THE LIGHTS IN THE TUNNEL / 18
CopyrightedMaterial–Paperback/Kindle available @ Amazon
erallyproduce onlyverydimlights. Inthirdworld coun-
trieswithlittleornosafetynet,theseunluckypeoplewill
likelybecastoutfromthetunnel,andtheirlightwilldis-
appearentirely.
Theimpactofautomationisstillverydifficulttodis-
cernamongthemultitudeoflightsinthetunnel.Weno-
tice,however,thatsomeofthebrightestlightsinthetun-
nel are beginning to shine with even more intensity. As
jobsareeliminated,many ofthebusinessesinthetunnel
becomemoreprofitable.Someofthiswealthisthentrans-
ferredtotheownersandtopexecutivesofthebusinesses.
Asthisprocesscontinues,weseethebrighterlightscon-
tinuetoslowlygainstrengthasmoreoftheaveragelights
graduallydimorflickerout.Thedistributionofincomeis
becomingmoreconcentratedinthetunnel.
Now,finally,webegintoseearealdifferenceinthe
tunnel.Itbecomesobviousthattherearefewerlightsand
thatthenumberiscontinuingtodiminish.Justasthisrea-
lizationstrikesus,weimmediatelyfeelthatthereisanew
senseofurgencypervadingthepanelsthatlinethewallsof
thetunnel.Thepanelsbegintodancewithmoreandmore
desperatemotionandcolorastheyattempttoattractthe
dwindlingnumberoflights.
The businesses on the walls of the tunnel are now
suddenly seeing significantly slower demand for their
products and services. This is happening even though
manyofthebrightestlightsinthetunnelhavecontinued
togaininstrength.
Imagine that your job is to sell as many $50 cell
phones as you can in one hour. You are offered two
The Tunnel / 19
CopyrightedMaterial–Paperback/Kindle available @ Amazon
doors:Behinddoor# 1sitBillGatesandWarrenBuffet,
thetworichestpeopleinAmerica.Behinddoor# 2area
thousand average people. You may well be tempted to
choose the first door just so you’ll get to meet Bill and
Warren,butintermsofgettingyourjobdone,youwould
probablyagreethatdoor# 2isclearlythebestchoice.This
isbecausethedemandforthemassmarketproductsthat
driveoureconomydependmuch moreonthenumber of
potential customers than on the wealth of any particular
customer. You are not going to be able to sell 40 cell
phonestooneperson,nomatterhowwealthytheyare.
Wecannowsensethatmanyofthebusinessesinthe
tunnelareclearlyintrouble.Eventhoughtheyarecontin-
uingtosavemoneyasautomationslowlyeliminatessome
oftheirremainingworkers,thisisnotenoughtomakeup
forthereductionin salestheyareexperiencing.Manyof
these companies are now at the point where they must
takeactiontosurvive.
Agreatdealofeachcompany’sresourcesisinvested
in factories, machines and equipment and offices. These
things, whichan economistmight referto as capital, are
very hard to quickly get rid of. For example, if you just
boughtalotofnewautomatedmachinesforyourfactory,
thenyouarestuckwiththem.Youcan’tjustreturnthem
and get your money back if demand for your products
suddenly starts to fall. For this reason, a business which
seesrapidlyfallingdemandusuallyhasonlyonechoicein
ordertosurvive:cutmorejobs.Weseethis,ofcourse,as
partofthenormalbusinesscycle.Businessesroutinelylay
offworkersinbadtimesandthenrehireingoodtimes.
THE LIGHTS IN THE TUNNEL / 20
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Inthetunnel,wenowseethatthebusinessesarebe-
ginning to cut more and more jobs. They are becoming
more desperate and, in many cases, they must eliminate
evenkeyemployeesthattheyformerlyfeltwerecrucialto
theiroperations.Asthishappens,webegintoseesomeof
thebrighterlightsinthetunnelrapidlybegintodim.
The continuing decrease in demand falls especially
heavilyonthemanufacturingbusinesseslocatedindevel-
opingnationslikeChina.Thesebusinessesrelyonproduc-
ingveryhighvolumeproducts,whichtheyexporttofirst
worldnations.Theyarenowseverelycuttingjobsandthe
flowofnewmiddleclasspeopleintothetunnelhasallbut
stopped.
As a result of the job cuts, the lights are becoming
evenmoresparseinthetunnel.Manyofthebusinessesare
now failing and whole regions of the tunnel walls are
growingdark.Nowweseethatmanyoftheverybrightest
lightsinthetunnelfinallyfeeltheimpactandalsobeginto
losetheirlight.Theownersofthebusinessesinthetunnel
areseeingmuchoftheirwealthgraduallydrainaway.
