Wiener Cynernetique

  • Upload
    comaide

  • View
    228

  • Download
    0

Embed Size (px)

Citation preview

  • 7/26/2019 Wiener Cynernetique

    1/194

  • 7/26/2019 Wiener Cynernetique

    2/194

    TITUTEOFTECHNOLOGY

    eofmustnot

    withoutthe

    ublisher.

    Y,1949

    sof America

  • 7/26/2019 Wiener Cynernetique

    3/194

  • 7/26/2019 Wiener Cynernetique

    4/194

  • 7/26/2019 Wiener Cynernetique

    5/194

  • 7/26/2019 Wiener Cynernetique

    6/194

  • 7/26/2019 Wiener Cynernetique

    7/194

  • 7/26/2019 Wiener Cynernetique

    8/194

    d hadinfactbeena participantin

    minaronthesubjectin 1911-1913.

    wasessentialtohavesomeone

    emathematicalquestionscritically.

    emberofthegroupuntilDr. Rosen-

    944andthegeneralconfusionofthe

    eetings.

    bluethandIhad sharedtheconvic

    areasforthegrowthof thesciences

    n neglectedasano-man'slandbe

    hedfields.SinceLeibnizthere has

    has hadafullcommandofall the

    ay.Since thattime,sciencehas

    ofspecialists,in fieldswhichshowa

    sivelynarrower.Acenturyago there

    butthere wasaGauss,a Faraday,

    are fewscholarswhocancall

    ansorphysicistsorbiologistswithout

    atopologistoranacousticianor a

    d withthejargonof hisfield,and

    andallits ramifications,but,more

    regard thenextsubjectassomething

    threedoorsdown thecorridor,and

    itonhis ownpartasan unwarran

    econtinuallygrowingandinvading

    likewhatoccurredwhenthe Ore

    adedsimultaneouslybytheUnited

    theMexicans,andthe Russiansan

    oration,nomenclature,andlaws.

    c work,asweshallsee inthebodyof

    nexploredfromthe differentsidesof

    cs,electricalengineering,andneuro-

    singlenotionreceivesa separate

    d inwhichimportantworkhas been

    ed;whilestill otherimportantwork

    bilityin onefieldofresultsthat may

    sicalinthenext field.

    nsofsciencewhichofferthe richest

  • 7/26/2019 Wiener Cynernetique

    9/194

    edinvestigator.Theyareatthe

    torytothe acceptedtechniquesof

    onoflabor.If thedifficultyofa

    athematicalinessence,tenphysio

    maticswillgetpreciselyasfar asone

    athematics,andnofurther.Ifaphy

    athematics,workstogetherwithama

    ophysiology,theonewillbeunable

    msthattheothercanmanipulate,and

    o puttheanswersinany formthat

    r.Rosenbluethhasalwaysinsisted

    ftheseblank spacesonthemapof

    ebyateam ofscientists,eachaspecia

    chpossessingathoroughlysoundand

    hthefieldsofhisneighbors;all inthe

    ofknowingoneanother'sintellectual

    ngthesignificanceofacolleague'snew

    kenona fullformalexpression.The

    avetheskilltoconductaphysiological

    avetheskill tounderstandone,to

    stone.Thephysiologistneednot be

    hematicaltheorem,buthemustbe

    gicalsignificanceandtotellthemathe

    dlook.Wehad dreamedforyears

    dentscientists,workingtogetherin

    fscience,notassubordinatesofsome

    utjoinedbythedesire,indeedby the

    erstandtheregionasa whole,andto

    gthof thatunderstanding.

    atterslong beforewehadchosen

    igationsandourrespectivepartsin

    nthisnewstep wasthewar.1had

    methatif anationalemergency

    initwouldbe determinedlargely

    ontactwiththeprogramofcompu

    byDr.VannevarBush,andmyown

    g Leeonthedesignof electricnet

    important.Inthesummerof

    of myattentiontothedevelopmentof

  • 7/26/2019 Wiener Cynernetique

    10/194

    hesolutionofpartialdifferentialequa-

    restedinthese,andhadconvinced

    blem,ascontrastedwiththeordinary

    elltreatedbyDr.Bush onhisdifferen

    e representationoffunctionsofmore

    so becomeconvincedthatthepro

    oyedintelevision,gavetheanswer

    ctthattelevisionwasdestinedto

    ingbythe introductionofsuchnew

    pendentindustry.

    ningprocessmustvastlyincrease

    th ascomparedwiththenumber

    inarydifferentialequations.Toac

    tsinareasonabletime,itthus became

    edofthe elementaryprocessesto

    dinterruptingthestreamof these

    sentiallyslowernature.It also

    ormtheindividualprocesseswithso

    thattheenormousrepetitionof theele

    dnotbringaboutacumulativeerror

    ccuracy.Thusthefollowingrequire

    and multiplyingapparatusofthe

    dbenumerical,asinanordinaryad

    onabasis ofmeasurement,asinthe

    ;

    s,whichareessentiallyswitchingde

    ectronictubesratherthanongears

    dertosecurequickeraction;

    ththepolicyadoptedinsome

    ellTelephoneLaboratories,itwould

    icalinapparatustoadoptthescale

    tiplication,ratherthanthe scaleoften;

    eof operationsbelaidouton

    ere shouldbenohumaninterven

    awereentereduntilthefinal results

    hatalllogicaldecisionsnecessaryfor

    emachineitself;

  • 7/26/2019 Wiener Cynernetique

    11/194

    ainanapparatusfor thestorageof

    hemquickly,holdthemfirmlyuntil

    y,erasethemquickly,andthenbe

    thestorageofnewmaterial.

    ogetherwithtentativesuggestions

    hem,were sentintoDr. Vannevar

    inawar.Atthatstageof theprepara

    seemtohavesufficientlyhighpriority

    onthemworthwhile.Nevertheless,

    ichhavebeenincorporatedintothe

    utingmachine.Thesenotionswere

    ofthethoughtof thetime,and1do

    claimanythinglikethesolerespon

    n.Nevertheless,theyhaveproved

    hatmymemorandumhadsomeeffect

    gengineers.Atanyrate,as weshall

    k,theyareallideas whichareofin

    hestudyof thenervoussystem.

    thetable,andalthoughithas

    ledtonoimmediateprojectby

    elf.Ouractualcollaborationresul

    hichwaslikewiseundertakenforthe

    thebeginningofthewar,the Ger

    dthe defensivepositionofEngland

    nyscientiststothe improvement

    enbeforethewarit hadbecome

    airplanehadrenderedobsoleteall

    rectionoffire,andthat itwasneces

    rolapparatusallthecomputations

    nderedmuchmoredifficultbythefact

    ncounteredtargets,anairplanehas

    appreciablepartofthevelocityof

    tdown.Accordingly,itisexceeding

    missile,notatthetarget,but insuch

    getmaycometogetherinspaceat

    e musthencefindsomemethodof

    onoftheplane.

    xtrapolatethe presentcourseofthe

  • 7/26/2019 Wiener Cynernetique

    12/194

    This hasmuchtorecommendit.

    nd curvesinflight,thelessis its

    stimeithasto accomplishamission,

    nadangerousregion.Otherthings

    asstraightacourseas possible.

    rstshell hasburst,otherthingsare

    lprobablyzig-zag,stunt,orin some

    tion.

    elyatthe disposalofthepilot,and

    sortofintelligentuseof hischances

    d pokerplayer,forexample,hehas

    odifyhisexpectedpositionbeforethe

    ouldnotreckonthechancesofhitting

    ptperhapsin thecaseofavery waste

    er hand,thepilotdoesnot havea

    omaneuverathiswill.Foronething,

    exceedinglyhighspeed,andanytoo

    coursewillproduceanacceleration

    scious,andmaydisintegratethe

    ycontrolthe planebymovinghis

    ewregimenofflowthatis established

    evelop.Evenwhenit isfullydeve

    heaccelerationoftheplane,andthis

    ustbeconverted,firstintochangeof

    ngeofposition,beforeiti sfinally

    viatorunderthestrainofcombat

    moodto engageinanyverycompli

    oluntarybehavior,andisquitelikelyto

    tivityin whichhehasbeentrained.

    onof theproblemofthecurvilinear

    hile,whethertheresultsshould prove

    ortheactualuseofa controlapparatus

    prediction.Topredictthefuture

    certainoperationon itspast.The

    annotberealizedbyanyconslructible

    ertainoperatorswhichbearit acertain

    ctrealizablebyapparatuswhichwe

    rofessorSamuelCaldwelloftheMass

    nologythattheseoperatorsseemed

  • 7/26/2019 Wiener Cynernetique

    13/194

    diatelysuggestedthatwe trythem

    tialanalyzer,usingthisasaready-

    fire-controlapparatus.Wedid so,

    discussedinthebody ofthisbook.

    fengagedinawar project,inwhich

    myselfwerepartnersinthe investi

    iction andoftheconstructionof

    etheories.

    econdtime 1hadbecomeengaged

    co-electricalsystemwhichwasdesigned

    anfunction inthefirstcase, the

    dpatternofcomputation;and inthe

    thefuture.In thissecondcase,we

    sionof theperformanceofcertain

    fire-controlapparatus,itis true,the

    mesin directlybyradar,butin

    s ahumangun-pointeroragun-

    thefire-controlsystem,andactingas

    ssentialto knowtheircharacteristics,

    mmathematicallyintothe machines

    eirtarget,theplane,isalso humanly

    e toknowitsperformancecharacter

    otheconclusionthatan extremely

    ryactivityiswhat thecontrolengineers

    ussthis inconsiderabledetailinthe

    noughto sayherethatwhenwe

    agivenpattern,thedifferencebetween

    yperformedmotionis usedasanew

    ulatedtomoveinsuch awayasto

    hatgivenbythepattern.For example,

    eof ashipcarriesthe readingofthe

    tillerwhichsoregulatesthe valvesof

    movethetillerinsuch awayasto turn

    ller turnssoasto bringtheotherend

    setamidships,andinthatway registers

    wheel astheangularpositionof

    onor otherdelayingforcewhich

    tiller willincreasetheadmissionof

  • 7/26/2019 Wiener Cynernetique

    14/194

    side,andwilldecreaseiton theother,

    sethetorquetendingtobring thetiller

    sthe feedbacksystemtendstomake

    eringenginerelativelyindependent

    ertainconditionsofdelay etc.,a

    uewillmakethe rudderovershoot,

    eed-backintheotherdirection

    vershootstillmore,untilthe steering-

    doscillationorhunting,and breaks

    ksuchasthat byMcColl',we

    onoffeed-back,theconditionsunder

    andtheconditionsunderwhichit

    omenonwhichweunderstandvery

    ativepointofview.

    p alead-pencil.TodothisI have

    However,forallofus butafew

    notknowwhatthesemusclesare;

    omists,therearefewif anywhocan

    ouswillingin successionofthecon

    ncerned.Onthecontrary,whatwe

    .Oncewehavedeterminedonthis,

    haway thatwemaysayroughly

    hepencil isnotyetpickedupi s

    Thispartofthe actionisnotin full

    ha manner,theremustbea

    m,consciousorunconscious,ofthe

    failedtopickthe pencilupateach

    onthepencil,thisreportmay be

    itis moregenerallykinaesthetic,

    gue,proprioceptive.Iftheproprio-

    nting,andwedonotreplacethemby

    e,weareunabletoperformtheact

    dfind ourselvesinastateof what

    iaof thistypeisfamiliarin the

    tralnervoussystemknownastabes

    anixms.VanNostrand,1945.

