T132uA129147u149128153147D134134 - PDF document

1 T„uA§¶“u•±€
T„uA§¶“u•±€§™“D†ê†•uI•o†€€u§u•huJonathanWeisbergUniversityofTorontoAf«±§Zh±.Iarguethattherationalebehindthene-tuningargumentfordesignisself-undermining,refutingtheargument'sownpremisethatne-tuningistobeexpectedgivendesign.In(Weisberg,óþÕþ)Iarguedoninformalgroundsthatthispremiseisunsupported.White(óþÕÕ)counteredthatitcanbederivedfromthreeplausibleassumptions.ButWhite'sthirdassumptionisbasedonafallaciousratio-nale,andisevenobjectionablebythedesigntheorist'sownlights.eargumentthatshowsthis,theargumentfromdivineindišerence,simultaneouslyexposesthene-tuningargument'sself-underminingcharacter.esameargumentalsoanswersBradley's(forthcoming)replytomyearlierobjection.T„une-tuningargumentfordesignrestsonarelativelynewdiscoveryincos-mology:thatouruniverse'sconstantsandinitialconditionsareprecariouslybalancedtoallowfortheexistenceofintelligentlife.Outofthewiderangethesevaluescouldhavetaken,onlyasmallsubsetyieldauniversecapableofsupportingintelligentlife.Andyettheactualvaluesdolieinthatsmallsubset.isdiscoveryissurprisingifouruniversewasnotdesigned.But,theargumentalleges,itistobeexpectedifouruniversewascreatedbyadesignerintentoncreatingintelligentlife.usthediscoveryofne-tuningtsbetterwiththedesignhypothesisthanwithitsnegation.LetDbethedesignhypothesisandNthenewdiscoverythatouruniverseisne-tuned.eargumentturnsoncomparingtheprobabilitiesthatDand Deachconferonthenewevidence,N.Accordingtotheargument,pˆNSDApˆNS D,soNsupportsDover D.Butinmy(óþÕþ)Iworriedthatthiscomparisonoverlooksanoldpiecebackgroundknowledge,thatlifeexists.LettingObetheoldnewsthatlifeexists,thecorrectstatementofthene-tuningargumentis:pˆNSD,OApˆNS D,O.(Õ)eLikelihoodPrinciple:ifpˆESHApˆES HthenEsupportsHover H.(ó)SoNsupportsDover D(givenO).(ì) IamgratefultoDarrenBradley,LisaCassell,KennyEaswaran,ChrisMeacham,RogerWhite,ananonymousreferee,andaudiencemembersatAmherstCollegefortheirhelpfulfeedbackandcriticism.Õ óJ™•Z±„Z•Wu†«fu§Myobjectionwasthatpremise(Õ)isnotcompelling.Wehaveknownformanyyearsthatouruniversecontainsintelligentlife,andthusthatthecons

2 tantsandinitialcon-ditionshadtobeinthera
tantsandinitialcon-ditionshadtobeintherangenecessarytosupportsuchlife.Whatwedidnotknowwaswhetherthatrangewaswideornarrow.InWhite's(óþÕÕ)helpfulterminology,whatwedidnotknowwaswhetherthelawsofouruniverseareªstringentºorªlaxº.Wehavenewlylearnedthattheyarestringent.But,theobjectiongoes,thisisnotsomethingwehavereasontoexpectatthehandsofadesigner,sinceshecouldhavechosenlaxlawsinstead.FollowingWhite,letSbethefactthatthelawsofouruniversearestringent,i.e.thattheywillonlysupportintelligentlifeonafewsettingsoftheconstantsandinitialconditions.SandNareequivalentgivenO,so(Õ)isequivalentto:pˆSSD,OApˆSS D,O.(Õ*)Myobjectionwasthatwehavenoreasontoaccept(Õ*),sincewehavenoreasontothinkthatadesignerwouldchoosestringentlawsasherwayofcreatingintelligentlife.Shecouldeasilyhavechosenlaxlawsasameansofcreatingintelligentlife.Whiterepliesthat(Õ*)canbederivedfromthreeplausibleassumptions.erstisthatstringencyandlife'sexistencearenegativelydependentifwesupposethereisnodesigner:pˆOSS, D@pˆOS S, D.