The tunnelhas becomea far darker and more stag-
nant place. We sense clearly that the hopes of even the
remainingbrighterlightsaregraduallyevaporatingintothe
newemptinessofthetunnel.
The Tunnel / 21
CopyrightedMaterial–Paperback/Kindle available @ Amazon
A Reality Check
Clearly,oursimulationdidnotturnoutwell.Perhapsour
initialassumptionaboutjobsbeingautomatedwaswrong.
But, again, let’s leave that for the next chapter. In the
meantime, we might wonder if we have made a mistake
somewhereinthesimulation.Let’sseeifwecanperform
sometypeof“realitycheck”onourresult.Perhapswecan
looktohistorytoseeifthereisanythinginthepastthat
mightsupportwhatwesawhappeninoursimulation.
Let’sleaveourtunnelandtravelbackintimetothe
year 1860. Inthe southern part ofthe United States, we
knowwillfindthegreatestinjusticeeverperpetratedinthe
historyofournation.Here,longbeforethenewlightof
advancedtechnologyfirstbegantoshine,menhaddiscov-
eredafarmoreprimitiveandperverseformofjobauto-
mation.
The injustice and moral outrage associated with sla-
veryrightlyattractsnearlyallofourattention.Forthisrea-
son, most of us don’t have occasion to think about the
overall economic impact of slavery. At the time Abraham
Lincoln was elected president, we know that while the
Northern population’s moral objection to slavery was a
primary divisive issue, there were also significant differ-
ences and debate about issues relating to the differing
economicsystemsoftheNorthandtheSouth.
The Northern economy was built on free labor and
entrepreneurshipandtendedtospreadopportunitymore
equallythroughoutthepopulation.Incontrast,theSouth-
ernstates relied onslavelabor, andwealthwasprimarily
THE LIGHTS IN THE TUNNEL / 22
CopyrightedMaterial–Paperback/Kindle available @ Amazon
concentratedinthehandsofwhiteplantationownerswho
ownedmanyslaves.Oneresultofthissystemwasthatit
wasveryhardforpoorerwhitestoadvancetheirsituation
becauserelativelyfewfreelaboropportunitieswereavaila-
ble.
Documentedobservationsillustratetheimpactofsla-
veryontheSoutherneconomy.InherbookTeamofRivals:
The Political Genius of Abraham Lincoln, Doris Kearns
Goodwin describes a journey that William Seward, who
would years later become Lincoln’s Secretary of State,
took in 1835. Seward traveled with his family from his
homeinNewYorkStatetotheslavestateofVirginia.
5
As
theSewardscrossintoVirginiatheyleavebehindthebus-
tling towns and cities to which they had become accus-
tomed. Instead, they travel a rough, deserted road with
fewhomes,businessesortaverns.Dilapidatedshacksdot
thelandscape,andthelanditselfseemstohavebeenas-
saultedbypoverty.Duringhisjourney,Sewardobserved:
“How deeply the curse of slavery is set upon this vene-
rated and storied regionof theold dominion. Of all the
countriesIhaveseenFranceonlywhoseenergieshavefor
forty years been expended in war and whose population
hasbeenmoredecimatedbytheswordisasmuchdecayed
asVirginia.”
6
Itseemsclearthattherearesomedefiniteparallelsbe-
tweenwhatwesawinoursimulationandtheslaveecon-
omy in the South. We noticed that in our tunnel, the
brightestlightsinitiallybecameevenbrighterastheaver-
agelightsbegantodimandflickerout.Thisfitswellwith
thefactthatmostwealthintheSouthwasconcentratedin
The Tunnel / 23
CopyrightedMaterial–Paperback/Kindle available @ Amazon
thehandsofrichplantationowners,whilethemajorityof
thepopulationwastrappedinpoverty.
Thereisoneimportantdiscrepancy,however.Inour
simulation, the situation continued to deteriorate until
even the brightest lights eventually began to lose their
strength.Incontrast,slaveryintheSouthernstateslasted
forovertwo hundredyears.Theplantationowners were
abletoholdontotheirwealthatleastuntilthestartofthe
CivilWarin1861.Ifoursimulationseemstoindicatethat
aslave(orautomation-based)economyisdestinedtoun-
dergo continuing decline, how is it that the slave states
wereabletomaintainstabilityforsolong?
TheanswerliesinthefactthattheSouthwasprimari-
lyanexporteconomy.Thelargeplantationsproducedraw
cotton which was then shipped to Europe and to the
Northern states where it was manufactured into textiles
andclothing.Itwasthisconstantwealthflowinginfrom
the outside that was able to maintain the economy over
time.
Our simulation, of course, was of the entire world
mass market, so there was obviously no export market
available. In the simulation, we found that across-the-
boardautomationofjobseventuallyreduceddemandfor
productsandservicesasthenumberoflightsinthetunnel
decreased. You can imagine that, if the South had been
completely isolated economically with no outside trade
allowed, it would likely have followed a path of decline
similartotheonewesawinthesimulation.
