1

On Evil’s Vague Necessity 1. Introduction . Itisthestandardpositionontheofevilthataperfectcould notpermitevenoneinstanceofevilthatservesnopurpose.Amongthosewho havedefendedthestandardpositiononevilisWilliamRowe.

Anomniscient,whollygoodbeingwouldpreventtheoccurrence of any intense evil it could ,unlessitcouldnotdosowithout therebylosingsomegreatergoodorpermittingsomeevilequally badorworse. 1 AccordingtoPetervanInwagenthestandardpositionontheexistenceofevil isfalse.Aperfectbeingwouldpreventtheoccurrenceofanyintensesuffering itcouldonlyifthereexistssomeminimumamountofevilthatisnecessaryto thepurposesofaperfectbeing.Theminimumevilnecessaryfordivine purposesistheleastamountofevilsuchthatanygreaterevilisunnecessary fordivinepurposes. 2ButvanInwagenurgesthatitisimplausibletosuppose thatthereissome amountofevilthatistheminimumnecessary.

Itisnotveryplausibletosupposethatthereisawayinwhich evilcouldbedistributedsuchthat(i)thatdistributionofevil wouldserve’spurposesaswellasanydistributionofevil couldand(ii)God’spurposeswouldbelesswellservedbyany distributioninvolvinglessevil. 3 Itisplausibletosupposeinsteadthatforanyamountofevilsuchthatany greaterevilisunnecessaryforGod’spurposesthereissomelesseramountof evil suchthatanygreaterevilisunnecessaryforGod’spurposes.Callthat 2 the No Minimum Thesis .Thereisthereforenoleastamountofevilsuchthat anygreaterevilisunnecessaryfordivinepurposes.Accordingtovan

Inwagenthe No Minimum Thesis togetherwithsomeuncontroversial assumptionsentailsthatthestandardpositiononevilisfalse.Andwearrive attheincredibleconclusionthataperfectbeingmightexistalongwith pointlessandpreventableevil.

IpresentnextvanInwagen’s No Minimum argumentagainstthe standardpositiononevil.Ishowin(2.1)thatthepremisesintheargument cannotallbetruetogetherandthatthereforetheargumentoffersnoreason tobelievethatthestandardpositiononevilisfalse.Contrarytothe No

Minimum argumentthestandardpositiononevildoesnotentailthatthereis adiscretetransitionbetweenevilthatisunnecessaryfordivinepurposesand evilthatisnecessaryfordivinepurposes.In(2.2)Iprovideareformulationof the No Minimum argumentthatassumesagreatestlowerboundonevilthat isunnecessaryfordivinepurposes.Thereformulatedargumentalsocontains inconsistentpremises.Insection(2.3)Iofferthe Vague Minimum Thesis .The thesisguaranteesthatthereisnodiscretetransitionbetweenamountsofevil thatarenecessaryandamountsofevilthatareunnecessaryfordivine purposes.The Vague Minimum Thesis posesnothreattothestandard positiononevil.In(3)Iconsidersomealternativeformulationsofthe No

Minimum Thesis .Iarguethatthealternativeformulationsareconsistent withthe Vague Minimum Thesis anddonotadvancevanInwagen’s 3 argument.In(3.1)and(3.2)Ishowthatdegreetheoreticandepistemicist interpretationsofthe No Minimum argumentdonotadvancevanInwagen’s argumentagainstthestandardpositiononevil.In(4)Iconsiderthehigher orderofunnecessaryevilandshowthatittoopresentsnoproblem forthestandardposition.In(5)Iconsiderwhether,foreveryamountofevil k thatisunnecessaryfordivinepurposes,anomniscientbeingwouldknowthat kisunnecessaryfordivinepurposes.Iarguethatanomniscientbeingwould knowandconcludethattheknowledgeofomniscientposesnoproblem forthestandardpositiononevil.Ioffersomeclosingcommentsin(6).

2.Van Inwagen’s No Minimum Argument.

Supposethereisadivinepurposetobringingaboutaworldcontaining higherlevelsentientbeings.Perhapstheexistenceofhigherlevelsentient beingsisnecessarytosomeveryimportantmoralgood—amoralgoodthat outweighsallsentientsuffering.Itmightbetruethataseriesofmiraculous interruptionsinnaturallawcouldpreventeveryinstanceofsentient suffering.Butsupposethatpreventingeveryinstanceofsentientsuffering wouldproduceamoraldefect—thebadeffectsofamassiveirregularityin naturallaw—thatisatleastasgreatasanyofsufferingamong sentientbeings.Itmightthenbetruethataperfectbeingcouldpermitan amountofevilthatisnottheminimumevilnecessaryfordivinepurposes.

Hereisthe No Minimumargument.

Butwhatofthehundredsofmillions(atleast)ofinstances[of intensesufferingsimilartoRowe’sfawn]thathaveoccurred 4

duringthelonghistoryoflife?Well,Iconcede,Godcouldhave preventedanyoneofthem,oranytwoofthem,oranythreeof them...withoutthwartinganysignificantgoodorpermitting anysignificantevil.ButcouldHehavepreventedallofthem? No—notwithoutcausingtheworldtobemassivelyirregular. Andofcoursethereisnosharpcutoffpointbetweenaworldthat ismassivelyirregularandaworldthatisnot...Thereis, therefore,nominimumnumberofcasesofintensesufferingthat Godcouldallowwithoutforfeitingthegoodofaworldthatisnot massivelyirregular. 4 Butifthereisnominimumamountofevilnecessaryfordivinepurposes,van

Inwagenurges,wecannotconcludethatGodisunjustorcruelforpermitting morethantheleastamountofevilnecessaryforthosepurposes.

ButifthereisnominimumofevilthatwouldserveGod’s purposes,thenonecannotarguethatGodisunjustor cruelfornot“gettingbywithless”—anymorethanone canarguethatalawthatfinesmotorists$25.00forillegal parkingisunjustorcruelowingtothefactthatafineof $24.99wouldhaveanidenticaldeterrenteffect. 5 Accordingtothestandardpositiononevilaperfectbeingcannotpermitmore thantheminimumevilnecessaryfordivinepurposes.Everyinstanceofevil thatexceedstheminimumnecessaryis,ofcourse,unnecessaryorpointless.

Butaccordingtothe No Minimum argumentitistrueboththatsomeevilis necessaryfordivinepurposesandthatnoevilistheminimumnecessaryfor divinepurposes.Aperfectbeingcanactualizedivinepurposesonlyifhe 5 permitsmorethantheminimumevilnecessary.ButvanInwagenurgesthat certainlyaperfectbeingispermittedtoactualizethegreatgoodsinhis divinepurposes.Thereforethestandardpositiononevilismistaken.

Accordingtothe No Minimum argumentanomnipotentbeingmight allowpointlessandpreventableevilanddisplaynomoralimperfectionatall.

Andthatisagoodreasontosuspectthatthe No Minimum argumenthas gonewrong.Supposewereconsiderthepremisesintheargument.Let kbe thetotalamountofevilintheactualworldincludingeveryinstanceof intensesufferingamongsentientbeings.The No Minimum argument assumesthattheamountofevilin kissuchthatanygreateramountofevil isunnecessaryfordivinepurposes.IndeedvanInwagenproposesthe strongerassumptionthattheamountofevilin kexceedstheamount necessaryfordivinepurposes.Hesimplyconcedesthatmanyactual instancesofintensesentientsufferingdonotserveanydivinepurpose.The initialpremiseofthe No Minimum argumentstatesthefollowing.