(¦)Ifthereisnodesigner,stringentlawsmakelifelesslikely.Second,stringencyandlife'sexistenceareindependentontheassumptionthatthereisadesigner:pˆOSS,D�pˆOS S,D.(¢)Ifthereisadesigner,shewillcreatelifecomewhatmay.Andthird:pˆDSSCpˆDS S.(ä)Insupportof(ä)Whitesays:[...]thefactthatthelawsputstringentconditionsonlifedoesnotbyitselfprovideanyevidenceagainstdesign[...]Ofcourseitispossiblethatadesignerhasapreferenceforlawsthatputstringentconditionsonlife'sexistence,orapreferenceforlaxconditions.Butaswehavenoreasontosuspectsoeitherway,SbyitselfhasnobearingonD.(White,óþÕÕ,p.äߘ)But(ä)isnotsupportedbythisrationale;indeed,thedesigntheorist'sownreasonsfor(Õ)actuallytellagainst(ä).IwillrstdescribeacasethatunderminesWhite's T„uA§¶“u•±€§™“D†ê†•uI•o†€€u§u•huìrationalefor(ä).enI'llarguethatthedesigntheorist'sownreasoningactuallyrefutes(ä),andeven(Õ).SupposeÕþþprisonersaresentencedtodeath,halfhousedincellblockAandhalfincellblockB.elawrequiresthatexactlyoneprisonerbepardoned,andtheluckyprisonerwillbeselectedeitherbyrandomlotteryorbyajudgewhowillbeappointedtomakethedecision.Ifappointed,thejudgewillpardonsomeon

3 ewhoisinnocent.Wehavenoreasontothinkthej
ewhoisinnocent.Wehavenoreasontothinkthejudgecareswherethepardonedprisonerishoused,butasithappensthereareÉinnocentprisonersincellblockA,andonlyÕincellblockB.Itiskeptsecrethowtheluckypardoneeisselected.NowsupposewelearnthatthepardonedprisonerwashousedincellblockB.isdiscoveryhasanegativebearingonthehypothesisthatthejudgewasappointed.ForshewasÉtimesmorelikelytopardonaprisonerfromcellblockAthanfromcellblockB,whereasitwasa¢þ/¢þshotatthehandsofchance.So,eventhoughwehavenoreasontothinkthejudgehasanypreferenceaboutwherethepardoneeishoused,wherethepardoneewashousedstillbearsonthehypothesisthatthejudgemadethedecision.So,thatwehavenoreasontosuspectadesignerwouldhaveanypreferencebetweenstringentandlaxlawsdoesnotshowthatScannotbearnegativelyonD.isshowsthat(ä)isinadequatelysupported,butitalsosuggeststhat(ä)isfalse.JustasthepardonedprisonerbeingfromcellblockBhasanegativebearingonthehypothesisthatthejudgeissuedthepardon,thesuppositionthatouruniverse'slawsarestringentmayhaveanegativebearingonthehypothesisthatitisdesigned.Foradesignerintentoncreatinglifeismorelikelytochooseoneoftheplentifullaxoptions,justasthejudgeismorelikelytochooseaprisonerfromthecellblockwithmoreinnocentprisoners.isisasurprisingsuggestion,as(ä)lookedplausibleenoughonitsface.Soratherthanrelyonanalogy,Iwillpresentanexplicitargumenttothisešect:bythedesigntheorist'sownlights,both(ä)and(Õ*)arefalse.eargument'smainideacanbeconveyedpictorially.Picturethespaceofpossibleuniversesarrangedinalineaccordingtothestrictnessoftheirlaws,withstrictnessincreasingtotheleŸ: «±§†h± Zì O Y Y Y Y Y Y Y Y Y Y Y Y