Infact,oneofPresidentLincoln’sfirstactsafterthe
SouthernstatessecededfromtheUnionwastoimplement
THE LIGHTS IN THE TUNNEL / 24
CopyrightedMaterial–Paperback/Kindle available @ Amazon
acompleteblockadeoftheSouth.The blockadebecame
increasingly effective as the years progressed—ultimately
achieving a 95 percent reduction in Southern cotton ex-
ports—and was certainlyan important factor in the out-
comeofthewar.Bythetimethewarendedin1865,the
Southerneconomywasincompleteruin.Onecanspecu-
latethatiftheblockadecouldhavebeenmaintainedwith-
outanactualshootingwartakingplace,theeconomicim-
pactalonemighthaveintimeledtotheendofslavery.
*
Summarizing
Both our tunnel simulation and our examination of the
Southern slave economy seem to support the idea that
oncefullautomationpenetratesthejobmarkettoasub-
stantial degree, an economy driven by mass-market pro-
duction must ultimately go into decline. The reason for
this is simply that, when we consider the market as a
whole, the peoplewho rely onjobs fortheirincome are
thesameindividualswhobuytheproductsproduced.
Anotherwayofexpressingthisistosaythatalthough
machines may take over people’s jobs, the machines—
unlesswearereallygoingtojumpintothestuffofscience
*
Isitreallyreasonabletodrawacomparisonbetweentheeconomic
effects of slaveryand advanced machine automation? I would argue
that the comparison almost certainly underestimates the economic im-
pactofautonomousmachines.Becauseofitsinhumanity,slaverycar-
rieswithitobviouscosts.Theseincludeboththedirectcostsofen-
slavingunwillinghumanbeingsaswellaslostproductivity.Theown-
ersofmachineswould,ofcourse,seenoneofthesecosts.Inaddition,
machines,whichcanoperateessentiallycontinuously,obviouslyhave
the potential to be far more productive than even a willing human
workercouldbe.
The Tunnel / 25
CopyrightedMaterial–Paperback/Kindle available @ Amazon
fiction—do not participate in the market as consumers.
Recallfromourexampleofsellingcellphonestothetwo
billionairesortoathousandregularpeople,thatmakinga
fewpeoplericherwillnotmakeupforlosingalargenum-
berofpotentialcustomers.Thatmayworkforyachtsand
Ferrarisbutnotforthemassproducedproductsandser-
vicesthatarethebackboneofoureconomy.
Attheverybeginningoftheautomationprocessthis
effectwasnotatallclear.Thefirstbusinessestoautomate
sawasignificantreductionintheircostsastheycutwork-
ers, while the impact on the demand for their products
was negligible—orin fact,demand mayhaveactuallyin-
creasedforatime,astheywereabletolowertheirprices.
Asaresult,theirprofits,andthereforethewealthoftheir
topemployeesandshareholdersincreased.Thesewerethe
brighterlightsinthetunnelthatinitiallybecamestronger.
However,asnearlyallbusinessesinthetunnelcontinued
toautomatejobs,atsomepointthedecreaseinthenum-
berofpotentialcustomersbeganto outweightheadvan-
tagesgainedfromautomation.Oncethishappened,busi-
nesseswereforcedtocutevenmorejobs,whicheliminat-
edevenmoreconsumersfromthemarketandcausedde-
mandtofallstillfurther.Fromthispointon,theeconomy
enteredacontinuingdownwardspiral.
Nota veryhappy ending. However,we still needto
examineourinitialassumption.Isitreallypossiblethat,at
some point in the future, machines or computers could
takeoverthejobsperformedbyalargepercentageofav-
erage workers without new jobs within the capability of
thesepeoplebeingcreated?Couldthatreallyhappen?
THE LIGHTS IN THE TUNNEL / 26
CopyrightedMaterial–Paperback/Kindle available @ Amazon
We’lllookatthatquestioninthenextchapter.We’ll
alsolookatsomethingcalledtheLudditefallacy—whichis
an established line of economic reasoning that strongly
contradictstheresultwesawinoursimulation.
CopyrightedMaterial–
Paperback/Kindle available @ Amazon
Chapter 2
ACCELERATION
Let’snowturntothequestionofwhetherornottheas-
sumptionwemadeaboutjobsbeingautomatedinthefu-
ture is a reasonable one. It might be helpful to start by
turningthatassumptioninsideoutandlookingatitscon-
verse.Ifyoubelievetheassumptionwemadeisincorrect,
thenyoumustbelievethat:
Technology will never
advance to the point where the bulk of jobs
performedbytypicalpeoplewillbeautomated.Theeconomywillal-
wayscreatejobsthatarewithinthecapabilitiesofthevastmajorityof
thehumanpopulation.