1. Theactualamountofevil kissuchthatanygreateramountofevilis unnecessaryfordivinepurposes. Butaccordingtothe No Minimum argumenttheamountofevil kn isnotthe least amountofevilnecessaryfordivinepurposesonlyif,forsomepositive incrementinevil i,theamountofevil kn - i is nottheleast amountofevil necessaryfordivinepurposes.InfactvanInwagenfindsitaveryplausible assumptionthatforanyamountofevilthatwouldhaveservedGod’s 6 purposes,slightlylessevilwouldhaveservedHispurposesjustaswell. 6

Howeverbadtheactualworlditseemsunlikelythatthereisaninfinite amountofactualevil.Butlet’ssupposethattheamountofevilbetween0 and kisatleastinfinitelydivisible. 7Let ibesomeincrementinevilbetween

0and k.Thereisthenasequence Sintotalamountsofevilthatbeginsatthe actualamount kandendsat0andissuchthat k> k– i > k– 2i > k– 3i >...>

0. Thesecondpremiseintheargumentexpressesthe No Minimum Thesis.

2. Foranyamountofevil kn,( k>kn>0)if knissuchthatanygreater

amountofevilisunnecessaryfordivinepurposesthenforsome i( i >0) kn-

iissuchthatanygreateramountofevilisunnecessaryfordivine purposes. 8 The No Minimum Thesis expressesthepropositionthat,foreachamountof evil kn(k>kn>0)inthesequence S,if knissuchthatanygreateramountof evilisunnecessaryfordivinepurposesthen,forsomepositiveincrementin evil i,thesameistrueofthelesseramountofevil kn – i .

Butfrompremise(1)togetherwiththe No Minimum Thesis wecanderive premise(3).

3. Thereisnoamountofevil kn( k>kn>0)thatisthe least amountofevil suchthatanygreateramountofevilisunnecessaryfordivinepurposes. From(1)and(2). Let’sshowthat(3)followsfrompremise(1)andthe No Minimum Thesis .

Supposepremise(3)isfalse.Itfollowsthatthereissome least amountofevil kn( k>kn>0)suchanygreateramountofevilisunnecessaryfordivine 7 purposes.Butitfollowsfromthe No Minimum Thesis andpremise(1)that foreveryamountofevil kn( k>kn>0)thereisa lesser amountofevil kn – i suchthatanygreateramountisunnecessaryfordivinepurposes.Soour suppositionisfalse.Itfollowsthatpremise(3)istrue.

Theminimumevilnecessaryfordivinepurposesisjusttheleast amountofevilsuchthatanygreaterevilisunnecessaryfordivinepurposes.

Butfrompremise(3)weknowthatthereisnoleastamountofevilsuchthat anygreaterisunnecessaryfordivinepurposes.Andsowearriveatpremise

(4).

4. Thereisnominimumamountofevil kn( k>kn>0)necessaryfordivine purposes.Frompremise(3). NowaccordingtothethatvanInwagenoffers—astoryofferedas aplausibleelaborationonthedataofChristianrevelation—acertainamount ofsufferingandevilis necessary fordivinepurposes.Itisimpossible,for instance,forGodtopreventeveryinstanceofintensesufferingamong sentientbeingswithouttherebycausingamoraldefectthatisatleastasbad.

The No Minimum argumentthereforeassumesthatatleastsomeinstances ofevilarenecessaryfordivinepurposes.

5. Thereissomeamountofevil kn( k>kn>0)thatisnecessaryfordivine purposes. Assumption Weknowfrompremise(4)thatthereis no minimum amountofevil necessaryfordivinepurposes.Andweknowfrom(5)thatsomeamountof 8 evilisnecessaryfordivinepurposes.Nowcertainlyaperfectbeingwould permitenoughevilandsufferingtorealizedivinepurposes.Sowearriveat premise(6).

6. Aperfectbeingwouldpermitanamountofevil kn( k>kn>0)thatis sufficienttorealizedivinepurposes.Assumption. Anamountofevil kn( k>kn>0)issufficientfordivinepurposesjustincase thereisnogreateramountofevilnecessaryfordivinepurposes.Butforevery amountofevil kn( k>kn>0), kniseither less than theminimumnecessary fordivinepurposesor knmore than theminimumnecessaryfordivine purposes.Anyamountofevilthatislessthantheminimumnecessaryfor divinepurposesisanamountthatisinsufficienttorealizethosepurposes.

The No Minimum argumentthereforeconcludesthataperfectbeingwould permitanamountofevilthatisgreaterthantheminimumnecessaryforhis purposes.Frompremises(4),(5)and(6)wearriveatpremise(7).

7. Aperfectbeingwouldpermitsomeamountofevil kn( k>kn>0)greater thantheminimumamountofevilnecessaryfordivinepurposes.From (4),(5)and(6). Ofcourse,indefenseofthestandardpositiononevilitmightbeurgedthata perfectbeingsimplycouldnotpermitmorethantheminimumevilnecessary fordivinepurposes.Premise(7)wouldthenconstituteanabsurdconsequence oftheassumptionthattherearedivinepurposesforwhichthereisnoleast amountofevilnecessary.Butsupposewesetthisobjectionaside.Itfollows directlyfrompremise(7)thatthestandardpositiononevilismistaken. 9

8. Thereforethestandardpositiononevilisfalse.From(7) Thestandardpositiononevilstatesthataperfectbeingwouldpreventthe occurrenceofanypointlessevilitcould.Butaccordingto(7)aperfectbeing wouldpermitanamountofevilthatisgreaterthantheminimumnecessary fordivinepurposes.Butthenaperfectbeingwouldnotpreventthe occurrenceofallthepointlessevilitcould.Thestandardpositiononevilis thereforefalse.

2.1 Why the No Minimum Argument Fails

The No Minimum argumentisdesignedtoshowthatthestandardposition onevilisfalse.Premise(1)assumesthatanyamountofevilexceedingthe actualamountisunnecessaryfordivinepurposes.

1. Theactualamountofevil kissuchthatanygreateramountofevilis unnecessaryfordivinepurposes.

Andpremise(2)expressesthe No Minimum Thesis .

2. Foranyamountofevil kn,( k>kn>0)if knissuchthatanygreater amountofevilisunnecessaryfordivinepurposesthenforsomeincrement

i (i >0), kn- i issuchthatanygreateramountofevilisunnecessaryfor divinepurposes. Andsinceourtheodicyassumesthatsomeamountofevilisnecessaryfor divinepurposes,the No Minimum argumentalsoincludespremise(5).

5.Thereissomeamountofevil kn( k>kn>0)thatisnecessaryfordivine purposes. 10

Butthesepremisescannotallbetruetogether.Premises(1)and(2)entail thatpremise(5)isfalse.Weknowfrompremises(1)and(2)thatevery amountofevilin Sissuchthatanygreateramountofevilisunnecessaryfor divinepurposes.Supposefor reductio ad absurdum that(5)istrueand knis necessaryfordivinepurposes.Weknowthatthereissomeamountofevil kin

S,suchthat kisunnecessaryfordivinepurposes.Butfromrepeated applicationsofpremise(2)—theincrementsinpremise(2)arealways positive—itfollowsthat knisalsounnecessaryfordivinepurposes.That’s impossible.Thereforeoursuppositionisfalse.Itfollowsfrompremises(1) and(2)thatnoamountofevil—notsomuchasthescratchingofafinger—is necessaryfordivinepurposes.Socontraryto(5)noamountofevil kn( k>kn>

0)in Sissuchthat knisnecessaryfordivinepurposes.

Weknowthatpremises(1),(2)and(5)inthe No Minimum argument cannotallbetrue.The No Minimum argumentmustcontainatleastone falsepremise.Theargumentisthereforeunsound.Iconsiderin(2.2)an illuminatingreformulationofthe No Minimum argument.

2.2No Minimum Reformulated .

Accordingtothe No Minimum argumentthereissomeamountofevilthatis necessaryfordivinepurposes.Wecanstipulatewithoutlossofgenerality thatsomeamountorothergreaterthan50turpsofevilisnecessaryfor divinepurposes.Let’ssupposefurtherthatforanyamountofevil kngreater than50turpsthereissomeincrement i=½( kn–50 )suchthat kn – i serves 11 divinepurposesjustaswell.Wearriveattheconclusionthat some amount or othergreaterthan50turpsofevilisnecessaryfordivinepurposesandeach particularamountofevilgreaterthan50turpsisunnecessaryfordivine purposes.