edotsrepresentthelife-supportinguniverses,thosewhereOholds.Asthepicturesuggests,thesebecomemorecommonaswemovetotheright;bydenition,laxerlawsyieldalife-supportinguniverseonmoreofthepossiblesettingsoftheconstants ¦J™•Z±„Z•Wu†«fu§andinitialconditions.NowconsiderhowprobabilitiesaredistributedoverthesepossibleuniversessupposingDandsupposing D.Ifthereisnodesigner,itisamatterofªblindchanceºwhichuniverseisactual,resultinginauniformdistributionoverthewholeline;chanceªthrowsadartºattheinterval.Ifthereisadesignerthough,onlythedotsarelivepossibilities,sincethepostulateddesignerisintentoncreatinglife.Andsinceweknownothingmoreaboutheraimsandmethodst

4 hanthis,auniformdistributionoverthedotsr
hanthis,auniformdistributionoverthedotsrepresentsourexpectations.Nowthepunchline:chance'sdartismorelikelytolandtowardstheleŸendofthespectrumthanthedesigner's.edesignerwillªlandherdartºononeofthedots,sohersismorelikelytolandclosertotherightthantotheleŸ.Butifthereisnodesigner,thedartmaylandanywhereintheline,makingitmorelikelythanthedesigner'stolandtowardstheleŸ.Contra(ä)then,strictnessisnegativelyrelevanttodesign.Wecangofurtherandseewhy(Õ*)isfalsetoo.SupposewelearnthatthedartlandedononeoftheO-possibilities.isdoesnotchangethewayprobabilitiesaredistributedsupposingdesign;givendesign,wealreadyknewanO-possibilitywouldbehit.Supposing Dthoughtheprobabilitiesdochange:toexactlythesameprobabilitieswegetsupposingD.Conditionalizingauniformprobabilitydistributionresultsinauniformdistributionovertheremainingpossibilities.Sonow,given D,eachO-possibilityhasequalprobabilityofbeinghit.us,onceweknowO,theprobabilityofselectingapointtowardsthestrictendofthespectrumisthesamegivenDandgiven D,contra(Õ*).Let'snowmaketheargumentrigorous.efollowingthreeassumptionsshouldbeacceptabletotheproponentoftheoriginalne-tuningargument.First:DivineIntent:pˆOSD�Õ.ejusticationhereisthesameasforWhite's(¢):thepostulateddesignerisintentoncreatinglife,and(wemaysuppose)canbecountedontodosocomewhatmay.Second:BlindIndišerence:pˆ�SO, DisauniformdistributionovertheO-possibilities,wherepˆ�SO, DistheprobabilityfunctionobtainedbyconditionalizingponO, D.BlindIndišerenceisjustiedbythedesigntheorist'sownrationaleforsayingthatpˆSS D,Oislowandthusthat(Õ)istrue.Ifthereisnodesigner,itisamatterofªblindchanceºhowtheworldturnsouttobe,sopˆ�S Disauniformdistributionoverallpossiblecosmologies.ismakespˆ�SO, DauniformdistributionoverthepossiblecosmologieswhereOholds.ethirdpremiseis: T„uA§¶“u•±€§™“D†ê†•uI•o†€€u§u•hu¢DivineIndišerence:pˆ�SO,DisauniformdistributionovertheO-possibilities.DivineIndišerenceismotivatedbythethoughtthat,absentanyinformationorstipulationaboutthedesigner,savethatshewillcreateoneoftheO-possibilities,eachO-possibilityshouldberegardedasequallyprobable.Mo

5 reneedstobesaidaboutDivineIndišeren