  • 7/26/2019 Wiener Cynernetique

    15/194

    theticsenseconveyedbythespinal

    troyed.

    ed-backislikelytobeas seriousa

    vityasa defectivefeed-back.Inview

    owand myselfapproachedDr.

    ecificquestion.Is thereany

    hichthepatient,in tryingtoper

    kepickingup apencil,overshoots

    uncontrollableoscillation?Dr.

    yansweredustliatthereissuch awell-

    calledpurposetremor,andthatit is

    ytothe cerebellum.

    nificantconfirmationofourhypo

    reofat leastsomevoluntaryacti

    urpointof viewconsiderablytrans

    gneurophysiologists.Ihecentral

    appearsasa self-containedorgan,

    ensesanddischarging intothemus

    eof itsmostcharacteristicactivities

    ar processes,emergingfromthe

    uscles,andre-enteringthenervous

    organs,whethertheybe propriocep-

    alsenses.Thisseemedtous tomark

    hatpart ofneurophysiologywhich

    mentaryprocessesofnervesand

    anceofthenervoussystemas aninte

    snewpoint ofviewmeriteda

    andpublishedx.Dr.Rosenblueth

    percouldonlybe astatementotpro

    xperimentalwork,andwedecided

    ourplanfor aninterscientificinsti

    wouldfurnishanalmostidealcenter

    gineeringplane,ithadalready

    wand myselfthattheproblemsof

    communicationengineeringwere

    gelow, Behaviour.Purpose&Teleology,Philo-

    p.18-24(1943).

  • 7/26/2019 Wiener Cynernetique

    16/194

    centerednotaroundthe technique

    utaroundthemuch morefundamen

    whetherthis shouldbetransmitted

    ornervousmeans.Themessageis

    quenceof measurableeventsdistri

    whatiscalleda time-seriesbythe

    nofthefutureof amessageisdone

    itspast,whetherthis operatoris

    athematicalcomputation,orbya

    paratus.Inthis connection,wefound

    echanismswhichwe hadatfirst

    bytwotypesof error,ofaroughly

    thepredictionapparatuswhichwe

    madetoanticipateanextremely

    ddegreeof approximation,this

    salwaysattainedatthe costofan

    bettertheapparatuswasfor smooth

    besetintooscillationby smalldepar

    ndthelongerit wouldbebeforesuch

    Thusthe goodpredictionofasmooth

    more delicateandsensitiveapparatus

    ictionof aroughcurve; andthe

    aratusto beusedina specificcase

    sticalnatureof thephenomenonto

    ngpairof typesoferrorseemedto

    nwiththecontrastingproblemsof the

    momentumtobe foundinthe

    hanics,asdescribedaccordingtohis

    pedthatthesolutionofthe problem

    sonlyto beobtainedbyanappealto

    eriestobepredicted,it wasnotdifficult

    yseemedto beadifficultyin thethe

    wasactuallyanefficienttoolfor sol

    ion.Assumingthe statisticsofatime

    oderiveanexplicit expressionforthe

    ctionby agiventechniqueandfora

    his,wecouldtranslatethe problem

    edeterminationofa specificopera-

  • 7/26/2019 Wiener Cynernetique

    17/194

    toa minimumaspecificpositivequan

    ator.Minimizationproblemsofthis

    dbranchofmathematics,thecalcu

    anchhas arecognizedtechnique.

    que,wewereabletoobtainan expli

    blemofpredictingthefutureofa time

    nature;andevenfurther,to achieve

    ssolutionby aconstructibleappa

    east oneproblemofengineering

    ynewaspect.Ingeneral,enginee

    tobeanart ratherthanascience.By

    sort toaminimizationprinciple,we

    tona farmorescientificbasis.Itoc

    notanisolatedcase,butthat therewas

    ngworkin whichsimilardesignpro

    he methodsofthecalculusofvaria

    hersimilarproblemsbythe same

    astheproblemofthedesignof wave

    ssagecontaminatedbyextraneous

    lbackgroundnoise.Wethenfacethe

    ginalmessage,orthe messageunder

    agemodifiedbyagivenlag,byan ope

    edmessage.Theoptimumdesignof

    pparatusbywhichitisrealized depends

    he messageandthenoise,singlyand

    cedinthe designofwavefilters

    merlyofanempiricalandratherhapha

    witha thoroughscientificjustification.

    deofcommunicationengineering

    e,abranchofstatisticalmechanics.The

    nicshasindeedbeenencroachingon

    rmorethana century.Weshallsee

    isticalmechanicsinmodernphysics

    eforthe interpretationofthenature

    mmunicationengineering,however,

    sticalelementisimmediatelyapparent.

    ationisimpossiblesaveasa trans.

  • 7/26/2019 Wiener Cynernetique

    18/194

    nlyonecontingencyis tobetrans

    mostefficientlyandwiththeleast

    sageatall. Thetelegraphandthe

    irfunctiononlyif themessagesthey

    riedina mannernotcompletely

    ndcanonly bedesignedeffectivelyif

    sagesconformstosomesort ofstatis

    mmunicationengineering,wehadto

    yofthe amountofinformation,inwhich

    tionwasthat transmittedasasinglede

    bablealternatives.Thisideaoccurred

    severalwriters,amongthemthestatisti

    nonoftheBellTelephoneLabora

    er'smotivein studyingthissubjectis

    tisticaltheory;thatofShannonin the

    tion;andthatofthe authorinthe

    ageinelectrical filters.Letitbe

    thatsomeofmyspeculationsinthis

    estotheearlierwork ofKolmogo-

    aconsiderablepartofmywork was

    wascalledto theworkofthe Russian

    f informationattachesitselfvery

    on instatisticalmechanics:thatof

    tof informationinasystemisa

    ganization,sothe entropyofasystem

    ofdisorganization;andtheone is

    other.This pointofviewleadsus to

    sconcerningthesecondlaw ofther

    yofthe possibilityoftheso-called

    estionsariseindependentlyinthe

    rcatalysts,andtheir studyisessential

    ngofsuch fundamentalphenomena

    ismandreproduction.Thethird

    oflife,that ofirritability,belongsto

    terpolationuudExtrapolationvonstationarenZufal-

    Sci. U.S.S.R.Sec. Math.,5.3-14(1941).

  • 7/26/2019 Wiener Cynernetique

    19/194

    tiontheory,andfallsunderthegroup

    ndiscussing1.

    earsago,thegroupof scientists

    dmyselfhadalreadybecomeaware

    esetofproblemscenteringabout

    ndstatisticalmechanics,whetherin

    ssue.Ontheotherhand,wewere

    elackofunityof theliteratureconcer

    bytheabsenceof anycommontermi

    nameforthefield.Aftermuchcon

    totheconclusionthatallthe existing

    yabiasto onesideoranotherto serve

    thefieldas wellasitshould ;andas

    sts,wehavebeenforcedto coinat

    ekexpressiontofillthe gap.Wehave

    eld ofcontrolandcommunication

    chineorin theanimal,bythename

    mfromtheGreekxugepv/lTY^orsteers

    ,wewishto recognizethatthefirst

    backmechanismsisanarticleongover

    byClerkMaxwellin1868 2,and

    oma LatincorruptionofxugspwiT/is.

    efactthatthe steeringenginesofa

    earliestandbestdevelopedforms

    s.

    eticsdoesnotdatefurtherbackthan

    all finditconvenienttousein refer

    edevelopmentofthefield.From1942

    opmentofthesubjectwentaheadonin

    asof thejointpaperofBigelow,

    weredisseminatedbyDr.Rosenblueth

    Yorkin1942underthe auspicesof

    n,anddevotedtoproblemsofcentral

    stem.Amongthosepresentat that

    cCulloch,oftheMedicalSchoolof

    hohadalreadybeen intouchwith

    tisLife"! CambridgeU.Press.Cambridge,Eng.,

    .Soc.(London).March5,1868.

  • 7/26/2019 Wiener Cynernetique

    20/194

    lf,andwhowasinterestedin the

    fthe cortexofthebrain.

    nelementwhichoccurs repeatedly

    cstheinfluenceofmathematical

    patron saintforcyberneticsoutof

    ouldhavetochooseLeibniz.The

    ersabouttwoclosely relatedconcepts

    mbolismandthatofacalculus of

    descendedthemathematicalnotation

    hepresentday.Now,justas the

    sitselfto amechanizationprogres

    dthe deskcomputingmachinetothe

    chinesofthepresentday,sothe

    bnizcontainsthe germsofthemachina

    gmachine.Indeed,Leibnizhimself,

    ,wasinterestedin theconstruction

    hemetal.It isthereforenotinthe

    meintellectuali mpulsewhichhasled

    hematicallogichasat thesametime

    echanizationofprocessesofthought.

    chwecan followisonewhich can

    berofsymbols.Thesesymbols,in

    tothe notionofinfinity,butthis

    nsumup inafinitenumberof stages;

    aticalinduction,whereweprovea

    arameternforn =0,andalso prove

    fromthe case/7,thusestablishingthe

    uesofn. Moreover,therulesof

    mechanismmustbefinitein number,

    peartobe otherwise,througharefe

    finitywhichcanitselfbe statedin

    sbecomequiteevident,bothtothe

    dtothe intuitionistslikeWeyl,that

    hematico-logicaltheoryissubjectto

    s asthosethatlimitthe performanceof

    weshall seelater,itis evenpos

    ytheparadoxesofCantorand of

    entofRussell,andowemuchto his

  • 7/26/2019 Wiener Cynernetique

    21/194

    kforhisdoctor'sthesis attheMassa

    ologytheapplicationofthe techniques

    ebraof classestothestudyof

    icalengineering.Turing,whois

    whohavestudiedthe logicalpossibi

    intellectualexperiment,servedthe

    thewaras aworkerinelectronics,

    programwhichthe NationalPhysi

    onhasundertakenforthedevelop

    nesofthemoderntype.

    mthefieldof mathematicallogic

    s.He hadbeenastudentof Car-

    so beenincontactwith Professor

    ofbiophysicists.Letit beremarked

    hascontributedmuchtodirecting

    maticallymindedtothepossibilities

    lthoughit mayseemtosomeofus

    dbyproblemsof energyandpoten

    assicalphysicstodothe bestpossible

    mslikethenervoussystem,which

    ed energetically.

    unetofallunderMcCulloch'sinflu

    workquiteearlyonproblemscon

    ibres bysynapsesintosystems

    es.IndependentlyofShannon,

    eof mathematicallogicforthedis

    allswitchingproblems.Theyadded

    rominentin Shannon'searlierwork,

    ysuggestedbytheideas ofTuring:

    rameter,theconsiderationofnets

    ynapticandother delays1.

    metDr.J.Lettvinof theBoston

    ymuchinterestedin matterscon

    sms.Hewasaclosefriend ofMr.Pitts2,

    withhis work.HeinducedMr.Pitts

    dtomaketheacquaintanceofDr. Ro-

    ndon MathematicalSociety,II,2.42,230-265(1936).

    tts,W., Alogicalcalculus oftheideasimmanent

    Math.Biophys.,5 115-133(1943).