Inthereformulated No Minimum argumentpremise(1)istruejustin casetheactualamountofevil kexceeds50turps.

1.Theactualamountofevil kissuchthatanygreateramountofevilis unnecessaryfordivinepurposes.

Premise(2)istruejustincaseforanyamountofevil kngreaterthan50turps thereissomeincrement i( i >0)suchthat kn – i isunnecessaryfordivine purposes.

2.Foranyamountofevil kn,( k>kn>0)if knissuchthatanygreater amountofevilisunnecessaryfordivinepurposesthenforsome i( i >0)

kn- i issuchthatanygreateramountofevilisunnecessaryfordivine purposes. Andpremise(5)istruejustincasethereissomeamountofevilgreaterthan

50turpsthatisnecessaryfordivinepurposes.

5.Thereissomeamountofevil kn( k>kn>0)thatisnecessaryfordivine purposes.

Butpremise(5)isfalseunderthecurrentassumptionthateach particular amountofevilgreaterthan50turpsisunnecessaryfordivinepurposes. 9

Sincethescopeoftheexistentialquantifierinpremise(5)iswideitasserts thatthereissomeparticularamountofevilthatisnecessaryfordivine 12 purposes.IndeedifvanInwagenisright,thenquitealargeamountofevilis necessaryfordivinepurposes.Thetheodicyheoffersisdesignedtoshowthe essential contributiontoGod’splanofagoodportionofactualevil.

Whatthetheodicistmustdo,giventhefactsofhistory,istosay whatcontribution—whatessentialcontribution—toGod’splan ofAtonementismadebythefactsaboutthetypes,magnitude, duration,anddistributionofevilthataremadeknowntousby historiansandjournalists,nottomentionourown. 10 Butsupposewegivetheexistentialquantifiernarrowscopeandweaken premise(5).Onanarrowscopereadingthepremisestatesthatitisnecessary todivinepurposesthatthereis some amount of evil or other .Consider whetherpremises(1),(2)andtheweakenedversionof(5)areconsistent.

Supposetheactualamountofevil kequals55turps.Since,by hypothesis,anyamountofevilgreaterthan50turpsisunnecessaryfor divinepurposes,itfollowsthat55turpsissuchthatanygreateramountof evilisunnecessaryfordivinepurposes.Accordingtopremise(2),forany amountofevil kn,if knissuchthatanygreateramountofevilisunnecessary fordivinepurposesthenforsome i (i >0) kn–iissuchthatanygreater amountofevilisunnecessaryfordivinepurposes.Sincewehavedefined i =

½( kn–50),itfollowsthat kn–i =( kn/2+25).Itisevidentthatanyamountof evilgreaterthan50turpssatisfiespremise(2).Butsince50isincludedas thegreatestlowerboundonthesequence,thesequencemustconvergeto50 turpsofevil.Andpremise(2)isfalsifiedontheassumptionthat kn =50 13

turps.If kn =50then knissuchthatanygreateramountofevilis unnecessaryfordivinepurposes.Butthereisnoincrementinevil i (i >0) suchthatanyamountofevilgreaterthan kn – i isunnecessaryfordivine purposes.Inshortif kn =50,then i= kn–( kn/2+25)=0.Butthenpremise(2) isfalse.Theantecedentinpremise(2)istruebutitisfalsethatforsome i( i

>0) kn- i issuchthatanygreateramountofevilisunnecessaryfordivine purposes.

Ofcoursewemightalsoconsiderweakeningpremise(2)inthe reformulated No Minimum argument.In(2 a)thecondition i( i >0)is weakenedto i( i >0).

2a.Foranyamountofevil kn,( k>kn>0)if knissuchthatanygreater amountofevilisunnecessaryfordivinepurposesthenforsome i( i >0)

kn- i issuchthatanygreateramountofevilisunnecessaryfordivine purposes.

Butthe No Minimum Thesis in(2 a)istrivial.Foranyamountofevil kn,( k> kn>0)if knissuchthatanygreateramountofevilisunnecessaryfordivine purposesthencertainly kn- i , i( i =0),issuchthatanygreateramountofevil isunnecessaryfordivinepurposes.

Itshouldbeclearthatthemetaphysicalclaiminthe No Minimum

Thesis isnotintendedtobeatrivial.Stillsupposethereformulated No

Minimum argumentincludespremise(1),theweakenedpremise(5)and premise(2a).Thesepremisesentailthatthereissomeleastamountofevil suchthatanygreateramountofevilisunnecessaryfordivinepurposes.At 14 precisely 50turpswearriveatanamountofevilthatislessthanthetotal amountofevilnecessaryfor divine purposes and at any amount of evil greater than 50 turps we arrive at an amount of evil that is greater than the total amount necessary for divine purposes. Here is van Inwagen again.

ButcouldHehavepreventedall[instancesofevilsimilarto Rowe’sfawn]?No—notwithoutcausingtheworldtobe massivelyirregular. And of course there is no sharp cutoff point between a world that is massively irregular and a world that is not ... 11 Butaccordingtothereformulated No Minimum argumentthereisasharp cutoffpointbetweenaworldthatismassivelyirregularandaworldthatis not.Atprecisely50turpsofeviltheworldismassivelyirregularandatany amountofevil—eversosmall—greaterthan50turpstheworldisregular.So thereformulated No Minimum argument entails that there is a precise border between an amount of evil that is less than the total necessary for divine purposes and an amount of evil that is more than the total amount necessary for divine purposes. Van Inwagen notes elsewhere,

OnemightaswellsupposethatifGod’spurposesrequirean impressivelytallprophettoappearatacertainandplace, thereisaminimumheightthatsuchaprophetcouldhave. 12 Butgiventhereconstructed No Minimum argument,thereisaprecise height,say6',suchthatanyprophetthatisexactly6'orlesswillfailtoserve

God’spurposesandanyprophetthatexceeds6'byanyamount—saya prophetthatis6.0000000000001'—wouldserveGod’spurposesperfectlywell. 15

Supposethata6'prophetwouldnotserveGod’spurposesandthata worldcontaining50turpsofevilismassivelyirregular.Itisevidentthatthe worldwouldnotsuddenlybecomeregularat50.0000000000001turpsofevil andaprophetwouldnotsuddenlyserveGod’spurposesperfectlywellat

6.0000000000001feet.

Thereformulated No Minimum argumentcontainspremisesthatare inconsistent.Andthepremises(2a)and(5)intheweakenedreformulation aresimplyimplausible.Theweakenedreformulationalsoentailsthatthere isapreciseborderbetween an amount of evil that is less than the total necessary for divine purposes and an amount of evil that is more than the total amount necessary for divine purposes. So the reformulated No Minimum argument fares no better than the initial No Minimum argument.

2.3 Vague Minimums: A Supervaluation Solution .

Supposeit’sagreedthatthepredicate‘isnecessaryfordivinepurposes’ doesnotsharplydivideitspositiveandnegativeextensions.Thereisan amountofevil kn – j thatisnecessaryfordivinepurposesandanamountof evil knthatisunnecessaryfordivinepurposes.Butthereareamountsofevil inthesequence Sthatareneithernecessaryfordivinepurposesnor unnecessaryfordivinepurposes.Nowsupposethattheactualamountofevil kfallsjustoutsidetherangeofevilthatisclearlyunnecessaryfordivine purposes.Figure(1)displaysthesituationthatweareconsidering.

| ------|------... > 0NecessaryEvil kUnnecessaryEvil 16

Fig. 1 Theamountsofevilintheshadedareainfigure(1)arethoseamountsthat arenotdefinitelynecessaryfordivinepurposesandnotdefinitely unnecessaryfordivinepurposes.So kfallsoutsidetherangeofevilthatis definitelyunnecessaryfordivinepurposes.Supervaluationsemanticsurges thatthetruthoftheproposition“theamountofevilin kisnecessary fordivinepurposes”canbedetermined only if wesharpenorprecisifythe vaguepredicate‘isnecessaryfordivinepurposes’.