When you look at things this way, you might see
some cause forconcern. The real problem, ofcourse,is
that one offending word: “never.” Never is a very long
time:itisthreehundredoraevenathousandyears.Never
is,well,forever.
Tomakethingsmorereasonable,let’slowerthestan-
dardsomewhat.Let’sthinkintermsofourownlifetimes
or the livesof ourchildren. Thatshould make theissue
much more approachable and personal. After all, surely
noneofuswouldwantsomethingdramaticallynegativeto
THE LIGHTS IN THE TUNNEL / 28
CopyrightedMaterial–Paperback/Kindle available @ Amazon
happenduring thelives ofour own children,even ifwe
weren’taroundtoseeit.
With that standard in mind, let’s just assume a rea-
sonableaveragelifespanof80yearsforababyborntoday.
That gives us the year 2089 as a cutoff date. So the as-
sumptionthatwewanttotestnowbecomes:
Technologywillnotadvancetothepointwherethebulkofjobsper-
formed by typical people will be automated before the year 2089.
Priortothatyear,theeconomywillalwayscreatejobsthatarewithin
thecapabilitiesofthevastmajorityofthehumanpopulation.
Canwebankonthat?
The Rich Get Richer
Nearlyallof us sense that our worldis changing rapidly
and that perhaps things seem to be speeding up. We’ve
become accustomed especially to continuous improve-
ment in technology. Wenotice that thelaptop computer
we buy today is dramatically fasterand lighter and more
feature-packed that the one we bought just a few years
ago,andyetitcostsless.Ournewcellphoneissmalleror
lighter,butitdoesmore.
As human beings, we are geared toward thinking in
termsofconstantmotionorgradualchange.Wetendto
analyzethingsintermsofstraightlines.Forthemostpart,
thisishowthephysicalworldaroundusworks.
Weare,ofcourse,familiarwiththeconceptofaccele-
ration.Weexperienceitwhiledrivingorwhenanairplane
takesoff.Butinthecourseofourdailylives,acceleration
is—almost without exception—something that isofvery
Acceleration / 29
CopyrightedMaterial–Paperback/Kindle available @ Amazon
short duration: something typically limited to seconds.
Perhapsforthisreason,itisnoteasyforustoreallycom-
prehendtheideaofan accelerationthatcontinuesrelen-
tlessly for decades. It is difficult for us to really get our
mindsaroundtheimplicationsofthis.
In1965, Gordon E.Moore,theco-founderof Intel
Corporation,observedthat,asaresultofconstantinnova-
tion, thenumberof transistors onasiliconchip roughly
doubled at a consistent pace. Moore speculated that the
rateofgrowthwouldcontinueintotheforeseeablefuture,
andintheyearssince,hisforecasthasproventobecor-
rect.Moore’sobservationinitiallyrelatedto thenutsand
bolts of how chips are fabricated, but over time it has
evolvedintoabroaderruleofthumbthatgivesusauseful
frameworkforthinkingabouthowourabilitytomanipu-
lateandprocessinformationincreasesovertime.Thisrule
ofthumbhasbecomeknownasMoore’sLaw,
*
anditcanbe
expressedasfollows:
Astechnologyprogresses,thecomputationalcapabilityofacomputer
willroughlydoubleeverytwoyears.
Moore’s Law is, of course, not a “law” at all—
certainlynotinthesensethatphysicalrulesliketheones
postulated by Isaac Newton are laws. It is, however, an
accurateobservation andprojection, andnearlyeveryone
inthetechnologyfieldacceptsit.Moore’sLawisanover-
all estimate. Different facets of technology, in fact,
progressatdifferentrates.Still,wecanprobablyagreethat
*
SomeversionsofMoore’sLawuse18monthsratherthan2yearsas
thedoublingstandard.Ihavechosenthemoreconservativenumber.
THE LIGHTS IN THE TUNNEL / 30
CopyrightedMaterial–Paperback/Kindle available @ Amazon
itisourexpandingabilitytomanipulateandcommunicate
informationthatisthedrivingforcebehindthetechnical
innovationweseeallaroundus—andMoore’sLawdoes
anespeciallygoodjobofcapturingtherateofprogressin
thatarena.
When something doubles at a regular pace, we say
thatitgrowsgeometrically,orexponentially.
*
Toillustratethe
extraordinaryaccelerationthat thisimplies,imagine start-
ingwithapennyandthendoublingtheamountyouhave
everydayforamonth.You beginwithonecent; onthe
seconddayyouhavetwocentsandthenfourcentsonthe
thirdday,andsoon.