  • 7/26/2019 Wiener Cynernetique

    22/194

    welcomedhimintoourgroup.

    chusettsInstituteofTechnologyin

    er toworkwithme,andto strengthen

    oundforthestudyofthenewscience

    bythattimebeenfairlyborn, but

    alreadythoroughlyacquaintedwith

    europhysiology,buthadnot hadthe

    yengineeringcontacts.Inparticular,

    hDr.Shannon'swork,andhe hadnot

    epossibilitiesofelectronics.Hewas

    nIshowedhimexamplesofmodern

    edtohimthatthesewere idealmeans

    heequivalentsofhisneuroniccircuits

    me,itbecameclearto usthatthe

    chine,dependingasitdoesonconse

    mustrepresentalmostanidealmodel

    thenervoussystem.Theall-or-none

    eoftheneuronsis preciselyanalogous

    n determiningadigitonthe binary

    eofus hadalreadycontemplatedas

    sofcomputingmachinedesign.The

    mechanismfordeterminingwhethera

    tputfromotherselectedelements

    equatestimulusfor thedischarge

    usthaveits preciseanalogue

    .Theproblemofinterpretingthe

    moryinthe animalhasitsparallelin

    gartificialmemoriesforthemachine.

    onofcomputingmachineshad

    alfor thewareffortthanthefirst

    aveindicated,andwas progressing

    esnot toodifferentfromthosewhich

    ated.Harvard,AberdeenProving

    yofPennsylvaniawerealreadycon

    heInstituteofAdvancedStudiesat

    husettsInstituteofTechnologywere

    eld.Inthisprogramtherewas a

    mechanicalassemblytotheelectrical

  • 7/26/2019 Wiener Cynernetique

    23/194

    oftento thescaleoftwo,fromthe

    ectricalrelay,fromhumanlydirected

    ydirectedoperation;andinshort

    hanthelastwasin conformitywith

    ntDr.Bush.There wasacontinual

    nterestedin thesefields.Wehadan

    ateourideastoourcolleagues,inpar

    ard,Dr. vonNeumannoftheInsti

    ,andDr.Goldstineofthe Eniacand

    versityofPennsylvania.Everywhere

    chearing,andthevocabularyofthe

    ontaminatedwiththetermsofthe

    psychologist.

    edingsDr.vonNeumannandmyself

    ntmeetingof allthoseinterested

    etics,andthismeetingtook placeat

    of1943-1944.Engineers,physiolo

    swereallrepresented.Itwasimpossi

    hamongus,as hehadjustacceptedan

    thelaboratoriesofphysiologyofthe

    ologiainMexico,butDr.McCulloch

    heRockefellerInstituterepresented

    nwasunabletobe present;how

    eofa groupofseveralcomputing-

    articipatedinthemeeting,whileDr.von

    yselfwerethemathematicians.The

    resentationofcyberneticproblems

    milarlythe computing-machinedesign

    dsandobjectives.Attheend of

    ecleartoall thattherewasasub

    deasbetweentheworkersinthe

    eineachgroupcouldalreadyuse

    tterdevelopedbytheothers,and

    bemadeto achieveacommon

    orethis,thewarresearchgroup

    Weaverhadpublishedadocument,

    ted,coveringtheworkof Mr.Bige-

    orsandwavefillers.It wasfound

  • 7/26/2019 Wiener Cynernetique

    24/194

    aircraftfiredid notjustifythedesign

    rvilinearprediction,buttheprinciples

    ctical,andhavebeen usedbythe

    gpurposes,andinseveralfieldsofrela

    typeof integralequationtowhich

    roblemreducesitselfhasbeenshown

    roblems,andinmanyotherproblems

    alinterest.Thusinonewayor ano

    wtheideasof predictiontheoryandof

    communicationengineeringalready

    hestatisticiansandcommunication

    tesandGreat Britain.Italsosawmy

    owoutofprint,andaconsiderable

    ersbyLevinson,Wallmann,Daniell,

    nto fillthegap.I myselfhave

    xpositorypaperunderwayforseve

    avedoneon permanentrecord,but

    etelyundermycontrolhavepreven

    Finally,afterajoint meetingatthe

    cietyandtheInstituteof Mathemati

    orkin thespringof1947,and devo

    sticprocessesfromapointofview

    cs,IhavepassedonwhatI haveal

    scripttoProfessorDoobof theUniver

    opedinhisnotationandaccording

    eMathematicalSurveysseriesof

    calSociety.Ihadalreadydeveloped

    eoflecturesinthe mathematics

    hesummerof1945.Sincethen,my

    or2,Dr.Y. W.Lee,hasreturnedfrom

    seonthenewmethodslorthe design

    pparatusin theM.I.T.electrical

    nthefallof1947,and hasplansto

    ecturesup intoabook.Atthesame

    ernmentdocumentistobereprinted.

    bluethreturnedtoMexicoabout

    thespringof1945,I receivedaninvi-

    hysics(M.I.T.), 1947.

    hysics(M. I.T.),1932.

  • 7/26/2019 Wiener Cynernetique

    25/194

    athematicalSocietytoparticipatein

    adalajarathatJune.Thisinvitation

    sionInstigadorayCoordinadorade

    undertheleadershipofDr.Manuel

    mIhavealreadyspoken.Dr.Rosen-

    some scientificresearchwithhim,

    eCardiologia,underits director

    dedmeitshospitality.

    Mexicoatthattime.Dr.Rosen-

    ntinuealineofwork whichwehad

    Walter13. Cannon,whowasalso

    visitwhichunfortunatelyprovedto

    odow iththerelationbetween,on

    nic, andphasiccontractionsinepi

    d, thetonicspasm,beat,andfibril

    thatheartmusclerepresentedan

    theinvestigationofconductionmecha

    furthermore,thatthe anastomoses

    artmusclefibres presenteduswith

    antheproblemofthe nervoussynapse.

    fulto Dr.Chavezforhisunquestion

    hasneverbeenthepolicyof the

    senbluethtotheinvestigationofthe

    haveanopportunitytocontribute

    directions: thestudyofphenomena

    yinuniformconductingmediaof two

    hestatisticalstudyof theconduc

    etsof conductingfibres.Thefirst

    atheoryofheart flutter,thelatter

    standingof fibrillation.Rolhlines

    a paper1,publishedbyus,and

    earlierresultshaveshowntheneed

    frevisionand ofsupplementation,

    grevisedbyMr.OliverG. Selfridge

    uteofTechnology,whilethestatis

    studyofheart-musclenetshasbeen

    h,A.,ConductionofImpulsesin CardineMuscle.

    205:265

  • 7/26/2019 Wiener Cynernetique

    26/194

    tofneuronalnetsby Mr.WallerPitts,

    monGuggenheimFoundation.The

    gcarriedonbyDr. Rosenbluethwith

    mosoftheInstitutoNacionaldeCar-

    ArmyMedicalSchool.

    goftheMexicanMathematicalSo

    dIpresentedsomeofour results.We

    onclusionthatourearlier plansofcol

    mselvestobepracticable.Wewerefor

    ancetopresent ourresultstoa larger

    946,Dr. McCullochhadmadearrange

    yfoundationforthefirstof aseriesof

    wYorkandtobe devotedtothepro

    meetingshavebeenconductedinthe

    kedoutmostefficientlybyDr.Frank

    nizedthemonbehalfoftheFounda

    gettogetheragroupof modestsize,

    ntyinnumber,ofworkersinvarious

    hem togetherfortwosuccessivedays

    papers,discussions,andmeals

    dthe opportunitytothreshouttheir

    rogressinthinkingalongthe same

    eetingshas beenthegroupthathad

    1944,butDrs. McCullochandFre

    eenthepsychologicalandsociological

    ,andhaveco-optedintothegroupa

    ogists,sociologists,andanthropolo

    gpsychologistshadindeedbeen

    g.Hewhostudiesthe nervoussystem

    dhewhostudies themindcannottor-

    uchofthepsychologyofthe pasthas

    morethan thephysiologyoftheorgans

    holeweightof thebodyofideaswhich

    intopsychologyconcernsthephy

    ehighlyspecializedcorticalareascon

    ofspecialsense.Fromthe beginning,

    heproblemofthe perceptionofGes-

    ormationofuniversals,wouldprove

    sthemechanismbywhichwerecog-

  • 7/26/2019 Wiener Cynernetique

    27/194

    irrespectiveofitsposition,its size,

    sistusinsuchmatters,and toinform

    htbemadeof ourconceptsfortheir

    gussuchpsychologistsasProfessor

    hicago,thelate Dr.KurtLewinof

    teofTechnology,andDr.M.Ericsson

    pology,itismanifestthat theim

    ndcommunicationasmechanismsof

    yondtheindividualintothecommunity.

    pletelyimpossibletounderstandsocial

    eofantswithouta thoroughinvestiga

    mmunication,andwewerefortunate

    Dr.Schierdainthismatter.Forthe

    organization,wesoughthelpfrom

    atesonandMargaretMead;while

    tuteof AdvancedStudieswasour

    eldof socialorganizationbelonging

    eryimportantjointbookongames

    theway,representsa mostinteres

    ationfromthe pointofviewof

    o,althoughdistinctfrom,thesubject

    Lewiriandothersrepresentedthe

    ofopinion samplingandtheprac

    dDr.F.C. S.Norlhrupwasinte

    osophicalsignificanceofourwork.

    eacompletelistof ourgroup.

    pto containmoreengineersandma

    lowandSavage;moreneuroanatomists

    chasvonBoninandLloyd,andso

    n thespringof1946,waslargely

    sbythoseofus whohadbeenpresent

    ndtoa generalassessmentoftheim

    present.Itwasthesenseof the

    indcyberneticsweresufficiently

    tothosepresentto warra'ntaconti

    ntervalsof sixmonths;andthat

    g,weshouldhavea smallmeeting

    mathematicallytrainedtoexplainto

  • 7/26/2019 Wiener Cynernetique

    28/194

    eas possiblethenatureofthe mathe

    d.

    eturnedtoMexicowiththesupport

    tionandthehospitalityoftheInstituto

    continuethecollaborationbetween

    elf.Thistimewedecidedtotakea

    romthetopicoffeed-back,andtosee

    xperimentally.Wechcsethecatas

    andthequadricepsextensorfemoris

    cuttheattachmentofthemuscle,

    owntension,andrecordeditscontrac

    nically.Wealsousedanoscillograph

    selectricalchangesinthemuscle

    ithcats,firstdecerebratedunder

    madespinalby athoracictransaction

    strychninewasused toincreasethe

    clewasloadedtothepoint wherea

    odicpatternofcontraction,whichis

    geofthe physiologist.Weobserved

    ,payingattentiontothephysiological

    don themuscle,thefrequencyof

    oftheoscillation,anditsamplitude.

    s weshouldanalyzeamechanical

    tingthesamepatternofhunting.We

    hemethodsofMcColl'sbookonservo-

    eplaceto discussthefullsignificance

    e nowrepeatingandpreparingto

    owever,thefollowingstatementsare

    probable:that thefrequencyofclonic

    sitivetochangesof theloadingcon

    ted;and thatitis muchmorenearly

    ntsoftheclosedarcefferent-nerve-

    body-afferent-nerve-central-synapse-

    hingelse.This circuitisnoteven

    inearoperatorsifwetake asour

    erof impulsestransmittedbythe

    ,butseemstobecomemuchmore

    numberofimpulsesbyitslogarithm.

    tthatthe formoftheenvelopeof

  • 7/26/2019 Wiener Cynernetique

    29/194

    nerveisnot nearlysinusoidal,but

    curveismuchmorenearlysinusoidal;

    gsystemwithconstantenergylevel,

    imulationmustbesinusoidalinall

    eroprobability.Again,thenotionsof

    remuchmorenearlymultiplicative

    rexample,acompleteinhibition

    zero,anda partialinhibitionmeans

    quantity.It isthesenotionsof

    whichhavebeenused1in thedis

    urthermore,thesynapseisa coin

    outgoingfibreis onlystimulatedif

    pulsesina smallsummation-time

    d.Ifthisthresholdis lowenoughin

    mberofincomingsynapses,the

    estomultiplyprobabilities,andthatit

    telylinearlinkis onlypossibleina

    pproximatelogarithmicityofthe

    rtainlyalliedtotheapproximate

    er-Fechnerlawofsensationintensity,

    lyafirstapproximation.

    thatonthislogarithmicbasis,and

    econductionofsinglepulsesthrough

    eneuromusculararc,wewereableto

    tionstotheactualperiodsof clonic

    quealreadydevelopedbytheservo-

    ationofthefrequenciesofhunting

    ystemswhichhavebrokendown.We

    ationsofabout13.9persecond,in

    oscillationsvariedbetweenfrequen

    rallyremainedwithinarangevarying

    nd17.Underthecircumstances,this

    notthe onlyimportantphenomenon

    ereis alsoarelativelyslowchangein

    slowerchangein amplitude.These

    bynomeanslinear.However,suffi-

    ClonusfromtheInstituto Nacional

  • 7/26/2019 Wiener Cynernetique

    30/194

    constantsofalinearoscillatingsys

    rstapproximationasthoughthey

    s thoughovereachpartofthe

    avedasitwould ifitsparameters

    atthetime.Thisis themethod

    fphysics asthatofsecular perturba

    udytheproblemsofbase-leveland

    thisworkhas notyetbeencom

    bothpossibleandpromising.There

    thoughthetimingofthe mainarcin

    o-neuronarc,theamplificationof

    ableinoneand perhapsinmore

    of thisamplificationmaybeaffected

    cesseswhichrunmuch higherinthe

    anthespinalchain primarilyrespon

    us.Thisvariableamplificationmay

    evelof centralactivity,bytheuse

    tics,bydecerebration,andby many

    tspresentedby Dr.Rosenblueth

    etingheld intheautumnof1946,and in

    kAcademyofSciencesheldat thesame

    fusingthenotionsofcyberneticsover

    werepleasedwithourresults,

    eneralpracticability ofworkinthis

    lessthatthetime ofourcollaboration

    t ourworkhadbeendone undertoo

    esirable topublishwithoutfurther

    n.Thisconfirmationwhichnatu

    futationwearenowseekingin the

    47.