Butofcoursethereisnouniqueandnonarbitrarywaytomakethe predicate‘isnecessaryfordivinepurposes’precise.Anysemanticdecisionto sharpenthepredicatewillmakesomearbitrarydistinction.

Supervaluationismthereforemakesit true thattheamountofevilin kis necessaryfordivinepurposesifandonlyifthatpropositionistrueonevery admissibleprecisificationof‘isnecessaryfordivinepurposes’.And supervaluationismmakesit false thattheamountofevilin kisnecessaryfor divinepurposesifandonlyifthatpropositionisfalseoneveryadmissible precisificationofthatpredicate.Otherwisethepropositionisneithertruenor false.

Supervaluationsemanticsplacessomeimportantrestrictionson admissibleprecisifications.Propositionsthatareclearlytrue(false)priorto precisificationmustremaintrue(false)afterprecisification.Theproposition expressedinpremise(5)ofthe No Minimum argument,forinstance,is assumedtobeclearlytrueandsoitmustremaintrueonevery 17

precisification.Butitisnodoubtfalsethattheamountofevilin kn( k>kn>

0)isnecessaryfordivinepurposesonlyifthelesseramountofevilin kn- j ( kn

> kn - j >0)isunnecessaryfordivinepurposes.Thepropositionmusttherefore remainfalseoneveryprecisification.

Reconsiderthe No Minimum Thesis .Supervaluationsemanticsmakes thethesistrueonlyifitistrueoneveryadmissibleprecisificationof‘is necessaryfordivinepurposes’.Butoneveryadmissibleprecisificationofthat predicatetherewillbesome least amountofevil kn,( k>kn>0)suchthatany greateramountofevilisunnecessaryfordivinepurposes.If knistheleast amountofevilsuchthatanygreaterevilisunnecessaryfordivinepurposes, thenofcourse kn – i is not suchthatanygreateramountofevilisunnecessary fordivinepurposes.Weknowthereisoneincrementofevilgreaterthan kn – i

(namely, kn)thatisnecessaryfordivinepurposes.Thereforenohow wemakethepredicateprecisetherewillbeanexceptiontothegeneralization expressedinthe No Minimum Thesis .Itfollowsthatthethesisisfalse.

Butitdoesnotfollowthatthetransitionfromtheamountofevil unnecessaryfordivinepurposestotheamountnecessaryfordivinepurposes isdiscrete.Weshouldreplacethe No Minimum Thesis in(2)withthethesis in(2b).

2b. Thereisnoamountofevil kn,( k>kn>0)in Ssuchthatforevery

increment i (i>0)and every admissibleprecisification, knis

unnecessaryfordivinepurposesand kn- i isnecessaryfordivine purposes. 18

Itfollowsfrom(2b)thatforeveryamountofevilkn( k>kn>0)thereis some admissiblewayofsharpeningthepredicate‘isnecessaryfordivinepurposes’ suchthat knisunnecessaryfordivinepurposesonlyif kn – i isunnecessaryfor divinepurposes.Thethesisin(2b)thereforeassertsthatthereisnodiscrete transitionfromtheevilunnecessaryfordivinepurposestotheevilnecessary fordivinepurposes.Call(2b)the Vague Minimum Thesis .

The Vague Minimum Thesis istrue. 13 Butdoesitpresentaproblem forthestandardpositiononevil?Supposethattheactualamountofevil kis accuratelyrepresentedinfigure(1)above.Accordingto(2b)thereis some admissibleprecisificationonwhichtheamountofevilin kisunnecessaryfor divinepurposesandthelesseramountofevilin k – i isalsounnecessaryfor divinepurposes.Butitdoesnotfollowthatthestandardpositiononevil prohibitsaperfectbeingfrompermittingtheevilin k.Thisisbecauseitis not true thattheamountofevilin kisunnecessaryfordivinepurposes.

Certainlyonsomeadmissiblewaysofsharpeningthepredicate‘isnecessary fordivinepurposes’ kisunnecessaryfordivinepurposes.Butonother admissiblewaysofsharpeningthepredicate kisnecessaryfordivine purposes.Supervaluationismmakesittruethat kisunnecessaryfordivine purposesonlyif kisunnecessaryfordivinepurposeson every admissible precisification.Thestandardpositionthereforedoesnotprohibitaperfect beingfromallowing k. 19

Butsupposethattheamountofevilin kisclearlyunnecessaryfordivine purposes.HereisvanInwagen.

Ifthereisapurposethatisservedbyallowingthe“ageofevil” tohaveacertainduration,doubtlessthesamepurposewouldbe servediftheageofevilwerecutshortbyaday,ayearorevena century. 14 Let’ssupposethattheactualamountofunnecessaryevilistheamount containedinanadditionaldayinthe“ageofevil”.Supposetheamountof evilin kisclosetotheamountdepictedinfigure(2).

| ------|------... > 0NecessaryEvil kUnnecessaryEvil Fig. 2 Theamountofevil kisclearlyunnecessaryfordivinepurposes.Butitisalso obviousinfigure(2)that kdoesnotincludeagreatdealofunnecessaryevil.

Stillitistrueoneveryadmissibleprecisificationthattheamountofevilin k isunnecessaryfordivinepurposes.Soevenundertheassumptionthat kdoes notcontainagreatdealofunnecessaryevilthestandardpositiondoesnot allowaperfectbeingtopermit k.

Considertheinterpretationofthestandardpositionthatismost favorableto.Itisalsothemostplausibleversionofthat.The standardpositiononevilholdsthataperfectbeingwouldnotpermitany amountofevil knthatis definitely unnecessary.Theamountofevil knis definitelyunnecessaryfordivinepurposesjustincaseitisunnecessaryon everyadmissibleprecisification. 20

SP1.Aperfectbeingwouldnotpermitanamountofevil kn( k>kn>0)in S

ifandonlyif,oneveryadmissibleprecisification, knisunnecessaryfor divinepurposes. Wecanthereforeconcludethataperfectbeingwouldnotpermittheamount ofevilwefindintheactualworldif—asvanInwagenproposesinthe No

Minimum argument—theactualamountofevilisdefinitelyunnecessaryfor divinepurposes.

The Vague Minimum Thesis in(2b)guaranteesthatthereisno discretetransitionbetweentheamountofevilthatisunnecessaryfordivine purposesandtheamountofevilthatisnecessaryfordivinepurposes.But thispresentsnoproblemforthestandardpositiononevil.Thestandard positioninSP1entailsthataperfectbeingwouldnotpermitanyamountof evilthatisdefinitelyunnecessaryfordivinepurposes.AndSP1and(2b)are perfectlyconsistent.

3. Alternative No Minimum Theses.

Supervaluationsemanticsprovidesasimplecountermodeltothe No

Minimum Thesis .Nomatterhowwemakethepredicate‘isnecessaryfor divinepurposes’precisetherewillbeanexceptiontothegeneralization expressedinthatthesis .Soitisreasonabletoreplacethatthesiswiththe

Vague Minimum Thesis .

JeffJordanhasproposedanotherversionofthenominimumthesis thatisconsistentwiththe Vague Minimum Thesis .15

Foranyamountofevil kn( k>kn>0)thatissufficientfordivine 21

purposes,thereissomeincrement isuchthat kn – i isalsosufficientfor divinepurposes. ConsiderwhethertheproposedthesisadvancesvanInwagen’s No Minimum argument.Supposethatthenewthesisasserts(2c).