Thefirstchartonthenextpageshowsthefirstfifteen
daysasourpennydoubles.Youcanseethatwestartout
very slowlyand then begin to accelerate.On dayfifteen,
we have about $164—which is not bad at all since we
startedwithonlyapenny!
Inournextchart,welookatdays15-30.Nowwe’ve
hadtogreatlyexpandthescaleofourbarchartsowecan
accommodate some very big numbers toward the end.
Youcanseethatwestartwhereweleftoffwith$164,but
nowthisamountissotinyagainstournewscalethatwe
don’tevenseeavisiblebar.Wehavetowaituntilday22
before we see a hint of progress—but still that amount
representsnearly21thousanddollars.
Thingsreallystarttoflyfromthere.Wepassthemil-
lion-dollarmarkatday28andenduponday30withover
fivemilliondollars.Notbadforamonth’swork.Ifwehad
*
Thesetermshaveslightly differenttechnical meanings,butforour
purposestheyareinterchangeable.
Acceleration / 31
CopyrightedMaterial–Paperback/Kindle available @ Amazon
been lucky enough to choose a month with 31 days for
ourexperiment,wewouldhavenearlyelevenmilliondol-
lars to show forit. If we couldcontinuethe process for
another thirty days, we would have an astonishing
$5,764,607,523,034,235—ornearlysixquadrilliondollars!
Asyou cansee,a geometricor exponentialprogres-
sionisreallytheultimatecaseof“therichgetricher.”The
moreyouhave,themoreyouget,anditjustkeepsgoing.
When we compare this with the more routine things we
encounter in life, the contrast is astonishing. Consider
economicgrowth,orperhapstheraiseyoumighthopeto
getatwork;inthesethings,wearehappytoseeagainofa
THE LIGHTS IN THE TUNNEL / 32
CopyrightedMaterial–Paperback/Kindle available @ Amazon
fewpercentage points. Canthisbereal?Isthecomputa-
tionalcapabilityofcomputersreallyexpandingthatfast?
Toillustratethatthisisindeedthecase,letmeusean
examplefrommyownexperience.In1981,Ienteredthe
UniversityofMichiganasafreshmanwithplanstostudy
computer engineering.Computer engineering was then a
new discipline just introduced at Michigan and at a few
other universities. Up until then, no one had been quite
surethatcomputerswereimportantenoughtomerittheir
ownengineeringfield.
TheUniversityofMichiganhadoneofthemostad-
vanced computing centers in thecountry. Thecomputer
then in use was a state-of-the-art mainframe machine
manufactured by the Amdahl Corporation. In my first
computerprogrammingcourse,wewereassignedthetask
ofwritingandrunningaprogramusingcomputerpunch
cards.
7
Todothis,youfirstwenttotheuniversitybookstore
and purchased a large box of blank punch cards. These
were similar to, but a little longer than, standard index
cards.
Youthenwroteyourprogramusingpencilandpaper,
andtookyourblankcardstoacardpunchmachineatthe
computing center. You inserted a blank card in the ma-
chineandentered,or“keyedin,”onelinefromyourpro-
gram.Asyoudidthis,themachinepunchedcorrespond-
ing holes in the card. You repeated this for each line in
your program. If you madea mistake,youhad to throw
the entire card away and start over. For a complex pro-
gram,youmighthavetopunchhundredsofcards.
Acceleration / 33
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Next,beingverycarefulnottoscrambleordropthe
stackof cards, you took them toa card reader machine.
Youfedthestackofcardsintothemachineandyourpro-
gramwasenteredintoalonglineofotherprogramswait-
ingfortheattentionofthecomputer.
After a time, in some cases hours, you went to the
print centerandretrievedapaperprintoutoftheresults.
Sinceitisvirtuallyimpossibletowrite aperfectprogram
the first (or usually even second) time, you had to go
through this process several times until you found and
fixedthe“bugs”inyourprogram.
Obviously,thewaywewithinteract withcomputers
haschangeddramatically.Ihadtoincludeadescriptionof
punchcardsaboveforthebenefitofyoungerreaderswho
maynothaveseenthese.Whataboutthecomputeritself?
The mainframe in use at Michigan then was an Amdahl
470/V8.Thiswasamachinethatprobablyoccupiedasig-
nificant portion of a room and cost somewhere in the
neighborhoodoftwomilliondollars.
In order tocompare the relative speeds of different
computers, engineers have developed a measurement
known as Millions of Instructions per Second, or MIPS.
TheMIPSratingofacomputerisabitlikethehorsepow-
erratingofanengine.Whileeachcomputerhasaunique
design,MIPSratingsgiveusausefulwaytomakerough
comparisons.