    onhadalreadygivenDr.Rosenblueth

    tofanewlaboratorybuildingat the

    ologia.Wefeltthatthe timewas

    ytothem thatis,to Dr.Warren

    partmentofphysical sciences,and

    arge ofthedepartmentofmedical

    e basisofalong-timescientificcollabo

    ourprogramata moreleisurelyand

  • 7/26/2019 Wiener Cynernetique

    31/194

    reenthusiasticallybackedbyour

    .GeorgeHarrison,DeanofSciences,

    veoftheMassachusettsInstituteof

    egotiations,whileDr.IgnacioChdvez

    eInstituto NacionaldeCardiologfa.

    becameclearthatthelaboratory

    houldbe attheInstituto,bothin

    onoflaboratoryequipmentandto

    stthe RockefellerFoundationhas

    tofscientificcentersinLatinAmerica.

    asforfive years,duringwhichIshould

    otheryear attheInstituto,while

    endsixmonthsoftheinterveningyears

    attheInstitutois tobedevotedtothe

    ofexperimentaldatapertainingto

    ermediateyearsaretobedevotedto

    ,andaboveall,totheverydifficult

    oplewishingtogo intothisnew

    whichwillsecurefor themboththe

    physical,andengineeringbackground,

    cewithbiological,psychological,

    McCullochandMr.Pittsdida pieceof

    erneticimportance.Dr.McCulloch

    mof designinganapparatustoenable

    edpagebyear.The productionof

    ughtheagencyof aphotocellisan

    tedbyany numberofmethods;the

    e patternofthesoundsubstantially

    of thelettersisgiven,whateverthe

    ogueoftheproblemofthepercep

    challowsus torecognizeasquare

    enumberofchangesofsizeand

    ch'sdeviceinvolvedaselective

    tfora setofdifferentmagnifica

    dingcanbeperfomedautomatically

    sscanning,toallowa compari

    givenstandardfigureoffixedbut

    ewhichIhad alreadysuggestedatone

  • 7/26/2019 Wiener Cynernetique

    32/194

    gramof theapparatusbywhich

    onecameto theattentionofDr.von

    sked,Isthis adiagramofthe

    rtex ofthebrain? Actingonthis

    ,withtheassistanceofMr. Pitts,

    the anatomyandthephysiologyof

    histheorytheoperationofscanning

    onsplaysanimportantpart.Thiswas

    1947bothattheMacy meetingand

    orkAcademyofSciences.Finally,

    vesacertain periodictime,which

    thetimeof sweepinordinary

    usanatomiccluestothis timein

    consecutivesynapsesnecessaryto

    erformance.Theseyieldatimeof

    condfora completeperformanceof

    dthis istheapproximateperiodof

    hm ofthebrain.Finally,thealpha

    dence,hasalreadybeenconjectured

    obeimportantin theprocessofform

    eivedaninvitationto participate

    nceinNancyon problemsarisingfrom

    ted,andonmyvoyagethereand

    weeksinEngland,chieflyas aguest

    B.S.Haldane.Ihad anexcellent

    se doingworkonultra-rapidcom

    lyatManchesterandatthe National

    ddinglon,andabovealltotalk over

    yberneticswithMr.Turing atTedding-

    hologicalLaboratoryatCambridge,

    cetodiscussthework thatProfessor

    weredoingonthe humanelementin

    gsuchanelement.I foundtheinte

    s greatandwellinformedin England

    dtheengineeringworkexcellent,

    the smallerfundsavailable.Ifound

    andingofitspossibility inmany

    aldane,H.LevyandBernalcertainly

  • 7/26/2019 Wiener Cynernetique

    33/194

    mosturgentproblemsontheagenda

    hilosophy.Ididnotfind, however,

    beenmade inunifyingthesubjectand

    ds ofresearchtogetheraswehad

    es.

    ancyonharmonicanalysiscon

    uniting statisticalideasandideas

    neeringina mannerwholelyinconfor

    wofcybernetics.Here1mustmention

    Blanc-LapierreandM.Loeve.I

    nterestinthe subjectonthepartof

    gists,andphysicalchemists,particu

    rmodynamicaspectsinso farasthey

    oblemofthenature oflifeitself.

    tsubjectin Bostonbeforemydepar

    Gyorgi,theHungarianbiochemist,and

    dantwithmyown.

    chvisitis particularlyworth

    ague,ProfessorG.deSantillanaof

    M.Freymann,ofthefirmof Her

    stedof methepresentbook.1am

    ehisinvitationasM Freymannisa

    fthepresentbook,as wellasagood

    gup toit,havebeendonein Mexico.

    neof thedirectionsofworkwhich

    acymeetingshassuggestedconcerns

    onandthe techniqueofcommunica

    iscertainlytruethatthe social

    kethe individual;thatitisbound

    mmunication;andthatit hasadyna

    cessesofafeedbacknatureplay an

    both inthegeneralfieldsof anthro

    ndinthe morespecificfieldofeco

    rtantwork,whichwehavealready

    nnandMorgensternonthetheoryof

    geof ideas.Onthisbasis,Drs.Gre

    etMeadhaveurgedme,in viewofthe

    ofthesociologicalandeconomicpro

    confusion,todevotealargepart

  • 7/26/2019 Wiener Cynernetique

    34/194

    ussionofthissideofcybernetics.

    their senseoftheurgencyofthe

    pe thattheyandothercompetent

    msofthis sortwhichIshall dis

    hisbook,1canshare neithertheir

    hefirstclaimonmy attention,nor

    cientprogresscanberegisteredin this

    ciabletherapeuticeffectinthepresent

    inwith,themainquantitiesaffecting

    cal,butthe runsofstatisticsonwhich

    velyshort.Thereis nogreatusein

    eeconomicsofsteelindustrybefore

    oftheBessemerprocess,norin com

    berproductionbeforeandafter the

    bileindustryandthecultivationof

    sthereany importantpointinrun

    ntsof venerealdiseaseinasingletable

    odbeforeandthat aftertheintroduc

    rthe specificpurposeofstudyingthe

    Foragoodstatisticof society,we

    ntiallyconstantconditions,justasfor

    weneedalenswith alargeaperture.

    lensis notappreciablyincreasedby

    perture,unlessthelensis madeofa

    sthatthedelayoflight indifferentparts

    properdesignedamountbylessthana

    h.Similarly,theadvantageoflongruns

    aryingconditionsisspeciousandspu

    ncesareverypoortesting-grounds

    chnique:aspooras thestatistical

    beto abeingofthe orderofsizeofa

    ctuationswhichweignorefroma

    epreciselythemattersofgreatest

    bsenceofreasonablysaferoutine

    eelementofthejudgmentoftheexpert

    estobe madeofsociological,anthro

    uantities,issogreatthatitis nofield

    otyet hadthebulkofexperience

    expert.I mayremarkparenthetically

  • 7/26/2019 Wiener Cynernetique

    35/194

    softhetheory ofsmallsamples,once

    nationofits ownspeciallydefined

    amethodforpositivestatisticalinfe

    otinspire mewithanyconfidence,

    tisticianbywhomthemain elements

    ationare eitherexplicitlyknownor

    dinwhichmy expectationsof

    temperedbyanunderstandingofthe

    hwemayhope toobtain.There

    1 ultimatelyhopetoaccomplish

    heaidofcyberneticideas,butin

    on furtherdevelopments.Oneof

    sthesesforlostorparalyzedlimbs.

    cussionofGestalt,theideasof com

    avealreadybeenappliedbyMcCulloch

    acementoflostsenses,inthe construc

    nabletheblindtoread printbyhea

    uggestedbyMcCullochlakesover

    efunctionsnotonlyof theeyebutof

    manifestpossibilityofdoing

    seof artificiallimbs.Thelossofa

    tonlythe lossofthepurelypassive

    mentor itsvalueasmechanicalexten

    elossofthecontractilepowerofits/

    lltheloss ofallcutaneousandkinars-

    nginit.Thefirst Iwolossesarewhat

    owtriestoreplace.Thethirdhasso

    Inthe caseofasimplepeg-legthis

    hatreplacesthemissinglimbhas no

    wn,and thekinaestheticmechanismof

    tetoreportitsownpositionand velo

    iththearticulatedlimbwithamo

    naheadbythe patientwiththeaid

    ure.Hehasnoadequatereportof

    andthis interfereswithhissureness

    ain.Theredoesnotseemto beany

    uippingtheartificialjointsandthe

    h strainorpressuregauges,which

  • 7/26/2019 Wiener Cynernetique

    36/194

    r otherwise,saythroughvibrators,

    epresentartificiallimbremoves

    edby theamputation,butleavesthe

    per receptors,muchofthisataxia

    ndthe patientshouldbeableto

    ewe alluseindriving acar,which

    ut withamuchsurer gait.What

    gshouldapplywithevenmore force

    reofthemanikinfamiliarto all

    ogyshowsthatthe sensorylossin

    baloneis considerablygreaterthan

    hip-jointamputation.

    reporttheseconsiderationstothe

    onow Ihavenotbeenable toaccom

    whetherthesameideas havealready

    ces,norwhethertheyhavebeentried

    mpracticable.Incase theyhavenot

    acticalconsideration,theyshould

    tefuture.

    herpointwhichlbelievetomerit at

    eartomethatthe modernultra-rapid

    principlean idealcentralnervous

    automaticcontrol;andthatits inputand

    formofnumbersordiagrams,butmight

    thereadingsofartificial sense-organs

    or thermometers,andtheperformance

    hthe aidofstrain-gaugesorsimilar

    manceofthesemotor organsandto

    ,tothecentralcontrol systemasanarti

    earealreadyin apositiontoconstruct

    stanydegreeofelaboratenessof

    Nagasakiandthepublicawareness

    doccurredtomethat wewerehere

    social potentialityofunheard-ofim

    vil. Theautomaticfactory,theas

    nagents,areonlysofarahead ofus

    nesstoput suchadegreeofeffort

    as spent,forexample,inthedeve-

  • 7/26/2019 Wiener Cynernetique

    37/194

  • 7/26/2019 Wiener Cynernetique

    38/194

    entsorlesshas nothingtosellthat

    ytobuy.

    haveasocietybasedonhuman

    rselling.To arriveatthissociety

    anningandagood dealofstruggle--

    othebest,maybe ontheplaneof

    hoknows ?Ithus feltitmydutyto

    dunderstandingofthe positionto

    nterest intheconditionsandthe future

    aborunions.I didmanagetomake

    ersonshighupin theC.I.().,and

    ryintelligentandsympathetichearing.

    ualsneitherInor anyofthemwasable

    asithad beenmypreviousobser

    thintheUnitedStales andinEngland,

    thelabormovementareinthehands

    el,thoroughlywell trainedinthe

    op-stewardshipanddisputesconcer

    ofwork; andtotallyunpreparedto

    al, technical,sociological,and

    concerntheveryexistenceof labor.

    asyenoughtosee:the laborunion

    omtheexactinglifeof aworkmaninto

    ministratorwithoutanyopportunity

    dforthosewhohave thistraining,

    rallyinviting;nor,quite naturally,

    such people.

    tributedtothenewscienceof

    moralpositionwhichis,to saythe

    e.Wehavecontributedtothei nitia

    h,asI havesaid,embracestech

    reatpossibilitiesforgoodand for

    verintotheworldthat existsabout

    Belsenand Hiroshima.Wedonot

    ppressingthese newtechnicaldeve

    heage,andthe mostanyofuscan do

    edevelopmentofthesubjectintothe

    sibleandmost venalofourengineers.

    e thatalargepublicunderstands

  • 7/26/2019 Wiener Cynernetique

    39/194

    ofthepresentwork,andto confine

    efields,such asphysiologyand

    fromwarandexploitation.Aswe

    whohopethat thegoodofa better

    societywhichisofferedbythis new

    eandoutweightheincidental

    gtotheconcentrationofpower

    ated,byitsveryconditionsofexistence,

    nscrupulous).1writein1947,and I

    isaveryslight hope.

    esshis gratitudetoMr.WalterPitts,

    eorgesDubeandMr.FredericWebster

    nuscriptand preparingthematerial

    ologia.