2c.Foranyamountofevil kn( k>kn>0)if knissuchthatanygreater amountofevilis definitely unnecessaryfordivinepurposesthenfor

someincrement i(i>0), kn – i issuchthatanygreateramountofevil isdefinitelyunnecessaryfordivinepurposes. (2c)entailsthateveryamountofevilisdefinitelyunnecessaryfordivine purposes.Butofcoursethe No Minimum argumentassumesthatatleast someevilisnecessaryfordivinepurposes.So(2c)doesnotadvancethe No

Minimum argument.Butsupposeinsteadthethesisasserts(2d).

2d.Forany kn( k>kn>0)if knissuchthatanygreateramountofevilis (atleast)indefinitelyunnecessaryfordivinepurposesthenforsome

increment i( i>0), kn – i issuchthatanygreateramountofevilis (atleast)indefinitelyunnecessaryfordivinepurposes. (2d)entailsthateveryamountofevilisindefinitelyunnecessaryfordivine purposes.(2d)isthereforeconsistentwiththestandardpositioninSP1.

AccordingtoSP1aperfectbeingisprohibitedfromallowinganamountof evilthatisdefinitelyunnecessaryfordivinepurposes.Soitisnoviolationof

SP1thataperfectbeingallowsanamountofevilthatisontheborderofevil thatisunnecessaryfordivinepurposes.

3.1 A Degree-Theoretic Solution 22

Accordingtodegreetheoreticanalysesofvaguenesstherearenotruthvalue gaps.Borderlinepredicationssuchas“theamountofevil knisnotclearly unnecessaryfordivinepurposes”areassignedsomerealvalueinthe(closed) interval[0,1].Theserealvalueassignmentsaretypicallyinterpretedas degreesoftruthwith1correspondingtotruesimpliciterand0corresponding tofalsesimpliciter. 16

Thedegreetheoreticinterpretationofthe No Minimum argumentis certainlyconsistent.Thedegreetheoreticinterpretationofpremise(1)states thatitistruesimpliciter—ortruetodegree1—thattheactualamountofevil kisunnecessaryfordivinepurposes.Wecouldofcourseweakenpremise(1) toitis assertable thattheactualamountofevil kisunnecessaryfordivine purposes.Insupervaluationsemanticsthereisarangeintheamountsofevil thataredefinitelyunnecessaryfordivinepurposes.Indegreetheorythereis arangeintheamountsofevilthatareassertablyunnecessaryfordivine purposes.

Supposeweassumeaconservativerangeofassertablity.Let’ssaythat theamountofevilin knisassertably(un)necessaryfordivinepurposesifand onlyif knisassertably(un)necessaryfordivinepurposestodegree n(.7< n<

1).Thedegreetheoreticsemanticsofvaguenessincludesatruthfunctional orquasitruthfunctionalsystemfortheclassicalconnectivesincluding|~p|

=|1–p|.Sotheamountofevilin knisnotassertably(un)necessaryfor divinepurposesifandonlyif knisassertably(un)necessaryfordivine 23 purposestodegree n(0<n<.7).Thereisthereforearangeintheamountsof evilthatisneitherassertablynecessaryfordivinepurposesnorassertably unnecessaryfordivinepurposes.Callthedegreetheoreticinterpretationof premise(2), No Assertable Minimum .

2e. There is no amount of evil kn, ( k > kn > 0) in S such that for every

increment i ( i > 0) kn is assertably unnecessary for divine purposes

and kn- i is assertably necessary for divine purposes. Premise(1)and No Assertable Minimum areconsistentwiththedegree theoreticinterpretationofpremise(5).Premise(5)nowstatesthatthereis someamountofevilthatisassertablynecessaryfordivinepurposes.

Butdoesthedegreetheoreticinterpretationofthe No Minimum argumentpresentaproblemforthestandardpositiononevil?VanInwagen suggeststhattheactualamountofevilin kisassertablyunnecessary.But themostthatwecanderivefromthe No Minimum argumentisthataperfect beingmayallowsomeinstancesofevilthatare not assertably unnecessary.

Perhapsaperfectlygoodbeingwouldnotpermitanyevilthatisevenso muchasnonassertablyunnecessaryfordivinepurposes.Butthe interpretationofthestandardpositionmostfavorabletotheismstatesthata perfectbeingwouldnotpermitanyinstanceofevilthatisassertably unnecessaryfordivinepurposes.

SP2.Aperfectbeingwouldnotpermitanamountofevil kn( k>kn>0)in Sif

andonlyif knisassertablyunnecessaryfordivinepurposes. 24

Wecanthereforeconcludethataperfectbeingwouldnotpermittheamount ofevilwefindintheactualworldif—asvanInwagenproposesinthe No

Minimum argument—theactualamountofevilisassertablyunnecessaryfor divinepurposes.

No Assertable Minimum in(2e)guaranteesthatthereisnodiscrete transitionbetweentheamountofevilthatisunnecessaryfordivinepurposes andtheamountofevilthatisnecessaryfordivinepurposes.Butthis presentsnoproblemforthestandardpositiononevil.Thestandardposition inSPentailsthataperfectbeingwouldnotpermitanyamountofevilthatis assertablyunnecessaryfordivinepurposes.AndSP2and(2e)areperfectly consistent.

3.2 Anti-Luminous Evil .

Consideranamountofevil kn,( k>kn>0)in Sconcerningwhichweare confident that anygreateramountofevilisunnecessaryfordivinepurposes.

Indeedwecanbefairlyconfidentthattheactualamountofevil kissuchthat anygreateramountofevilisunnecessaryfordivinepurposes.Suppose furtherthatwe know that kissuchthatanygreateramountofevilis unnecessaryfordivinepurposes.Canwereachtheconclusionthat,forany amountofevil kn,( k>kn>0),ifweknowthat kn issuchthatanygreater amountofevilisunnecessaryfordivinepurposesthen,forsome i( i >0),we knowthat kn- i issuchthatanygreateramountofevilisunnecessaryfor divinepurposes?Consideranepistemicistversionthe No Minimum Thesis .17 25

2f.Foranyamountofevil kn,( k>kn>0)in Sifwe know that knissuch thatanygreateramountofevilisunnecessaryfordivinepurposesthen

forsome i( i >0),weknowthat kn- i issuchthatanygreateramountof evilisunnecessaryfordivinepurposes. Sinceweknowthattheactualamountofevilissuchthatantgreateramount ofevilisunnecessaryfordivinepurposeswealsoreliably believe thatthe actualamountofevilissuchthatanygreateramountofevilisunnecessary fordivinepurposes.Butifwereliablybelievethat knissuchthatanygreater evilisunnecessaryfordivinepurposesthenforsomeincrement i( i >0), kn- i suchthatanygreaterevilisunnecessaryfordivinepurposes.Thereis,for some i( i >0),nonoticeabledifferencebetweentheamountofevil knandthe amountofevil kn-i.Sowesimplycannotbeconfidentthat knissuchthatany greateramountofevilisunnecessaryfordivinepurposesandnotbe confidentthat kn- i issuchthatanygreaterevilisunnecessaryfordivine purposes.Reliabilityinthiscontextdependsoncondition(R).

R.Foranyamountofevil kn,(k>kn>0)in Sifweknowthat knissuchthat anygreateramountofevilisunnecessaryfordivinepurposesthenfor

some i( i >0), kn- i issuchthatanygreateramountofevilisunnecessary fordivinepurposes. Butsupposeitistruethatforanyamountofevilkn,( k>kn>0)in Sif knis suchthatanythatanygreateramountofevilisunnecessaryfordivine purposesthenwe know that knissuchthatanythatanygreateramountof 26 evilisunnecessaryfordivinepurposes.Thesuppositionisthatunnecessary evilis luminous .

L.Foranyamountofevil kn,( k>kn>0)in Sif knissuchthatanygreater

amountofevilisunnecessaryfordivinepurposesthenweknow knissuch thatanygreateramountofevilisunnecessaryfordivinepurposes. Thereliabilitycondition(R)andtheluminositycondition(L)entailthe No

Minimum Thesis in(2f).Butitisevidentthat(L)isfalse.

