakofa“software”job,Idon’tmeanthataper-
sonwhohasthejobnecessarilyworkswithorprograms
software.Isimplymeanthatautomationofthejobpoten-
tiallyrequiresonlysufficientlyadvancedsoftware.Inother
words, someone with a softwarejob could eventuallybe
replaced by a computersimilar to theone that currently
sitsonhisorherdesk.Thereisnoneedforroboticarms
or, in fact,any moving parts at all. Another, more com-
mon, term for people with software jobs is, of course,
knowledgeworker.
Software jobs are also highly subject to offshoring.
The conventional wisdom used to be that becoming a
*
Inreality,thereisanotherfactorthatmightslowtheadoptionoffull
automationinRadiology:thatis malpracticeliability.Becausethe re-
sultofamistakeoroversightinreadingamedicalscanwouldlikelybe
dire for the patient, the maker of a completely automated system
wouldassumehugepotentialliabilityintheeventoferrors.Thisliabil-
ity,ofcourse,also exists for radiologists, but it is distributedacross
thousandsofdoctors.However,itiscertainlypossiblethatlegislation
and/or courtdecisions will largely remove this barrier inthefuture.
Forexample,inFebruary2008,theU.S.SupremeCourtruledinan8-
1decisionthat,incertaincases,medicaldevicemanufacturersarepro-
tected fromproductliability casesaslongasthe FDAhas approved
the device. In general, we can expect that non-technological factors
such as product liability or the power of organized labor will slow
automationincertainfields,buttheoveralltrend will remainrelent-
less.
THE LIGHTS IN THE TUNNEL / 68
CopyrightedMaterial–Paperback/Kindle available @ Amazon
knowledgeworkerrepresentedthebestpathtoaprosper-
ous future. The advent of offshoring has increasingly
calledthispropositionintoquestion.Today,offshoringis
impacting knowledge workers across the board. Jobs in
fields such as radiology, accounting, tax preparation,
graphic design, and especially all types of information
technologyarealreadybeingshippedtoIndiaandtoother
countries.Thistrendwillonlygrow,andasIhavepointed
out previously, where offshoring appears, automation is
oftenlikelytoeventuallyfollow.
Theautomationofsoftwarejobsistiedcloselytothe
fieldofartificialintelligence.Whenmostofusthinkabout
artificial intelligence, we are quickly sidetracked into the
world of science fiction. We think of the robots C3PO
and R2D2 from the Star Wars movies, or perhaps the
HAL2000computerfrom2001:ASpaceOdyssey.Asare-
sultofthis,wehavebeenluredintothefalsebeliefthatin
order to replace us, machines have to become like us—
that,infact,theyhavetosomehowreplicateourhumanity.
Thisissimplynottrue.Howoftenhaseachofussaid
“Iamnotmyjob.”Or“Iworktolive;nottheotherway
around.”Howmuchofyourcompleteidentityasahuman
beingreallygoesintoyourjob?Outsideofwork,youmay
read books; listento a certaintype ofmusic. Maybeyou
haveahobbyorpassion.Perhapsyoufeelstronglyabout
politicsortheenvironment.Certainlyyoucaredeeplyfor
yourchildren,yourfamilyandothersclosetoyou.Collec-
tively, all this makes up who you are. Duplicating all of
thatinamachinecertainlyremainsintherealmofscience
fiction.Buthowmuchofallthatisreallyrequiredtodoyourjob?
Acceleration / 69
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Thefactisthatthebarwhichtechnologyneedstohurdle
inordertodisplacemanyofusintheworkplaceismuch
lowerthanwereallyimagine.
To gain some insight into how artificial intelligence
worksintherealworld,let’sconsidercomputerchess.In
1989, Garry Kasparov, the world chess champion faced
offagainstaspecialcomputercalledDeepThought.Deep
ThoughtwasdesignedatCarnegieMellonUniversityand
IBM.Kasparoveasilydefeatedthemachineinatwogame
match.
In1996,Kasparovfacedanewcomputerdeveloped
by IBM called Deep Blue. Again Kasparov defeated the
computer.In1997,IBMcamebackwithanimprovedver-
sionofDeepBluethatfinallydefeatedKasparovinasix
gamematch.Thisrepresentedthefirsttimethatamachine
haddefeatedthetophumanchessplayer.
Since then, computer chess has continued to
progress.In2006,thenewworldchesschampion,Vladi-
mir Kramnik, lost a match against a German software
programcalledDeepFritz.WhileIBM’sDeepBluewasa
completelycustomcomputeraboutthesizeofarefrigera-
tor,DeepFritzisaprogramthatrunsonacomputerusing
twostandardIntelprocessors.Itseemshighlylikelythat,
inthenearfuture,aprogramlikeDeepFritz,runningon
virtuallyanycheaplaptopcomputer,willbeabletodefeat
thebestchessplayersintheworld.
Whenwethinkofwhatittakesforahumanbeingto
be a world chess champion, most of us would probably
agree that it takes a certain degree ofcreativity—atleast
within theconfinesofa highly definedsetof rules. Yet,