  • 7/26/2019 Wiener Cynernetique

    40/194

    ONIANTIME

    ngfamiliartoeveryGermanchild.

    estehen

    t?

    ehen

    hlet

    fehlet

    hl .

    owestthouhow manystarsstand

    ?Knowestthouhow manyclouds

    orld?TheLord Godhathcounted

    wholegreatnumberbelacking.

    stingthemeforthephilosopher

    ,in thatitputsside bysidetwo

    e similarityofdealingwiththeheav

    almosteveryotherrespectofferan

    myistheoldestofthe sciences,

    ngtheyoungesttobeginto deserve

    arastronomicalphenomenacanbe

    es,whilea precisepredictionof

    nerallynot easyandinmanyplaces

    heanswertothe firstquestionis

    knowhowmanystarsthereare.In

    minor uncertaintiesconcerningsome

    stars,a starisa definiteobject,emi

    gandcataloguing; andifahuman

    rsas wecallthesecatalogues

  • 7/26/2019 Wiener Cynernetique

    41/194

    ONIANTIME41

    ntensethanacertainmagnitude,

    antto usinthe ideaofadivineDurch-

    ther.

    weretoaskthe meteorologistto

    slerungofthe clouds,hemightlaugh

    atientlyexplainthatinall thelan

    ereisnosuch thingasacloud,defined

    permanentidentity;andthatif there

    sthefacilitiesto countthem,norishe

    ngthem.Atopologicallyinclinedme

    defineacloudas aconnectedregion

    ityofthepart ofthewatercontent

    xceedsacertainamount,hut this

    theslightestvaluetoanyone,and

    extremelytransitorystate.What

    rologistissomesuch statisticalstate

    17,1950.Sky38%overcast.Cirro-

    hofastronomywhichdealswith

    meteorology:thestudyofgalaxies

    ersandtheirstatistics, aspursuedfor

    hta,butthisisaveryyoung branch

    anmeteorologyitself,andissomething

    sicalastronomy.Thistradition,

    icatoiy,Durchmusterungaspects,

    atherwiththe solarsystemthanwith

    s.Itis theastronomyofthesolar

    associatedwiththenamesof

    o,andNewton,andwhichwasthe

    cs.

    plescience.Evenbeforetheexis

    amicaltheory,evenasfarback asthe

    dthateclipsesoccurred inregularpre

    backwardsandforwardsovertime.It

    f couldbetterbemeasuredbythe

    rcoursesthaninany otherway.The

    esolarsystemwasthe revolutionofa

    s,whetherinthe formofthePtolemaic

    Copernicantheoryoforbits,andinany

  • 7/26/2019 Wiener Cynernetique

    42/194

    era fashionrepeatsIhepast.The

    alindrome,andthebookofastro

    kwardsasforwards.Thereis nodiffer

    sand directionsbetweenthemotion

    sand oneruninreverse.Finally,

    byNewtontoa formalsetofpostu

    cs,the fundamentallawsofthis

    bythetransformationofthetime-

    e.

    otion-pictureofthe planets,speed

    epictureofactivity,and weretorun

    stillbeapossible pictureofpla

    wtonianmechanics.Onthe other

    motion-picturephotographofthe

    athunderhead,andreverseit,it

    ng.Weshouldseedown-dralts

    ,turbulencegrowingcoarserin tex

    steadoffollowingthe changesof

    eit, andsoonindefinitely.

    weenthe astronomicalsmdthe

    whichbringsaboutallthesedifferences,

    rencebetweentheapparentreversibi

    dthe apparentirreversibilityof

    efirst place,themeteorological

    ast numberofapproximatelyequal

    rycloselycoupledto oneanother,

    temofIhesolar universecontains

    berofparticles,greatly diversein

    eanother inasufficientlylooseway

    ingeffects donotchangethegeneral

    bserve,andtheveryhighordercoup

    negligible.Theplanetsmoveunder

    totheisolationofa certainlimited

    anyphysicalexperimentwecanset

    aredwiththe distancesbetween

    nthesun,are verynearlypoints.

    andplastic deformationstheysuffer,

    nearly rigidbodies;or wherethey

    sare atanyrateofa relativelyslight

  • 7/26/2019 Wiener Cynernetique

    43/194

    ONIANTIME43

    tivemotionoftheir centresiscon

    theymoveisalmostperfectlyfree

    din theirmutualattraction,their

    yconsideredtolie atIheir centresand

    ureofthe lawofgravityfromthe

    minute.Thepositions,velocities,

    f thesolarsystemareextremely

    ndthecomputationoftheirfuture

    oteasy indetail,iseasyand pre

    herhand,inmeteorology,thenum

    s soenormousthatanaccurate

    onsandvelocitiesisutterlyimpos

    reactuallymade,andtheir future

    mputed,weshouldhavenothingbut

    gureswliichwould needaradical

    ouldbeof anyservicetous.The

    mperature,turbulence,etc.,areall

    singlephysicalsituation,buttoa

    ationsofwhich onlyoneactualcase

    gsofallthe meteorologicalstations

    slylaken,theywouldnotgive abil

    essarytocharacterizetheactualstate

    ewtonianpointofview.Theywould

    sconsistentwithan infinityofdiffe

    most,togetherwithcertain apriori

    vingasaprobabilitydistribution,a

    ssibleatmospheres.UsingtheNew

    systemofcausallawswhatever,all

    futuretimeis aprobabilitydistribu

    esystem,andeventhispredictability

    oftime.

    ystem,in whichtimeisperfectly

    obabilityandpredictionleadto

    betweenpastandfuture,becausethe

    answersareasymmetrical.IfIset

    bring thesystemIamconsidering

    sentinsucha waythatIfix certain

    sonablerightto assumethatcertain

    wnstatisticaldistributions.1then

  • 7/26/2019 Wiener Cynernetique

    44/194

    ributionofresultsaftera giventime.

    hIcanreverse.Inorderto doso,

    ckout afairdistributionofsystems

    ononourpart,wouldendupwithin

    ndfindoutwhatthe antecedent

    meago.However,forasystem

    positiontoendup inanytightly

    sorare anoccurrencethatwemay

    dwecannotbaseourexperimental

    countingmiracles.Inshort,we are

    ation tothefutureis differentfrom

    lourquestionsareconditionedby

    ranswersto thesequestionsare

    micalquestionconcerningthe

    inconnectionwithIhe timeof

    areobservingremoteheavenlybodies

    din whichthereseemstobeno

    atureofourexperiment.Whythen

    ermodynamicswhichisbasedon

    bservationsstandusinsuch'good

    answeris interesting,andnottoo

    softhestarsarethroughthe agency

    semergingfromtheobservedobject,

    anperceiveincominglight,butcan

    t;or atleasttheperceptionof

    edbyan experimentassimpleand

    ght.In theperceptionofincoming

    yeora photographicplate.We

    eptionofimagesbyputting themin

    metimepast:wedark-condition

    ges,andwewrapour platesinblack

    Itis clearthatonlysuchaneyeand

    e tous:ifwe weregiventopre-images

    andifwehad toputourplates in

    emand developthembeforeusing,

    erydifficultartindeed.This being

    stars radiatingtousandto thewbole

    nystarswhoseevolutionis inthe

  • 7/26/2019 Wiener Cynernetique

    45/194

    ONIANTIME45

    attractradiationfrom thewhole

    ractionfromuswill notbeperceptible

    actthat wealreadyknowourown

    usthepartofthe universewhich

    t-futurerelations,asfarastheemission

    oncordantwithour own.Thevery

    ansthatits thermodynamicsislikeour

    tingintellectualexperimenttomake

    tbeing whosetimeshouldrunthe

    suchabeing allcommunication

    ble.Anysignalhemightsendwould

    eamofconsequentsfromhispoint of

    urs.Theseantecedentswouldalready

    ould haveservedtousas the

    ignal,withoutpresupposingan

    ntit. Ifhedrew usasquare,we

    is figureasitsprecursors,and it

    ouscrystallizationalwaysperfectly

    mains.Itsmeaningwouldseem to

    eswereadinto mountainsandcliffs.

    wouldappeartousas acatastrophe

    plainablebynaturallaws bywhich

    exist.Ourcounterpartwouldhave

    rningus.Withinanyworld wthwhich

    directionoftime isuniform.

    etweenNewtonianastronomyand

    esliein anintermediateposition,but

    meteorologythantoastronomy.

    veseen,containsacosmic meteoro

    atextremelyinterestingfieldstudied

    knownasthetheoryoftidal evolution.

    treattherelativemovementsofthe

    movementsofrigidbodies,butthis is

    rth,forexample,isnearlysurrounded

    erthe moonthanthecenterofthe

    ctedto themoonthanthesolid part

    rontheotherside islessstrongly

    ighteffectpullsthe waterintotwo

  • 7/26/2019 Wiener Cynernetique

    46/194

    nd oneoppositelothemoon.In

    esehills couldfollowthemoon

    reatdispersalofenergy,andconse

    ostpreciselyunderthemoonand

    ywouldconsequentlyhaveapull on

    greatlyinfluencetheangularposition

    ns.However,thetidalwavethey

    tangledupanddelayedoncoastsand

    eBehringSeaand theIrishSea.It

    thepositionofthe moon,andtheforces

    urbulent,dissipativeforcesof a

    cesmet inmeteorology,andneeda

    ed,oceanographymaybecalled

    drosphereratherthanofthe atmos

    themoonback initscourseabout

    herotationof theearthforward.They

    fthemonthandof thedayevercloser

    edayof themoonisthemonth,

    entsnearlythe samefacetotheearth.

    thisisIheresult ofanancienttidal

    containedsomeliquidorgas or

    dgive undertheearth'sattraction,

    patelargeamountsofenergy.This

    utionisnotconfinedto theearthand

    rvedto somedegreethroughoutall

    espastithas seriouslymoditiedthe

    oughin anythinglikehistorictimes,

    mparedwith therigidbody

    solarsystem,

    tronomyinvolvesfrictionalproces

    snotasingle sciencewhichconforms

    tonianpattern.Thebiologicalsciences

    reof one-wayphenomena.Birthis

    ath,nor isanabolism,thebuildingup

    eof catabolism,theirbreakingdown.

    notfollowapatternsymmetricalin

    thegerm-cellstoformthefertilized

    arrow pointedthroughtimeinone

  • 7/26/2019 Wiener Cynernetique

    47/194

    ONIANTIME47

    ydirectedfrom thepastintothe

    yindicatesadefinitelong-timetrend,

    tedthoughitmightbe,fromthesimple

    dleofthe lastcenturythistrendhad

    entistswithanhonestlyopenmind,

    heproblemofdiscoveringitsmecha

    hroughthesamegreatstep bytwo

    ametime :CharlesDarwinand

    astherealizationthata merefor

    vidualsofa speciesmightbecarved

    essone-directionalorfew-directional

    e varyingdegreesofviabililyofthe

    romthepointofviewof theindivi

    antdogwithoutlegswill certainly

    rd thathasdevelopedthemechanism

    yhaveabetterchanceforsurvivali*

    edfromtheimpedingprojectionsof

    whetherfish,lizard,ormammal,will

    mshape,powerfulbodymuscles,and

    chwillcatchthewater;and ifitis

    e pursuitotswiftprey,its chances

    ts assumingthisform.

    samechanismbywhicha moreor

    combinedintoa ratherdefinitepat

    lholdstoday,thoughwehave a

    fthemechanismonwhichitdepends.

    enus afarmorepreciseanddis

    tythanthatheldby Darwin,whilethe

    etimeof deVrieson,hascom

    tionofthestatisticalbasisof mutation.