If you imagine a software program running on a
computer to be similar to someone playing a tune on a
piano, then each computerinstructionwouldcorrespond
toonestrikeofthepianokeys.TheAmdahlmainframeat
THE LIGHTS IN THE TUNNEL / 34
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Michiganhad a rating ofabout seven MIPS.
8
So wecan
thinkofourpianoplayerrippingalongatsevenmillionkeys-
trokespersecond.Obviously,thatisaveryfastpianoplay-
er,andatthetime,itwasprettygoodforacomputer.
By the time I graduated from Michigan in 1985,
thingsoncampushadchangedagreatdeal.Theyearbe-
fore, Apple Computer had released the MacIntosh. The
MacIntoshand its predecessor, the Apple Lisa, werethe
firstcommerciallyavailablepersonalcomputerstohavea
graphical interfaceand a mouse.Theuniversity had pur-
chaseddozensofthesenewcomputers,andstudentswere
now using them in their courses rather than the main-
frame.
The original MacIntosh ran at about one MIPS.
9
In
otherwords,itwasabout1/7asfastastheAmdahlmain-
frame.Thatseemedquiteimpressive.Afterall,theMacIn-
tosh was this tiny thing that sat onyour desk, while the
Amdahlwasa$2millionbehemoththatrequireditsown
room.
Now let’slookathow things progressed afterI left
college:
By1988,Intel’s386DXprocessorwasrunningat
8.5 MIPS. This microprocessor was used in the
firstIBMPC’sthatcouldrunearlyversionsMicro-
softWindows.Thusadesktopcomputerhadnow
exceededthespeedoftheAmdahlmainframe.
By1992,Intel’s486DXran at about54MIPS or
nearly eight times the speed of the old Amdahl
mainframe. 486-based PC’s were the first ma-
chinestoreally provideausefulplatformfor Mi-
Acceleration / 35
CopyrightedMaterial–Paperback/Kindle available @ Amazon
crosoftWindows.Windows3.1,alsointroducedin
1992, became an enormous commercial success
forMicrosoft.
By 1999, theIntel Pentium III was rated at over
1,300MIPS.Ourpianoplayerisnowgoingatover
a billion keystrokes per second. This is close to
200timesthespeedoftheoldAmdahl.
In 2008, an Intel Core 2 Extreme processor was
ratedatupto59,000MIPS.That’s59billionpiano
keystrokes per second and over 8000 times the
speed of our 1981-vintage $2 million Amdahl
mainframe.
Obviously, things have progressed very impressively
overthe 24or so years since I leftcollege. What weare
more interested in, however, is what will happen in the
future.
WeknowfromMoore’sLawthatcomputersarepro-
gressingatageometricora“richgetricher”ratewherewe
doublewhatwealreadyhaveeverytwoyears.Inthefirst
chapter,Iusedanexamplewherewethoughtaboutselling
cellphonestoBillGatesandWarrenBuffet.Let’sdragBill
backintothestorynowandperformanexperimentthat
mightgiveusanideaoftheleveloffutureprogressthat
wecanexpect.
BillGatesleftHarvardin1975tomovetoNewMex-
icoandfoundMicrosoftalongwithhispartnerPaulAllen.
Wecanmarkthat dateasbeingessentiallythe beginning
ofthepersonalcomputerindustry.AsBillstartsworkin
1975,let’simaginethatweslipourmagicpennyintohis
pocket.Bill’sfocusedonotherthings,andhewon’tnotice.
THE LIGHTS IN THE TUNNEL / 36
CopyrightedMaterial–Paperback/Kindle available @ Amazon
We’lldoublethepennyeverytwoyearsandseehowmuch
Billendsupwith:
The IBM PC, which uses Microsoft’s MS-DOS
software, isintroduced in August 1981. This sets
Microsoftonitspathtosuccess.Billnowhaseight
centsinhispocket.
InMarch1986,Microsoftgoespublicanditsstock
tradesforthefirsttimeontheNASDAQmarket.
Billhasabout45cents.
10
Windows 3.1 is introduced in 1992. For the first
time,Microsoft beganto offer some competition
toApple’sMacIntosh.Billnowhasabout$3.60in
hispocket.
WindowsXPisintroducedin2001.Billhasabout
$82.
Let’szoomforwardto2009andlookinBill’spocket:
about$1,300.Obviously,it’sa good thingBilldidn’tpin
hisfortunesonourmagicpenny.
ConsidereverythingthatBillGateshasaccomplished
overhiscareer.HebuiltMicrosoftintotheworld’sdomi-
nantsoftwarecompanyandhasnowretiredfromfulltime
workatthecompanytorunhischaritablefoundation.Af-
ter all that, in terms of our experiment to measure the
geometric acceleration of technology, Bill has less than
1,500 dollars.However,we canalsoseethat thingshave
accelerated quite dramatically in the years between 2001
and2009:injusteightyears,Billhasgainedover$1,200,
comparedwithagainofonly$82overthe26yearsleading
upto2001.