Accordingtoepistemicisttheoriesofvaguenesseveryamountofevilis eithernecessaryfordivinepurposesorunnecessaryfordivinepurposes.The phenomenonofvaguenessissimplyareflectionofourignoranceofthe preciseborderbetweenamountsofevilthatarenecessaryfordivinepurposes andamountsthatareunnecessaryfordivinepurposes.Therearecertainly amountsofevil knsuchthatinformedandcompetentlanguageusersarenot preparedtoassertthat knisunnecessaryfordivinepurposesandalsonot preparedtoassertthat knisnotunnecessaryfordivinepurposes.Theseare amountsofevilthataretooclosetothemarginsofunnecessaryevilforus— orforanyonewhosejudgmentinthesemattersissimilarlylimited—to reliablybelievetheyareunnecessaryfordivinepurposes.Sothereare certainlyamountsofevil kninSthataresuchthatanygreateramount ofevilisunnecessaryfordivinepurposesandalsosuchthatwe do not know thatanygreateramountofevilisunnecessaryfordivinepurposes.

Unnecessaryevilisnotluminous. 27

Restrictedtolimitedagents—agentsthatarenotomniperceptiveand notomniscient—wehavetheepistemicistinterpretationofthe No Minimum

Thesis in(2g).

2g. Thereisnoamountofevil kn,( k>kn>0)in Ssuchthatforevery

increment i( i >0) knis reliably believed unnecessaryfordivine

purposesand kn- i isreliablybelievednecessaryfordivinepurposes.

(2g)istrueanditisconsistentwiththerebeinganamountofevil knsuch that knistheleastamountofevilsuchthatanygreaterevilisunnecessary fordivinepurposes.Theinterpretationofthestandardpositionmost favorabletotheismstatesthataperfectbeingwouldnotpermitanyinstance ofevilthatarereliablybelievedunnecessaryfordivinepurposes.

SP3.Aperfectbeingwouldnotpermitanamountofevil kn( k>kn>0)in Sif

knisreliablybelievedtobeunnecessaryfordivinepurposes. Butthemostthatwecanconcludefromtheepistemicistinterpretationofthe

No Minimum argumentisthataperfectbeingmaynotallowanyinstancesof evilthatarereliablybelievedtobeunnecessaryfordivinepurposes.Of coursewecannotconcludethataperfectbeingmayallowinstancesofevil thatwedonotreliablybelieveareunnecessaryfordivinepurposes,since suchinstancesofevilmayinfactbeunnecessaryfordivinepurposes.

Wecanthereforeconcludethataperfectbeingwouldnotpermitthe amountofevilwefindintheactualworldif—asvanInwagenproposesinthe 28

No Minimum argument—theactualamountofevilisreliablybelievedtobe unnecessaryfordivinepurposes.

4. Supervaluation and Higher-Order Vagueness .

Thereisnoamountofevil knsuchthat,oneveryadmissible precisification, knisunnecessaryfordivinepurposesand kn – i isnecessaryfor divinepurposes.Thereisthereforenoprecisetransitionbetweenevilthatis unnecessaryfordivinepurposesandevilthatisnecessaryfordivine purposes.Butthispresentsnoproblemforthestandardpositiononevil.The standardpositionholdsthataperfectbeingwouldnotpermit kn justincase kn isunnecessaryfordivinepurposesoneveryadmissibleprecisification.

Butsupposethereisalsoavagueborderbetweentheamountofevil thatisunnecessaryoneveryprecisificationandtheamountofevilthatisnot unnecessaryoneveryprecisification.Sowhatcountsasasetofadmissible precisificationsisitselfamatterofsemanticindecision.Atsomepointinthe sequencebetween kand0thereissomeamountofevil knsuchthat,onevery precisificationinsomesetofadmissibleprecisifications, knisunnecessaryfor divinepurposes.Itistherefore definite that knisunnecessaryfordivine purposes.Butonsomeprecisificationsinanothersetofadmissible precisifications knis not unnecessaryfordivinepurposes.Itistherefore not definite that itis definite that k n is unnecessaryfordivinepurposes. The predicate‘isnecessaryfordivinepurposes’issecondordervague. 29

Ifitisindefinitewhetheritisdefinitethat knisunnecessary,thenitis indefinitewhether knisunnecessaryfordivinepurposesoneveryadmissible precisification. knisontheborderbetweentheamountofevilthatis definitelyunnecessaryfordivinepurposesandtheamountofevilthatis indefinitelyunnecessaryfordivinepurposes.Figure(3)depictsthis situation.

----|------|------... >

kn-jIndefinitelykn UnnecessaryEvilUnnecessaryEvil Fig. 3

Itissecondordervaguewhethertheamountofevilin knis unnecessaryfordivinepurposes.Itisfirstordervaguewhethertheamount ofevilin kn-jisunnecessaryfordivinepurposes.Buttheamountofevilin kn and kn-jare both on the border ofunnecessaryevil.Sothefactthatitis definitethat knisunnecessaryfordivinepurposes relative to some set of admissible precisifications shouldnotentailthataperfectbeingisprohibited frompermittingtheamountofevilin kn.Theamountofevilin knmight nonethelessbeontheborderofunnecessaryevil. 18 Itmightbesecondorder orthirdordervaguewhethertheamountofevilinknisunnecessaryfor divinepurposes.

Let’ssupposeitis superdefinite that knisunnecessaryfordivine purposesjustincasethepropositionisdefiniteateveryhigherorderof 30

vagueness.Ifitisnotsuperdefinitethat knisunnecessaryfordivinepurposes thenforsomeprecisificationatsomehigherorderofvagueness, knisnot unnecessaryfordivinepurposes.Thereforetheamountofevilin knisonthe borderlineofevilthatisunnecessaryfordivinepurposes.Wetherefore arriveattheconclusionthatthestandardpositiononevilprohibitsaperfect beingfrompermittingtheevilin knifandonlyifitissuperdefinitethat knis unnecessaryfordivinepurposes.Butthatisjusttosaythatthestandard positionprohibitsaperfectbeingfrompermittingknifandonlyif knis unnecessaryfordivinepurposesand knisnotontheborderlineofevilthatis unnecessaryfordivinepurposes.Ifthepredicate‘isunnecessaryfordivine purposes’ishigherordervague,theprincipleSP4specifiestheconditions underwhichaperfectbeingwouldnotpermitevil.Andtheseareagain conditionsmostfavorabletovanInwagen’sposition.

SP4.Aperfectbeingwouldnotpermitanamountofevil kn( k>kn>0)in S

ifandonlyifitissuperdefinitethatknisunnecessaryfordivine purposes. ThestandardpositioninSP4doesnotentailthatthereisadiscrete transitionbetweentheamountofevilthataperfectbeingwouldpermitand theamountofevilaperfectbeingwouldnotpermit.Thethesisin(2h) guaranteesthatthetransitionbetweenimpermissibleevilandpermissible evilisnotdiscrete.

2h.Thereisnoamountofevil kn,( k>kn>0)in Ssuchthatforevery

increment i (i>0)itissuperdefinitethat knunnecessaryfordivine 31

purposesandsuperdefinitethat kn- i isnotnecessaryfordivinepurposes. (2h)reformulatesthe Vague Minimum Thesis .Accordingto(2h)the transitionisvaguebetweentheamountofevilthatisunnecessaryfordivine purposesandtheamountofevilthatisnecessaryfordivinepurposes.But thispresentsnoproblemforthestandardpositiononevil.Thestandard positioninSP4entailsthataperfectbeingwouldnotpermitanyamountof evilthatissuperdefinitelyunnecessaryfordivinepurposes.AndSP4and

(2h)areperfectlyconsistent.

5. Omniscience and Vagueness .

Supervaluationistsurgethattherearenohiddenboundariesforvague predicatesthatareaccessibleonlytoomniscientbeings.Itisnottrue,for instance,thatGodknowsthepreciseboundarybetweeneverypossible instanceofbeingbaldandnotbeingbald.AnditisnottruethatGodknows thepreciseboundarybetweentheamountofevilnecessaryfordivine purposesandtheamountofevilunnecessaryfordivinepurposes.According tothesupervaluationiststhisisbecausethereisnopreciseboundarytobe known.Insteadsupervaluationistsmaintainthatvaguenessisamatterof semanticindecision.Andsincetherearenosharpboundariesforvague predicatesatanylevelofvaguenesssemanticindecisionmustcontinue upwardthroughourmetalanguageandsoon.

Supervaluationismconcludesthatwereplacethe No Minimum Thesis withthe Vague Minimum Thesis .Accordingtothe Vague Minimum Thesis 32 thereareborderlinecasesbetweenamountsofevilthatareunnecessaryfor divinepurposesandamountsofevilthatarenecessaryfordivinepurposes.

Soinsection(4)wesuggestedthatthestandardpositiononevilisbetter formulatedasrequiringthataperfectbeingnotpermitanyamountofevil kn thatis definitely unnecessary.Theamountofevil knisdefinitelyunnecessary fordivinepurposesjustincaseitisunnecessaryoneveryadmissible precisificationof‘unnecessaryfordivinepurposes’.

SP1.Aperfectbeingwouldnotpermitanamountofevil kn( k>kn>0)in S

ifandonlyif,oneveryadmissibleprecisification, knisunnecessaryfor divinepurposes. AccordingtoSP1aperfectbeingcannotpermitanamountofevilthat isdefinitelyunnecessaryfordivinepurposes.Let’ssupposethat knis definitelyunnecessaryfordivinepurposes.Ourcentralquestioniswhether anomniscientbeingwould know that knisdefinitelyunnecessaryfordivine purposes.Itdoesseemreasonabletoproposethatanomniscientbeingwould know(atleast)everypropositionthatisdefinitelytrue.Considerthenthe characterizationofomniscienceinO 1.19

O1. (∀x)(xisomniscient≡( ∀p)(xknowsp≡xp)&(xbelievesp≡Defp)))

AccordingtoO 1abeingisomniscientifandonlyifitknowsallandonlythose propositionsthataredefinitelytrue.Soiftherearepropositionspthatare indefinitelytruethennoomniscientbeingknowsthatp.Onanyviewof 33 omniscienceaccordingtowhichGodknowseverytrue(definiteorindefinite) propositionthisconsequenceisunwelcome. 20

Supposethatsupervaluationistsmaintainthatthereissimplyno differencebetweentruthanddefinitetruth.Thenofcourseomniscientbeings couldnotknowanyindefinitesincetherearenoindefinitetruths.Ifp isan“indefinitetruth”thenpisbydefinitionontheborderbetweenbeing definitelytrueandbeingdefinitelyfalse.Andnotevenanomniscientbeing coulddiscoverthatthereisnoadmissibleprecisificationunderwhichpisnot true.Soomniscientbeingswouldknowtheindefinitetruthponlyif omniscientbeingswouldfavoradmissibleprecisificationsunderwhichpis true.Buttheresimplyisnoreasontofavoradmissibleprecisificationsunder whichpistrueoverprecisificationsunderwhichpisfalse.

ThereisamoreseriousproblemforthecharacterizationinO 1.Itis quitereasonabletobelievethattheknowledgeofomniscientbeingsisclosed undertheS4axiom.AndsoweshouldexpectO 2tobetrueforallomniscient beings.

O2. (∀x)(xisomniscient≡( ∀p)(xknowsp≡xknowsthatxknowsp))

AccordingtoO 2anomniscientbeingknowseverythingthatheknowsandso onupward.ButifO 2governstheknowledgeofeveryomniscientbeing,then thecharacterizationinO 1cannotbetrue.Omniscientbeingscannotknow everypropositionthatisdefinitelytrue. 34

Supposepisdefinitelytrueortrueundereveryadmissible precisification.ItfollowsfromO 1thateveryomniscientbeingknowsthatpis true.ButthenfromO 2itfollowsthateveryomniscientbeingknowsthathe knowsthatpistrue.AndfromO 1againwederivetheconclusionthatitis definitethatitisdefinitethatpistrue.Butthisconclusionmightwellbe false.Noteverypropositionthatisdefinitelytrueis definitely definitelytrue.

Butthenomniscientbeingscannotknoweverypropositionthatisdefinitely true.

Ifomniscientbeingsdonotknoweverypropositionthatisdefinitely truethenweareforcedtorejectSP1.Itisnotingeneraltruethataperfect beingwouldnotpermitanamountofevil knthatisdefinitelyunnecessaryfor divinepurposes.Theinferenceisstraightforward.IfO 1isfalsethenaperfect being might not know that knisdefinitelyunnecessaryfordivinepurposes.

Butifhedoesnotknowthat knisdefinitelyunnecessaryfordivinepurposes thenofcoursehemightpermit kn.TheprincipleinSP1isthereforefalse.

CertainlysupervaluationistshavetheoptiontorejecttheS4axiomin

O2.ButitfollowsfromtherejectionofO 2thatomniscientbeingsdonotknow everythingthattheyknow.Andthatisanespeciallyunwelcomeconclusion.

Certainlyconsiderationsofantiluminosityandsafetyurgethatlessthan omniperceptiveandlessthanomniscientbeingscannotingeneralknowwhat theyknow.Butsuchconsiderationsurgenothingagainstomniscientand 35 omniperceptivebeings.Theonlyremainingalternativeistorejectthe characterizationinO 1.

AswehavenotedO 1isfalseonlyifperfectbeingsdonotknowevery propositionthatisdefinitelytrue.Andsoweareforcedtorejecttheprinciple inSP1.Theseconclusionsarelessseriousthantheyappear.Theprinciple

SP1wasnotintendedtogovernhigherordervaguenessforthepredicate‘is unnecessaryfordivinepurposes’.Theprinciplewasintendedtogovernfirst ordervaguenessforthatpredicate.Butcertainlysupervaluationistsare primarilyconcernedaboutthegeneralproblemofvaguenessforthat predicate.

Shouldweexpecteveryomniscientbeingtoknoweverydefinitelytrue proposition?Itseemsperfectlyreasonabletoholdthatomniscientbeings mightfailtoknowindefinitelytruepropositions.Thesepropositionsarenot truerelativetoeveryadmissibleprecisificationoftheirvagueterms.

Supervaluationistsmighturgethatitisatleastasreasonabletoholdthat omniscientbeingsdonotknowindefinitelydefinitepropositions.These propositionsarenotdefinitelytruerelativetoeveryadmissiblewaytoresolve thevaguenessof‘admissibleprecisification’.Suppose,forinstance,thatitis indefinitewhether knisunnecessaryfordivinepurposes. knisthereforeon theborderofevilthatisunnecessaryfordivinepurposes.Butsupposeitis indefinitewhetheritisdefinitethat knisunnecessaryfordivinepurposes. kn is again onaborderofevilthatisunnecessaryfordivinepurposes.Onthis 36

viewomniscientbeingsdonotknowthat knisunnecessaryfordivine purposesunless knisnotonsomeborderoftheamountofevilthatis unnecessaryfordivinepurposes.Sothispositiongivesconsistenttreatment toeverypropositionthatison some border ofthoseunnecessaryfordivine purposes.