    atomyof thechromosome,andhave

    e listofmoderngeneticistsislong

    lofthese,suchasHaldane,havemade

    delianismaneffectivetoolforthe

    fthetidal evolutionofSirGeorge

    son.Neithertheconnectionofthe

    fthefather,nor thechoiceofthe

  • 7/26/2019 Wiener Cynernetique

    48/194

    ortuitous.In tidalevolutionaswellas

    haveamechanismbymeansof

    ty,thatof therandommotionsofthe

    fthemoleculesof thewater,iscon

    cessintoa patternofdevelopment

    on.Thetheoryoftidalevolu

    stronomicalapplicationoftheelder

    fDarwins,SirCharles,isone ofthe

    ntummechanics.Thisfactmaybe

    essrepresentsanevenfurtherinva

    deas ofstatistics.Thesuccession

    ann-Gibbsrepresentsaprogressive

    micstostatisticalmechanics:lhatis,

    enaconcerningheatand temperature

    Newtonianmechanicsisappliedtoa

    ,notwitha singledynamicalsystem,

    tionofdynamicalsystems;andin

    ernnot allsuchsystemsbutan

    them.Abouttheyear1900it became

    methingseriouslywrongwith ther

    whereitconcernedradiation.The

    werto absorbradiationsofhighfre

    helawofPlanck thananyexisting

    ntheoryhadallowed.Planckgavea

    iationthequantumtheorywhich

    noughTorthesephenomena,butwhich

    eremainderofphysics; andNils

    asimilarlyadhoc theoryoftheatom.

    ohrformedrespectivelythethesis

    nantinomy.Thesynthesisis the

    edbyHeisenbergin1925,in whichthe

    micsof(iibbsisreplaced byastatis

    thatofNewtonandGibbsfor large-

    hichthe completecollectionofdata

    stisnotsufficientto predictthefuture

    thusnottoomuchto saythatnot

    omy,buteventheNewtonianphy

    oftheaverageresultsofa statistical

  • 7/26/2019 Wiener Cynernetique

    49/194

    ONIANTIME49

    countofanevolutionaryprocess.

    onian,reversibletimetoa Gibbs-

    adits philosophicalechoes.Berg-

    encebetweenthereversibletimeof

    ewhappens,andtheirreversibletime

    whichthereis alwayssomething

    heNewtonianphysicswasnotthe

    asperhapsthecentralpoint inthe

    vitalismandmechanism;althoughthis

    esiretoconserveinsomeform or

    of thesoulandof Godagainstthe

    heend,as wehaveseen,thevita-

    adofbuilding awallbetweenthe

    physics,thewallhasbeenerectedto

    ssthatboth matterandlifefindthem

    atthe matterofthenewerphy

    ewton,butitis sometingquiteas

    morphisingdesiresofthevitalists.

    m-theoreticianisnottheethicalfree

    dTycheisasrelentlessa mistressas

    reflectedinitstechnique.The

    ayswereland-surveyors,astronomers,

    e seventeenthandearlyeighteenth

    ersandgrindersoflenses.Asin

    enmadetheirtoolsinthe imageof

    thingbuta pocketorrery,moving

    stialspheres;and iffrictionandthe

    arole init,theyare effectstobeover

    motionof thehandsmaybeasperio

    The chieftechnicalresultofthis

    elofHuyghensandNewtonwasthe

    forthe firsttimeitwas possibleto

    respectableprecision,andtoconvert

    toceansfromathing ofchanceand

    erstoodbusiness.Itis theenginee

    edthemanufacturer,andtothe

    engine.FromtheNewcomenengine

  • 7/26/2019 Wiener Cynernetique

    50/194

    the centralfieldofengineeringhas

    vers.Heathas beenconvertedinfo

    ndtranslation,andthe physicsof

    entedbythatofRumford,Carnot,

    csmakesitsappearance,ascience

    yirreversible;andalthoughtheear

    seemtorepresentaregion ofthought

    hthe Newtoniandynamics,Ihetheory

    gyandthe laterstatisticalexplana

    r secondlawofthermodynamics

    ationofenergythatprinciplewhich

    iencyobtainablebyasteam-engine

    mperaturesoftheboiler andthecon

    used thermodynamicsandtheNewto

    isticaland thenon-statisticalaspects

    yeighteenthcenturiesare theage

    hteenthandthenineteenthcenturies

    m-engines,thepresenttimeis theage

    trol.Thereisin electricalengineering

    ermanyasthe splitbetweenthetech

    d thetechniqueofweakcurrents,and

    tinctionbetweenpowerandcommu

    his splitwhichseparatestheage just

    earenowliving.A ctually,commu

    ealwithcurrents ofanysizewhatever,

    nespowerfulenoughtoswing massive

    shesit frompowerengineeringisthat

    nomyofenergybuttheaccurate

    issignal maybethetapof akey,

    pofatelegraphreceiverat theother

    ransmittedandreceived through

    ne; oritmaybe theturnofa ship's

    ularpositionof therudder.Thus

    ngbeganwithGauss,Wheatstone,and

    eivedits firstreasonablyscienti

    f LordKelvin,afterthefailureof the

    he middleofthelastcentury ;and

    sperhapsHeavisidewhodidthe

  • 7/26/2019 Wiener Cynernetique

    51/194

    ONIANTIME51

    ernshape.Thediscoveryofradar

    worldwar,togetherwiththeexigencies

    fttire,havebroughtto thefielda

    edmathematiciansandphysicists.

    aticcomputingmachinebelongtothe

    hwascertainlyneversoactivelypur

    thepresentday.

    esinceDaedalusorHero ofAlexan

    icertoproduceaworkingsimulacrum

    waysintriguedpeople.Thisdesire

    tomatahasalwaysbeen expressed

    niqueoftheage.Inthe daysofmagic,

    ister conceptoftheGolem,that

    eRabbi ofPraguebreathedinlife

    neffableNameofGod. Inthe

    atonbecomestheclockworkmusic

    pirouettingstifflyon top.Inthe

    utomatonisaglorifiedheatengine,

    efuelinsteadoftheglycogenofthe

    hepresentautomatonopensdoors

    pointsgunsto theplaceatwhicha

    plane,or computesthesolutionofa

    magicalautomatonliesalongthe

    of developmentofthemodern

    mtohavehad muchofaninfluence

    ught.Itis fardifferentwiththe

    sideahasplayeda verygenuine

    earlyhistoryofmodernphilosophy,

    neto ignoreit.

    onsiderstheloweranimalsasau

    d questioningtheorthodoxChris

    avenosouls tobesavedordamned.

    matafunctionissomethingthat Des

    everdiscusses.However,the

    ofthemodeof couplingofthehuman

    n will,withitsmaterialenviron

    tesdoesdiscuss,althoughina very

    eplacesthiscouplingintheone

  • 7/26/2019 Wiener Cynernetique

    52/194

    nowntohim,thepinealgland.As

    ngwhetherornot itrepresentsa

    atterandofmatteronmind heis

    ydoesregard itasa directaction

    butesthevalidityof humanexperience

    worldtothe goodnessandhonesty

    inthismatteris unstable.Either

    whichcaseitis hardtoseehow

    allyexplainsanything,orHeisanactive

    ishardtosee howtheguarantee

    e anythingbutanactiveparticipation

    usthecausalchainofmaterialpheno

    usalchainstartingwiththe actof

    s inustheexperiencescorresponding

    n.Once thisisassumed,itis entirely

    espondencebetweenourwill and

    duceintheexternalworld toasimilar

    thepathfollowed bytheOcca-

    alebranche.InSpinoza,whoisin

    rofthis school,thedoctrineofOcca

    rereasonableformofassertingthat

    eenmindandmatteristhat oftwo

    fGod;butSpinozais notdynamically

    o attentiontothemechanismof

    whichLeibnizstarts,butL eibniz

    sSpinozais geometricallyminded

    ofcorrespondingelements,mind

    mofcorrespondingelements: the

    onceivedafterthepatternof the

    stanceswhichdo notrisetothe degree

    souls,andwhich formpartofthat

    uldhaveattributed tomatter.Each

    seduniverse,withaperfect causal

    fromminusinfinityintimeto the

    ;butclosedthoughthey are,they

    rthrough thepre-establishedharmony

    hemtoclocks, whichhavesobeen

  • 7/26/2019 Wiener Cynernetique

    53/194

    ONIANTIME53

    togetherfromthecreationforall

    madeclocks,theydonotdrift

    sis duetothemiraculouslyperfect

    or.

    orldofautomatawhich,asis natu

    ns,heconstructsafterthemodelof

    onadsreflectoneanother,thisreflec

    ransferofthecausalchainfrom one

    llyasself-containedas,orrather

    thepassivelydancingfigureson

    avenoreal influenceontheout

    ectivelyinfluencedbyit.As he

    ws.Theapparentorganizationof

    thingbetweenafigmentandamiracle.

    solarsystemwrit small.

    theautomatawhicharehumanly

    hernaturalautomata,theanimalsand

    estudiedfrom averydifferentaspect.

    degradationofenergyaretheruling

    vingorganismis aboveallaheat

    rglycogenorstarch,fats,andproteins,

    ,andurea.Itis themetabolicbalance

    tion;and ifthelowworkingtempe

    ttractattentionasopposedtothe

    esofaheatengineof similarefficiency,

    ornerandgliblyexplainedbya contrast

    ergyofthelivingorganismandthe

    engine.All thefundamentalnotions

    energy,andthechiefof theseisthatof

    gofthebodyis abranchofpower

    hisisthepredominatingpointof

    yminded,conservativephysiologists;

    ght ofsuchbiophysicistsasRashevsky

    ssto itscontinuedpotency.

    ealizethatthe bodyisveryfarfrom

    dthatits-componentpartsworkinan

    vailablepowerismuchlesslimited

    e.Theelectronictubehasshownusthat

    ourceof energy,almostallofwhich

  • 7/26/2019 Wiener Cynernetique

    54/194

    ffectiveagencyforperformingdesired

    sworked atalowenergylevel. We

    suchimportantelementsastheneurons,

    omplexof ourbody,dotheirwork

    itionsas vacuumtubes,withtheir

    pliedfromoutsidebythe circulation;

    whichis mostessentialtodescribe

    energy.Inshort,the newerstudy

    emetalor intheflesh,is abranchof

    ng,anditscardinalnotionsarethose

    turbanceornoise aterm taken

    ngineerquantityofinformation,

    n.

    withautomataeffectivelycoupled

    merelybytheirenergyflow,their

    ow ofimpressions,ofincoming

    nsofoutgoingmessages.Theorgans

    eceivedare theequivalentsofthe

    organs.Theycomprisephoto-electric

    r light;radarsystems,receivingtheir

    hydrogen-ion-potentialrecorders,

    e;thermometers,pressure-gaugesof

    sandsoon.Theeffectorsmaybe

    dsorheating-coilsor otherinstru

    s.Betweenthereceptoror sense-organ

    ntermediateset ofelements,whose

    ei ncomingimpressionsintosuch

    redtypeofresponseinthe effectors.

    scentral controlsystemwillvery

    concerningthefunctioningofthe

    secorrespondamongotherthingsto

    dotherproprioceptorsofthe human

    ganswhich recordthepositonofa

    tionofamuscle,etc. Moreover,the

    eautomatonneednotbeusedat

    or storedsoastobecomeavailableat

    e analogueofmemory.Finally,

    isrunning,itsveryrulesof operation

    angeon thebasisofthedata which

  • 7/26/2019 Wiener Cynernetique

    55/194

    ONIANTIME55

    ceptorsin thepast,andthisis not

    ng.

    arenowspeakingarenotthe

    t,northehopeof somefuturetime.

    mostats,automaticgyro-compassship-

    pelledmissilesespeciallysuchasseek

    ftfire-controlsystems,automatically

    s,ultra-rapidcomputingmachines,

    unto beusedlongbeforethe war

    am-enginegovernorbelongsamong

    echanizationofthesecondworldwar

    n,and theneedofhandlingthe

    ergyottheatomwillprobablybring

    tofdevelopment.Scarcelyamonth

    pearsontheseso-calledcontrol

    chanisms,andthepresentageisas

    mechanismsasthenineteenthcenturywas

    neortheeighteenthcenturytheage of

    omataofthepresentageare coupled

    forthereceptionofimpressionsand

    ons.Theycontainsense-organs,

    entofanervoussystemto integrate

    fromtheone totheother.Theylend

    scriptionin physiologicalterms.Itis

    y canbesubsumedunderonetheory

    ysiology.