reneedstobesaidaboutDivineIndišerenceandwewillreturntothematterinamoment.Firstletusseehowtheseassumptionsrefute(ä)and(Õ*).Together,DivineIndišerenceandBlindIndišerenceentailthatOªscreensošºD(and D)fromtheO-possibilities.OnceOisgiven,supposingD(or D)hasnoešectonthewayprobabilitiesaredistributedovertheO-possibilities.uswehave:DivineIrrelevance:pˆXSO,D�pˆXSOforanyXthatisaunionofO-possibilities.Wecanthenderive:pˆDSS�pˆDpˆSSD pˆSbyBayes'eorem(~ä)�pˆDpˆSSO,D pˆSbyDivineIntent�pˆDpˆSSO pˆSbyDivineIrrelevance�pˆDpˆOSS pˆObyprobabilitycalculus@pˆD.bythedefn.ofSWecanalsoderivedirectlyfromDivineIndišerenceandBlindIndišerence:pˆSSD,O�pˆSS D,O.(~Õ*)eseresultsvindicatemyearlierobjectionto(Õ)intworespects:acrucialassumptioninthederivationof(Õ*)isfalse,andsois(Õ*)itself.eseresultsalsoanswerBradley's(forthcoming)replytomyobjection.WhileIthinkSdoesnotsupportD,IdoallowthatOmayošersomesupportforDinitially;it'sjustthatSošersnoadditionalsupport.But,Bradleyobserves,theamountofsupportOlendstoDdependsonwhetherSor Sistrue.Plausibly,thedišerencebetweenpˆOSD,SandpˆOS D,SissignicantlygreaterthanthatbetweenpˆOSD, SandpˆOS D, S,sothatOošerssignicantlygreatersupporttoDgivenSthangiven S.usne-tuningdoessupportdesign,justindirectly,byamplifyingthesupportfromouroldevidenceO.WhileBradleymayberightthatlearningSampliestheevidentialsupportOlendstoD,thisdoesnotmeanthatlearningSinadditiontoOincreasesthenetsupport äJ™•Z±„Z•Wu†«fu§forD.ForSmaysimultaneouslybeevidenceagainstD,sothattheamplicationofO'ssupportisdrownedoutbythedisconrmationešectedbyS.Infact,theaboveargumenttellsusthatthisisexactlywhathappens.(~ä)showsthatStellsagainstD.And(~Õ*)tellsusthatthisdisconrmationofDexactlybalancesouttheamplicationofO'ssupport,sincelearningSaŸerlearningOneitherincreasesnordecreasestheprobabilityofD.Let'snowreturntoDivineIndišerencewithamorecriticaleye.Indišerence-basedreasoningisnotorious

6 lyproblematic,soit'snaturaltowonderwheth
lyproblematic,soit'snaturaltowonderwhethertheaboveargumentusesitillicitly.ereareseveralworrieshere.OneworryisthattheuniformprobabilitydistributionpositedbyDivineIndišer-ence(andBlindIndišerence)doesnotexist,sincethespaceofrelevantpossibilitiesisunbounded.erangeofpossiblelaws,constants,andinitialconditionsisnotbounded,soanypositiveuniformdistributionoveritwillbeimproper.Butthisisaproblemfortheproponentofthene-tuningargumenttosolve,sincesheassumesauniformdistributionoverthespaceofpossibleuniversestomotivatepremise(Õ)(Colyvanetal.,óþþ¢).Howevershesolvesit(perhapsbyre-parameterizingthespacetotanitearea,orbyimposinganitepartitionwhereeachcellgetsequalpriorprobability),wecanadopthersolutiontosaythatpˆ�SO,DisauniformdistributionoverthesubsetofpossibilitieswhereO.AnotherworryarisesinconnectionwithBertrand'sparadox.Auniformdistribu-tionoveranuncountablesetparameterizedonewaywillnotbeuniformunderallalternativeparameterizationsofthesameset.WhatparameterizationispresupposedbyDivineIndišerence?isagainisaproblemfortheproponentofthene-tuningargumenttosolve.Whateverparameterizationsheusestomotivatepremise(Õ)ofherargument,DivineIndišerenceistobeinterpretedusingit.Itisimportanttonote,however,thatherparameterizationcannotbethesortusuallypresupposedinstatementsofthene-tuningargument.esestatementsonlyprovideaparameteri-zationofthespaceofpossibleconstantsandinitialconditionsfortheactuallawsofouruniverse.Butweareassessingtheimportofthediscoverythatthesearetheactuallaws.Soweneedaparameterizationofthespaceofallpossiblelaws,notjustofthespaceofpossibleconstantsandinitialconditionsforourlaws.Itisuptothedesigntheoristtoprovidesuchaparameterizationifshewishestohaveanargumentatall,sinceshemustprovidesomereasonforthinking(Õ*)plausible.Presumably,shewillprovidethisreasonbypresentinguswithanaturalparameterizationonwhichStakesupasmallportionofthespaceofpossibilities.BlindIndišerencethenjustacceptsthisparameterization,sayingthatpˆ�S Disauniformdistributionoverit,andthusthatpˆ�SO, DisauniformdistributionovertheO-possibilities. T„uA§¶“u•±€§™“D†ê†•uI•o†€€u§u•hußAnalworryisthatthedesigntheoristmightobj

7 ecttoauniformdistributionoverthespaceofp
ecttoauniformdistributionoverthespaceofpossiblelawsgivendesign.Considerthejudge:shemightpickaninnocentprisoneratrandom,butshemightinsteadžipacointosettleonacellblockandthenpickaprisoneratrandom(orbysomeothermeans).Inthatcase,wheretheprisonerwashousedhasnobearingonwhetherthejudgewasappointed.Similarly,thedesignermightdealwithherindišerenceaboutSvs. SbyžippingacoinandthenchoosingfromamongtheSpossibilities(or S,asthecasemaybe).Ofcoursewehavenoreasontosuspectthedesignerwouldusesuchamethod.Butthedesigntheoristmayarguethatindišerenceshouldbeappliedtothepossiblemethodsthedesignermightuse,ratherthantothepossibilitiesthesemethodsselectfrom.AndanaturalwaytopartitionandparameterizethesemethodsisbythechanceeachhasofresultinginS,yieldingÕ/óprobabilityforeachofSand S,givenD.iswon'thelpthedesigntheorist'scausethough.edivisionbetweenªstringentºandªlaxºlawswasanarticialsimplicationweadoptedforconvenience.Really,stringencycomesonacontinuum.SothenaturalparameterizationisþBxBÕ,wherexistheportionofthepossiblesettingsoftheconstantsandinitialconditionsthatcansupportintelligentlife.Andauniformdistributionoverthisparameter(or,whatcomestothesamething,auniformaverageofthepossibledistributionsoverit)willyieldthesameuniformdistributionpostulatedbyBlindIndišerence.For,ifthedesigntheoristthinksthisparameterizationisreasonablegivenD,itisreasonablegiven Dtoo.AŸerall,thestringencyofourlawsisthediscoveryinquestion,soauniformdistributionoverthepossibledegreesofstringencyisnaturalgivennodesigner.Tomakethisresponseworkthen,thedesigntheoristwouldhavetosupplyanddefendasecondparameterizationshewantstoapplyindišerencetogiven D.Untilshedoes,herargumentiscollapsingunderitsownweight.Forthenaturalextensionofherownthinkingunderminesherkeypremise,yieldinginsteadourargumentfor(~Õ*),theargumentfromdivineindišerence.Ru€u§u•hu«Bradley,Darren.forthcoming.Weisbergondesign:Whatne-tuning'sgottodowithit.Erkenntnis.Colyvan,Mark,JayL.Gareld&GrahamPriest.óþþ¢.Problemswiththeargumentfromnetuning.SyntheseÕ¦¢(ì):ìó¢±ìì˜.Weisberg,Jonathan.óþÕþ.Anoteondesign:What'sne-tuninggottodowithit?Analysisßþ(ì):¦ìÕ±˜.White,Roger.óþÕÕ.Whatne-tuning'sgottodowithit:Replytoweisberg.AnalysisßÕ(¦):äßä±