Acceleration / 37
CopyrightedMaterial–Paperback/Kindle available @ Amazon
We know from the charts we looked at earlier that
Billwilleventuallyreachthemilliondollarmark.Whatcan
wesayaboutthefuture?
In2015,Billwillhaveabout$10,500oreighttimes
whathehasin2009.
In2021,Bill willhavenearly$84,000or64 times
the2009figure.
In 2025, Bill will have almost $336,000 or about
258timeswhathehasin2009.
In2031,Billbecomes amulti-millionaire. Hewill
have2.6milliondollarsor2000timeswhathehas
in2009.
Looking at these numbers, we can see that unless
technicalprogressslowssignificantly,computersaregoing
to get dramatically more powerful by 2031. That date is
nearly60yearsbeforethecutoffdateof2089thatweset
atthebeginningofthischapter.
WhatwouldBillhavein2089?1.4quadrilliondollars.
Thisisoveronetrilliontimesthe2009amountof$1,300!
Thesenumbersshouldgiveyouasenseoftheincred-
ible degree of technological acceleration we can expect
overthecomingyearsanddecades.Asfuturistandinven-
tor Ray Kurzweil writes, “Exponential [or geometric]
growthisdeceptive.Itstartsoutalmostimperceptiblyand
explodeswithunexpectedfury.”
11
HowconfidentcanwebethatMoore’sLawwillcon-
tinuetobesustainablein thecomingyears anddecades?
Evidencesuggeststhatitislikelytoholdtrueforthefore-
seeable future. At some point, current technologies will
THE LIGHTS IN THE TUNNEL / 38
CopyrightedMaterial–Paperback/Kindle available @ Amazon
runintoafundamentallimitasthetransistorsoncomput-
erchipsarereducedinsizeuntiltheyapproachthesizeof
individual molecules or atoms. However, by that time,
completely new technologies may be available. As this
book was being written, Stanford University announced
thatscientiststherehadmanagedtoencodetheletters“S”
and“U”withintheinterferencepatternsofquantumelec-
tron waves.
12
In other words, they were able to encode
digital information within particles smaller than atoms.
Advances such as this may well form the foundation of
future information technologies in the area of quantum
computing; this will take computer engineering into the
realmofindividualatomsandevensubatomicparticles.
Evenifsuch breakthroughs don’t arrive intime,and
integrated circuit fabrication technology does eventually
hit a physical limit, it seems very likely that the focus
wouldsimplyshiftfrombuildingfasterindividualproces-
sors to instead linking large numbers of inexpensive,
commoditizedprocessorstogetherinparallelarchitectures.
Aswe’llseeinthenextsection,thisisalreadyhappening
toasignificantdegree,butifMoore’sLaweventuallyruns
out of steam, parallel processing may well become the
primaryfocusforbuildingmorecapablecomputers.
Even if the historical doubling pace ofMoore’s Law
doessomedayprovetobeunsustainable,thereisnorea-
son to believe that progress would halt or even become
linearinnature.Ifthepacefelloffsothatdoublingtook
four years (or even longer) rather than the current two
years, thatwould still beanexponential progressionthat
Acceleration / 39
CopyrightedMaterial–Paperback/Kindle available @ Amazon
would bring about staggering future gains in computing
power.
13
Let’slookatouroriginalassumptionagain:
Technologywillnotadvancetothepointwherethebulkofjobsper-
formed by typical people will be automated before the year 2089.
Priortothatyear,theeconomywillalwayscreatejobsthatarewithin
thecapabilitiesofthevastmajorityofthehumanpopulation.
Does that seem reasonable now? But wait, there’s
more.
World Computational Capability
Back in 1975,itprobablywould havebeen quiteeasyto
makealistofeverycomputerintheworld.Primarily,we
would have found computers in government agencies,
universities, and large corporations. A manufacturer like
IBM could probably have given us a list showing where
eachcomputerwasinstalled.Intheprecedingsection,we
talkedabouthowthepowerandspeedofcomputershas
increased.Ifwe took thatgeometricrate ofincreaseand
justappliedittothecomputersthatexistedin1975,that
wouldbeanincredibleexpansionofcomputationalpower.
Butofcourse,weknowthatisnotwhathappened.
The number of computers in the world has also in-
creased at a fantastic rate. By some estimates, there are
now over a billion personal computers in use. But it
doesn’t stop there. Computersin theform ofembedded
microprocessorsare in ourcell phones, mp3players, car
engines, appliances and in countless other places. Com-
putersareeverywhere.