    anismstotime demandscareful

    hattherelation input-outputisa

    ndinvolvesa deQnitepast-future

    so clearisthatthe theoryofthe

    tisticalone.We arescarcelyever

    nceofa communication-engineering

    To functionadequatelyitmustgive

    eforawholeclass ofinputs,andthis

    actoryperformancefortheclassof

    yexpectedtoreceive.Thusits

    sianstatisticalmechanicsrather than

    mechanics.Weshall studythisin

  • 7/26/2019 Wiener Cynernetique

    56/194

    hapterdevotedtothetheoryofcom

    onexistsinthesame sortofBergso-

    anism;andhencethereisno reason

    swhytheessentialmodeoffunctioning

    ldnot bethesameasthat ofthe

    alismhaswon totheextentthateven

    othetime-structureofvitalism;but

    yisa completedefeat,forfromevery

    eslightestrelationto moralityor

    csisfully asmechanisticastheold.

    newpointofviewmaterialisticis

    :theascendancyofmattercharac

    th-centuryphysicsfarmorethan the

    alismhas cometobebut littlemore

    mechanism.Infact,thewhole

    versyhasbeenrelegatedtothelimbo

  • 7/26/2019 Wiener Cynernetique

    57/194

  • 7/26/2019 Wiener Cynernetique

    58/194

    termineinacompletelyNewtonian

    momentaandpositionsforany

    possibletomakestatementsabout

    meofthesewill havethecharacterof

    ystemwill havecertaincharacteristics

    tainother characteristicswithpro

    arenotionswhich includecomplete

    possibility,butincludemuchmore

    twith abulletofthe dimensions

    tIhit anyspecificpointonthe target

    ughit isnotimpossiblethatI hitit;

    ccase Imustactuallyhitsome

    eventofprobabilityzero.Thusan

    atofmy hittingsomepoint,may

    blageofinstancesofprobability

    rocesseswhichis usedinthetech

    ticalmechanics,althoughit isused

    hereclearly awareofit,is thereso

    encyintoan infinitesequenceof

    safirst,a second,athird,andso

    knownprobability;andthe expression

    rgercontingencyasthesumof the

    pecialcontingencies,whichform an

    cannotsumprobabilitiesin all

    probabilityof thetotaleventfor

    zerosiszero whilewecansum

    cond,a thirdmember,andsoon,

    tingenciesinwhich everytermhasa

    positiveinteger.

    esetwocases involvesrathersubtle

    gthenatureofsetsof instances,and

    werfulmathematician,wasnevera

    bleforaclassto beinfinite,andyet

    tiplicityfromanotherinfiniteclass,

    eintegers?Thisproblemwassolved

    tcenturybyGeorgCantor,andthe

    consider allthedistinctdecimal

  • 7/26/2019 Wiener Cynernetique

    59/194

    LMECHANICS59

    on-terminating,lyingbetweenzero

    heycannotbearrangedin one,two,

    trangelyenough,alltheterminating

    arranged.Thusthedistinction

    atisticalmechanicsis notontheface

    serviceofLebesguetothe Gibbs

    mplicitrequirementsofstatistical

    tingenciesofprobabilityzeroand

    itiesof contingenciescanactually

    siantheorydoesnotinvolvecon

    r,wasnotdirectlybasedon the

    nics,butonwhat lookslikeavery

    yoftrigonometricseries.Thisgoes

    nturyphysicsofwavesandvibrations,

    stionofthegeneralityofthesets

    emwhichcanbesynthesisedoutof

    system outofthosevibrations,

    thepassingoftimesimplymultiplies

    mfromequilibriumby aquantity,

    ndentonthetimealone,andnot on

    tionis expressedasthesumofa

    coefficientsareexpressedasaverages

    tiontoberepresented,multipliedby

    .Thewholetheorydependson the

    fa series,intermsofthe averageofan

    ttheaverageofa quantitywhich

    mzerotoA,and zerofromAto 1,

    astheprobabilitythatthe random

    rvalfrom0to A,ifit isknowntolie

    words,thetheoryneededforthe

    closetothetheorjr neededforanade

    bilitiescompoundedfromaninfinite

    he reasonwhyLebesgue,insolving

    solved thatofGibbs.

    sdiscussedbyGibbshavethemselves

    n.Ifweconsidera certainvery

    edynamicalsystem,withNdegrees

    positionandvelocitycoordinates

  • 7/26/2019 Wiener Cynernetique

    60/194

    alsetof2 Ncoordinates,Nofwhich

    positioncoordinatesandN thegene

    eterminea2N-dimensionalspace

    volume;andifwe takeanyregion

    ointsflowwiththe courseoftime,

    2N coordinatesintoanewset

    time,thecontinualchangeof the

    snotc hangeits2N-dimensional

    snotso simplydefinedastheseregions,

    ratesasystemof measureofthetype

    of measure,andintheconservative

    aretransformedinsucha wayasto

    nt,thereisone othernumerically

    mainsconstant: theenergy.Ifall

    ctonlyononeanother,andthere are

    positionsand fixedorientationsin

    expressionswhichalsoremain

    vectors :themomentum,andthe

    thesystemasawhole. Theyarenot

    tthe systemisreplacedbya system

    om.

    ms,theremaybeother quantities,

    rgy,themomentum,andthemomentof

    changedasthesystemdevelops.

    systemsinwhichanotherinvariant

    tontheinitialcoordinatesandmomenta

    dregularenoughto besubjecttothe

    donLebesguemeasure,areveryrare

    ensel.Insystemswithout otherinva

    thecoordinatescorrespondingto

    totalmomentofmomentum,andinthe

    rdinates,themeasuredeterminedby

    mcoordinateswillitselfdeterminea

    smeasurein spacewilldetermine

    surfaceoutof afamilyoftwo-dimen

    pleifourfamilyis thatofconcentric

    betweentwoconcentricspheresclose

    Measure-PreservingHomeomorphismsandMetri

    Math.(2),42,874-920(1941).

  • 7/26/2019 Wiener Cynernetique

    61/194

    LMECHANICS61

    dbytakingasonethe totalvolumeof

    spheres,willgiveinthe limitamea

    eofasphere.

    easureona regioninphasespacefor

    ntum,andtotalmomentofmomentum

    supposethat therearenoother

    ntitiesinthesystem.Letthetotal

    egionbe constant,oraswecan make

    e.Asour measurehasbeenobtained

    ntime,in awayinvariantintime,

    eshallcall thismeasurephase

    kenwithrespecttoit phaseaverages.

    ryingintimemayalsohavea time

    "(/)dependson/,itstime averagefor

    future:

    nics,bothtime averagesandspace

    rilliantideaofGibbsto trytoshow

    eragewere,insomesense,thesame.

    otypesofaveragewererelated,Gibbs

    nthemethodbywhich hetriedto

    utterlyandhopelesslywrong.For

    me.Evenatthetime ofhisdeath,

    ntegralhad justbeguntopenetrateto

    enyears,itwas amuseumcuriosity,

    ngmathematicianstheneedsand

    hematicianasdistinguishedasW.F.

    ingto dowithittill hisdyingday.

    whena groupofmathematicians

    Birkhoff2,finallyestablishedthe

    sgood'searlyworkrepresentedanimportantstep

    esgueIntegral.

    orie.Erg.d. Math.(1936).

  • 7/26/2019 Wiener Cynernetique

    62/194

    Gibbsstatisticalmechanics.Weshall

    ationswere,in thestudyofergodic

    ina systemfromwhichallthe

    vedasextracoordinates,almostall

    pacepassedthroughallcoordinates

    thesishecalledtheergodic hypothesis,

    v,work,and 6'coe,path.Now,

    herel,andothershaveshown,there

    rethathypothesisistrue.No diffe-

    area intheplane,evenif itisof

    rsofGibbs,includingat theend

    wthisin avagueway,andreplaced

    si-ergodichypothesis,whichmerely

    oftime,asystemgenerallypassesinde

    n theregionofphase-spacedeter

    ants.Thereis nologicaldifficulty

    merelyquiteinadequatefor the

    seson it.Itsaysnothingaboutthe

    temspendsin theneighborhoodof

    ageandof measuretheaverage

    ononeoverasetto bemeasuredand

    were mosturgentlyneededtomake

    n ordertoappreciatetherealsignifi

    eneed amorepreciseanalysisofthe

    as thenotionoftransformationgroup.

    lyfamiliartoGibbs, ashisstudyof

    vertheless,itispossibleto maintain

    mattheirfull philosophicalvalue.

    viside,Gibbsisoneof thescientists

    ticalacumenoftenoutstripshislogic,

    whileheisoften unabletoexplain

    ence,it isnecessarythatthere

    notstandisolated.In aworldruled

    performedbyanirrationalGod

    we shouldbeforcedtoawait each

    ofperplexedpassiveness.Wehave

  • 7/26/2019 Wiener Cynernetique

    63/194

    LMECHANICS63

    thecroquetgameinAlice inWon-

    sareflamingos;theballs,hedgehogs,

    abouttheirownbusiness;thehoops,

    wisesubjecttolocomotorinitiative

    sarethedecreesof thetesty,unpre

    erule foragameora usefullawof

    bleinadvance,andthatit applyto

    y,it shouldrepresentapropertyof

    hremainsthesame underthefluxof

    nthesimplestcase,it isaproperty

    oftransformationstowhich thesystem

    tothenotionsof transformation,

    nvariant.

    mis somealterationinwhicheach

    .Themodificationofthesolar system

    tionbetweentime/Aandtimetz isa

    ofcoordinatesofthe planets.The

    dinateswhen wemovetheirorigin,

    esto arotation,isatransformation.

    occurswhenweexamineapreparation

    onofa microscopeislikewiseatrans

    ansformationAbyatransforma

    mation,knownastheproductorresul

    ralitdependsonthe orderofA

    sformationwhichtakesthecoordi

    .andyintoo;,whilezisunchanged;

    o x,andyis unchanged;

    , y intozandzintox ;whileAB

    x , andzintoy . IfABandBA

    ythatAand Barepermutable.

    s,thetransformationAwill not

    fthesysteminto anelement,butwill

    ryelementisthe resultoftransfor

    ase,thereisaunique transformation

    ndA'*Aare thatveryspecialtransfor

    eidentitytransformation,whichtrans

    tself.Inthis casewecall A"1the

  • 7/26/2019 Wiener Cynernetique

    64/194

    tAisthe inverseo fA" 1, tha tIis its

    verse ofABis B^A1.

    transformationswhereeverytrans

    sethasan inverse,likewisebelonging

    esultantofany twotransformations

    belongstotheset.Thesesets are

    groups.Thesetofalltranslations

    orin athree-dimensionalspace,is

    ndevenmore,itis atransformation

    nownasAbelian,whereanytwotrans

    epermutable.Thesetof rotations

    otionsofarigid bodyinspace,are

    vesomequantityattachedtoallthe

    atransformationgroup.Ifthisquan

    h elementischangedbythesame

    p,whateverthattransformationmay

    ofthegroup.There aremanysorts

    fwhich twoareespeciallyimpor

    inearinvariants.Lettheelements

    groupbethe termswhichwerepre

    complex-valuedfunctionofthese

    ropriatepropertiesofcontinuityor

    ndsforthe elementresultingfrom

    nT,andif f(x)isafunctionof abso

    ofabsolutevalue1dependingonlyon

    characterofthegroup.It isan

    slightlygeneralizedsense.Iff(x)

    ers,clearlyf(x) g(x)isonealso,as

    esentanyfunctionh(x)definedover

    binationofthecharactersofthe group,

  • 7/26/2019 Wiener Cynernetique

    65/194

    LMECHANICS65

    rofthegroup,and a/r(T)bearsthesame

    oesto f(x)in(2.03),then

    x)intermsofa setofgroupcharac

    for allTin termsofthecharacters.

    acters ofagroupgenerateother

    ationandinversion,anditmaysimi

    stant1isa character.Multiplication

    generatesatransformationgroupof

    selves,whichisknownas thecha

    group.

    translationgroupontheinfinite

    Tchangesxintox +T,(2.03)be

    r),

    ''*,a(T)=eiVr.Thecharacterswillbe

    charactergroup willbethe groupof

    ox+ T,thushavingthesamestructure

    willnotbethe casewhentheoriginal

    onsabout acircle.Inthis case,the

    anumberbetween0and2*, differing

    multipleof2*, andwhile(2.06)will

    raconditionthat

    before,weshallobtain:

    bearealinteger,positive,negative,or

    huscorrespondsto thetranslations

    heotherhandthe originalgroupis

    theintegers,xandT in(2.05)arecon

    de'Xx involvesonlythenumberbet

    rsfromxby anintegralmultipleof2*.

    essentiallythegroupofrotations

    a givencharacter/",thevaluesof

  • 7/26/2019 Wiener Cynernetique

    66/194

    h awaythatthedistribution isnot

    multipliedby

  • 7/26/2019 Wiener Cynernetique

    67/194

    MECHANICS67

    ofitsmetricalinvariants.Theseare

    measurewhichdo notundergoany

    ansformedbythe grouparepermu

    egroup.Inthis connection,weshould

    ofgroup measure,duetoHaar1.

    oupitselfisa collectionofobjects

    ngmultipliedby theoperationsof

    mayhavean invariantmeasure.

    rtainratherwideclassof groupsdoes

    inedinvariantmeasure,definablein

    hegroupitself.

    ationofthetheoryof themetrical

    ansformationsistoshowthejustification

    ofphaseaveragesandtimeaverages

    yseen,Gibbstriedin vaintoestablish.

    sbeenaccomplishedisknown asthe

    emsstartwithan ensembleE,

    measure1, transformedintoitself

    ransformationT,orbyagroup of

    sformationsT*,whereoo

  • 7/26/2019 Wiener Cynernetique

    68/194

    )rfx,

    ergesinthemeantoa limitf*(x)as

    spectively,in thesensethat:

    0.

    ereergodictheoremofBirkhoff,

    ;whichmeansthat

    x)aredeflnedasin(2.16)and(2.17).

    at,exceptfor asetofvaluesof x

    theso-calledergodicormetrically

    transformationT,orthesetof

    sinvariantnosetofpointsx which

    or0.In suchacase,the setofvalues

    m)forwhich f*(x)takesona certain

    waj's either1or0. Thisisimpossible

    sconstant. Thevaluewhichf*(x)

    ys,is

    eorem,wehave

  • 7/26/2019 Wiener Cynernetique

    69/194

    MECHANICS69

    m,wehave

    ofa:ofzero measureorprobability0.

    continuouscase.Thisisan adequate

    changeofphaseaveragesand time

    TorthetransformationgroupT> is

    hasshownunder verygeneralcondi

    ucedtoergodiccomponents.Thatis,

    ofa;ofzero measure,Ecanbeseparated

    blesetofclassesEnand acontinuum

    measureisestablishedon eachEand

    derTorT>. Thesetransformationsare

    eintersectionof SwithE(y)and Snwith

    asure(S(y))dy +Vmeasure(Sn).

    heoryofmeasure-preservingtransfor

    the theoryofergodictransformations.

    y,let usremarkinpassing, may

    nsformationsmoregeneralthan

    ranslationgroupon theline.In

    dtothe translationgroupinvjdimen

    mensionsisphysicallyimportant.

    mporalequilibriumisspatial homoge

    thatofthehomogeneousgas,liquid,or

    tionof three-dimensionalergodic

    ergodicgroupoftranslationtransfor

    nsappearsasthe setoftranslationsofa

    uch thatoneoranotherexistsat agiven

    h.

  • 7/26/2019 Wiener Cynernetique

    70/194

    sofstatisticalmechanics,which

    onintheclassicalthermodynamics,is

    rilyapropertyofregionsin phase-

    ogarithmoftheirprobabilitymeasure.

    erthedynamicsofn particlesina

    ts,Aand B.Ifmparticlesare inA,

    racterizedaregioninphase-space,

    robabilitymeasure.Thelogarithm

    bution:mparticlesinA, n-minB.

    tof itstimeina statenearthatof

    nsethatfor mostofthetime,nearlyint

    yn-n^inB,wheretheprobabilityof

    -min Bisa maximum.Forsystems

    rticles,andstateswithinthelimitsof

    hismeansthatifwetakeastate of

    opy,andobservewhathappenstoit,

    sincreases.

    amicproblemsoftheheatengine,

    onsinwhichwe havearoughthermal

    slikean enginecylinder.Thestates

    opyare statesinvolvingmaximum

    atureandvolume,fora smallnumber

    umesandatthegiventemperature

    efineddiscussionsofthermalengines,

    inesliketheturbine,in whichagasis

    licatedmannerthanina cylinder,do

    nstooradically.Wemaystilltalk of

    veryfairapproximationeventhough

    ydeterminedexceptinastateof equi

    volvingthisequilibrium.However,in

    hofeventhis roughhomogeneity.The

    eas shownbytheelectronmicroscope

    essandfinenessoftexture,andits

    correspondingfinenessoftexture.

    thanthatofthespace-and-timescale

    er,andsothetemperaturesreadby

    ivingtissuesaregrossaverages,and

    ofthermodynamics.Gibbsianstatis

    beafairlyadequatemodelof what

  • 7/26/2019 Wiener Cynernetique

    71/194

    ALMECHANICS71

    picturesuggestedbytheordinary

    t.Thethermalefficiencyofmuscle

    ng,andcertainlydoesnotmean

    tisticalmechanicsisthat ofthe

    pposeagasin whichtheparticles

    edistributionof velocitiesinstatisti

    emperature.Fora perfectgas,thisis

    etthisgasbe containedinarigid

    sit,containinganopeningspanned

    y agatekeeper,eitherananthropo

    emechanism.Whenaparticleof

    tyapproachesthegatefromcompart

    sthanaveragevelocityapproachesthe

    ,thegatekeeperopensthegate,and

    h;butwhena particleoflessthanave

    fromcompartmentAoraparticleof

    cityapproachesfromcompartmentB,

    way,theconcentrationofparticles

    din compartmentBandisdecreased

    oducesanapparentdecreasein

    compartmentsarenowconnectedby

    obtainaperpetual-motionmachine

    estionposedbythe Maxwell

    othingis easierthantodenythe

    rstructures.Weshall actuallyfind

    estrictestsensecannotexisti na

    fwe acceptthisfromthebeginning,

    rateit,weshallmissan admirable

    thingaboutentropyandabout

    al,andbiologicalsystems.

    t,it mustreceiveinformationfrom

    ncerningtheirvelocityandpointof

    ertheseimpulsesinvolvea transfer

    involveacouplingofthedemon

    oftheincreaseofentropyapplies

    stem,butdoesnotapply toanon-

    tem.Accordingly,theonlyentropy

  • 7/26/2019 Wiener Cynernetique

    72/194

    ofthesystemgas-demon,andnot

    gasentropyismerelyoneterm

    argersystem.Canwefind terms

    lwhich contributetothistotal

    demoncan onlyactoninforma

    rmation,aswe shallseeinthe next

    ativeentropy.Theinformationmust

    lprocess,say someformofradiation.

    sinformationiscarried ataverylow

    ransferof energybetweenparticle

    erabletimefarlesssignificantthan

    .However,underthequantum

    toobtain anyinformationgivingthe

    ofaparticle, muchlessthetwo

    eeffectonthe energyoftheparticle

    nimumdependentonthefrequencyof

    ation.Thusallcouplingisstrictlya

    ;andasysteminstatistical equilibrium

    attersconcerningentropyandthose

    ongrun, theMaxwelldemonisitself

    ncorrespondingto thetemperatureof

    bnizsaysof someofhismonads,it

    small impressions,untilitfallsinto

    ndisincapableofclear perceptions.

    Maxwelldemon.

    beaquiteappreciableintervaloftime

    ditioned,andthistime maybeso

    akofthe activephaseofthedemon

    reasonto supposethatmetastable

    st;indeed,itmaywellbe thatenzymes

    emons,decreasingentropy,perhaps

    eenfastand slowparticles,butby

    cess.Wemaywellregardliving

    mself,inthis light.Certainlythe

    nismarealikemetastable:thestable

    edeconditioned,andthestablestate

    edead.Allcatalysts areultimately

    esof reaction,butnottrueequili-

  • 7/26/2019 Wiener Cynernetique

    73/194

    ALMECHANICS73

    lystsandManalikehavesufficiently

    bilitytodeservetherecognitionofthese

    nentconditions.

    hapterwithoutindicatingthat

    rablywidersubjectthanwehave

    ecertainmoderndevelopmentsof

    emeasuretobe keptinvariantunder

    efineddirectlyby thesetitself,

    vance.Ireferespeciallyto thework

    f,andtosomeofthe workofHure-

    ool.

    dtothe statisticalmechanicsof

    field inwhichtheconditionsare

    thestatisticalmechanicsofheatengines

    suitedtoserveasa modelofwhat

    ism.

  • 7/26/2019 Wiener Cynernetique

    74/194

    ON,

    enomenainwhichwhat isobserved

    asequenceofnumericalquantities,

    mperatureasrecordedby aconti

    eter;ortheclosing quotationsofa

    akenday byday;orthecompleteset of

    shedfromdaytodaybythe Weather

    s,continuousordiscrete,simpleor

    sarerelativelyslowlychanging,and

    entemployinghandcomputationor

    uchasslide-rulesand computing

    ongstothemoreconventionalparts

    edis thattherapidlychanging

    telephoneline oratelevisioncircuit

    tusbelongjustastruly tothefieldof

    lthoughtheapparatusby meansof

    ndmodifiedmustin generalbevery

    ctmustbe abletoputout resultspan

    alterationsofinput.Thesepiecesof

    eceivers,wavefilters,automaticsound

    oderof theBellTelephoneLabora

    tingnetworksandtheircorresponding

    sencequick-actingarithmeticaldevices,

    eapparatusofcomputingmachines

    ffof computers,ofthestatistical

    neededintheirusehas beenbuilt

    s ithasintothe automaticrange-finders

    -aircraftfirecontrol system;andfor

  • 7/26/2019 Wiener Cynernetique

    75/194

    N,ANDCOMMUNICATION75

    nof operationhastoworktoo

    nlinks.

    dtheapparatustodeal withthem

    aboratoryorin thetelephonecircuit,

    rding,preservation,transmission,and

    this information,andhowisit

    plest,mustunitary formsofinforma

    oicebetweentwo equallyprobable

    theother ofwhichisbound to

    example,betweenheadsandtailsin

    hallcall asinglechoiceof thissort

    or theamountofinformationinthe

    mentofaquantityknownto liebetween

    iforma prioriprobabilitylieany

    a l lsee tha t i fweputA=0andB=1,

    inthe binaryscalebytheinfinite

    ... .whereal, a2,....,eachhas the

    berofchoicesmadeand the

    ormationisinfinite.Here

    +a2++an+...

    ntwhichweactuallymakeis per

    on.If themeasurementhasauni

    ngovera rangeoflengthblb2... bn....

    ot equalto0,we shallseethatall the

    andpossiblytoa*,are significant,

    sarenot.Thenumberof decisions

    rom:

    "

    ntityastheprecise formulaforthe

    ditsdefinition.

    hefollowingway:weknow opriori

    n0 and1,anda posteriorithatitlies

    e(0,1).Thenthe amountofinforma

    sterioriknowledgeis

  • 7/26/2019 Wiener Cynernetique

    76/194

    era casewhereouraprioriknowledge

    acertainquantityshouldlie between

    the aposterioriprobabilityisf2(x)dx.

    ndoesoura posterioriprobability

    thatof attachingawidthtothe

    =fl(x)andy=/2(*).It willbenoted

    gthevariabletohaveafundamental

    resultswillnotin generalbethe

    ,oranyother functionofx.Since

    ,we shallhave:

    hmofthe breadthoftheregionunder

    somesort ofaverageoftheheightof

    calof fl(x).Thusareasonablemea

    ormationassociatedwiththecurve

    )dx.

    eas amountofinformationisthe

    uallydefinedasentropyin similar

    eregivenisnot theonegivenby

    problems,althoughitisa statistical

    to replaceFisher'sdefinitioninthe

    stantover(a, b),andiszero

    -^-^log, ^-=log,^^

    informationthatapointisin the

    he measureofthedifference,

    =log

    se ofapersonalcommunicationofJ.von