THE LIGHTS IN THE TUNNEL / 40
CopyrightedMaterial–Paperback/Kindle available @ Amazon
Infact,we mightspeculatethatboththe power and
thenumberofcomputersintheworldareincreasingata
geometric rate—or at least something closetoit.That is
clearlyanincomprehensibleincreaseinourtotalabilityto
manipulate information. If you consider the number of
obsoletedevicesthathavebeenthrownawaysincethePC
wasintroduced,it’seasytoseethatthecomputingpower
in landfills today is many orders of magnitude beyond
whatexistedintheworldin1975.
Itseemsimpossibletoimaginethatsuchanincredible
advanceinourabilitytocomputeandtoprocessinforma-
tioncouldtakeplacewithoutithavingadramaticeffecton
general technology, economics and society in general. In
fact, however, in many areas, change has not come as
quicklyasweperhapsmighthaveexpected.
Cars and airplanes now incorporate computers, but
theiroveralldesignandoperationisstill,forthemostpart,
whatitwasin1975.NASAmanagedtheApollomissions
andreachedthemoonwithoutaccesstomoderncompu-
tingpower.Eventhespaceshuttledatesbacktotheintro-
duction of the first PCs. Likewise, economists speak of
something called the productivity paradox, which basically
saysthat,atleastuntilquiterecently,theeconomyhasnot
reallyshowntheproductivitygainsyoumightexpectgiven
all the new computers that have been introduced into
workplaces. The computer revolution seems, so far, to
havelargelyturneditsenergyinwardonitself,resultingin
Acceleration / 41
CopyrightedMaterial–Paperback/Kindle available @ Amazon
advancesprimarilyintheinformationandcommunication
areas.
*
Ihavethefeelingthatthisstaggeringincreaseinour
computational capability represents a pent up resource
thatispoisedtoburstoutinnewandunexpectedways.In
thefuture,wecanexpectthatmanymoretraditionaltech-
nologies,andinfact nearlyeveryaspectofourlives,will
change—perhaps very rapidly—in ways that we cannot
foresee.Asexamplesofwhatwemightexpect,let’slook
attwothingsthathavealreadyoccurred:onethat,atleast
sofar,hasbeengenerallypositive,andonethathasbeen
decidedlynegative.
Grid and Cloud Computing
Gridcomputingisarapidlygrowingfieldthatfocuseson
leveraging not justthe powerofan individual computer,
butalsothelargenumberofsuchcomputersnowavaila-
ble.Theideaistotiemanycomputerstogetherusingspe-
cial software. A big computational problem can then be
brokendownintopiecesanddistributedacrosshundreds
oreventhousandsofcomputerssothattheycanworkon
it simultaneously. Grid computing has the potential to
bringanunprecedentedlevelofcomputingpowertobear
ondifficultproblemsintheareasofscienceandengineer-
ing.
Oneofthefirstandmostnotableapplicationsofgrid
computing was in theHuman Genome Project.This in-
*
Even much of biotechnology and genetics could be considered a
type of information science because it is focused on cataloging and
understandingtheinformationinourDNA.
THE LIGHTS IN THE TUNNEL / 42
CopyrightedMaterial–Paperback/Kindle available @ Amazon
ternational project began in 1990 and was completed in
2003—twoyearsaheadofschedule. Theprimarygoalof
theprojectwastosequencetheentirehumanDNAmole-
culeandtoidentifythe25,000orsoindividualgenesthat
comprise ourgenetic code. The processofdecoding our
DNAmoleculeandidentifyingeachgenetookatremend-
ousamountofcomputationalresources,andgridcompu-
tingplayedasignificantroleinthis.
Thegeneticinformationobtainedthroughtheproject
isstoredindatabasesandcanbeaccessedbyscientistsand
researchersviatheInternet.Theresultisafantasticsource
ofknowledgethatcontinuestobeanalyzedandwhichis
certain to result in innumerable future advances in the
fieldsofgenetics,bio-engineeringandmedicine.
An especially interestingdevelopment in thefield of
gridcomputingistheideathatunusedpoweronvirtually
anycomputerconnectedtotheInternetcanbeintegrated
intoavoluntarygridanddeployedtosolvebigproblems.
Mostcomputers,iflefton,donothingatallduringlarge
blocksoftime,especiallyovernight.Theideatotiethese
computerstogetherbyhavingtheirownersdonateunused
computingpowerhassprungupinanumberofplaces.
Stanford University’s folding@ home project is geared
towardsolvingdifficultproblemsinaspecializedareaof
biochemistryknownas“proteinfolding.”Advancesinthis
areahavethepotentialtoprovidefuturesolutionsforcan-
cerandfordiseasessuchasHuntington’sandParkinson’s.
AnothermajorplayerinthisareaistheBerkeleyOpenInfra-
structureforNetworkComputing(BOINC).This specialsoft-
ware,developedattheUniversityofCalifornia,Berkeley,