Supervaluationistscanresolvetheseproblemsandretainasimple characterizationofomniscience.AccordingtoSP5apropositionpis superdefinitelytruejustincasepisdefinitelytrueatevery(higher)orderof vagueness.SupposepisthepropositionIndef 2A,orthepropositionthatitis indefinitethatitisindefinitethatA.Inthatcasepissuperdefinitelytrueif andonlyifDef nIndef 2Aforeveryorder n( n >3).Sopissuperdefinitelytrueif andonlyif‘Indef 2A’isdefinitelytrueateveryorderofvaguenessgreater thanorequaltothethirdorder.InshortAisnotonanyborderofbeing secondorderindefiniteA.ButsupposepisthepropositionA.Inthatcasepis superdefinitelytrueifandonlyifDef nAforeveryorder n( n>0). 21

Supervaluationistsshouldurgethatomniscientbeingsknowevery propositionthatissuperdefinitelytrue.Nowsupposewerestrictthe propositionalquantifiertoallandonlythosepropositionsthatare superdefinitelytrue.Anomniscientbeingknowseveryproposition characterizedinthesimplerO 3.

O3. (∀x)(xisomniscient≡( ∀p)(xknowsp≡p)) 37

O3statesthatforeverypropositionp,everyomniscientbeingknowsp.O 3is perfectlyconsistentwithO 2,sincethepropositionthatxknowsthatxknows thatpdoesnotentailthatitisdefinitethatitisdefinitethatp.Itfollows fromO 2 thatanomniscientbeingknowsponlyifheknowsthatheknowsp.

Andthatisaconsequenceofomniscienceweshouldexpect.

Nowsupposepisthepropositionthatitisindefinitethat knis unnecessaryfordivinepurposes.Ifitissuperdefinitethatpthenan omniscientbeingknowsthatitisindefinitethat knisunnecessaryfordivine purposes.SoaccordingtoO 3itisperfectlypossiblethatanomniscientbeing knowsthatitisindefinitethatAandthatheknowsthatit’sindefinitely indefinitethatBandsoon.

Wenotedinsection(6)thatabetterformulationofthestandard positiononevilprohibitsaperfectbeingfromallowingtheevilin knifand onlyifitissuperdefinitethat knisunnecessaryfordivinepurposes.The principleSP4specifiestheconditionsmostfavorabletotheismunderwhicha perfectbeingwouldnotpermitevil.

SP4.Aperfectbeingwouldnotpermitanamountofevil kn( k>kn>0)in S

ifandonlyifitissuperdefinitethat knisunnecessaryfordivine purposes. Supposethatit’ssuperdefinitelytruethat knisunnecessaryfordivine purposes.AccordingtoO 3thepropositionthat knisunnecessaryfordivine purposesisamongthepropositionsthateveryomniscientbeingknows.So happilyO 2andO 3areconsistentwiththestandardpositioninSP4.Aperfect 38

beingwouldknowthat knisunnecessaryfordivinepurposesaccordingtoO 3 andaperfectbeingwouldnotpermit kn accordingtoSP4.

6.Conclusions

AccordingtovanInwagenthestandardpositiononevilistrueonlyif thereisaleastamountofevilsuchthatanygreaterevilisunnecessaryfor divinepurposes.Butthatpositionismistaken.The Vague Minimum Thesis in(2b)guaranteesthatthereisnodiscretetransitionbetweentheamountof evilthatisunnecessaryfordivinepurposesandtheamountofevilthatis necessaryfordivinepurposes.Andthethesisin(2a)isperfectlycompatible withthestandardpositiononevilin(SP1).

Certainlythereareotherinterpretationsofthe Vague Minimum

Thesis forthcoming,forinstance,fromdegreetheoreticaccountsofvagueness andepistemicistaccountsofvagueness.But(2e)and(2g)donotadvancevan

Inwagen’s No Minimum argumentandasithappensneitherdo(2c)and(2d).

Itisperhapstruethattheamountofevilnecessaryfordivinepurposes issecondorderorthirdordervague.Ifthepredicate‘isnecessaryfordivine purposes’isahigherordervaguepredicate,thenthestandardposition prohibitsaperfectbeingfrompermittingtheamountofevil knifandonlyifit issuperdefinitethat knisunnecessaryfordivinepurposes.Ifitis not superdefinitethat knisunnecessaryfordivinepurposes,thenforsome admissibleprecisificationatsomeorderofvagueness knisnotunnecessary fordivinepurposes.Itfollowsthat knisontheborderlineofevilthatis 39 unnecessaryfordivinepurposes.Andaperfectbeingneednotpreventevil thatisontheborderlineofunnecessaryevils.Buthigherordervagueness presentsnoproblemforthestandardpositiononevil.Thereformulated

Vague Minimum Thesis in(2h)guaranteesthatthereisnodiscretetransition betweentheamountofevilthatisunnecessaryfordivinepurposesandthe amountofevilthatisnecessaryfordivinepurposes.Andthethesisin(2h)is perfectlyconsistentwiththestandardpositioninSP4.

Omniscientbeingsknoweverypropositionthatissuperdefinitelytrue.

Ifitissuperdefinitelytruethat knisunnecessaryfordivinepurposesthen

Godknowsthathecannotallowtheamountofevilin kn.Omniscientbeings donotknowanypropositionthatisnotsuperdefinitelytrue.Ifitisnot superdefinitelytruethat knisunnecessaryfordivinepurposesthenthe amountofevilin knisonthe border oftheamountofevilthatisclearly unnecessaryfordivinepurposes.If knisclearlyontheborderthenGod knowsthat knisontheborderofevilthatisunnecessaryfordivinepurposes.

Buthedoesnotalsoknowthat knisunnecessaryfordivinepurposes.

Theremainingquestionistheexactamountofactualevilin k.Suppose, asvanInwagenseemstosuggest,thatitissuperdefinitethattheamountof evilin kisunnecessaryfordivinepurposes.Thestandardpositionthen prohibitsaperfectbeingfrompermitting kandtheamountofevilwefindin theactualworldpresentsanimportantobstacletotheistic.But supposethatforsomeprecisificationatsomeorderofvaguenesstheactual 40 amountofevilin kisnotunnecessaryfordivinepurposes.Itthenfollows that kfallssomewhereontheborderofunnecessaryevilandthispresentsno problematallfortheism. 41

Notes 1SeeWilliamRowe,‘TheProblemofEvilandSomeVarietiesof’ collectedinDanielHowardSnyder(ed.) The Evidential Argument from Evil

(Indianapolis:IndianaUniversityPress,1996),pp.111.Myemphasis.

2Theunfortunatelocution‘theminimumamountofevilnecessaryfordivine purposesD’referstotheleastamountofevilsuchthatanygreaterevilis unnecessaryforD.If kistheleastamountofevilsuchthatanygreaterevilis unnecessaryfordivinepurposes,thenofcourse k– i isnecessaryforDand k

– 2i isnecessaryforDandsoonforanyamountofevillessthan kand greaterthan0.But kisthetotalquantityofevilnecessaryforD.Van

InwagencallsthattheminimumevilnecessaryforD.

3SeePetervanInwagen,‘TheMagnitude,Duration,andDistributionofEvil:

ATheodicy’inhis God, Knowledge and Mystery: Essays in Philosophical

Theology (Ithaca:CornellUniversityPress,1995)p.103.

4PetervanInwagen,‘TheProblemofEvil,theProblemofAir,andthe

ProblemofSilence’inhis God, Knowledge and Mystery: Essays in

Philosophical (Ithaca:CornellUniversityPress,1995)note11,page

77.

5‘TheMagnitude,Duration,andDistributionofEvil:ATheodicy’ op. cit .p.

103.

6‘TheMagnitude,Duration,andDistributionofEvil:ATheodicy’ op. cit .pp.

103104. 42

7Twopointsareworthmentioninghere.Firsttheinfinitedivisibilityofthe evilbetween0and kmightentailthatsomeevilsthatareimperceptibly small.Thatconclusioniscontroversial.JeffJordanarguesagainst imperceptibleharmsorevils.Seehis‘EvilandvanInwagen’, and

Philosophy Vol.20(2003)23639.FrankArntzeniusandDavidMcCarthy offerastrongargumentinfavorofimperceptibleharmsorevils.Seetheir

‘SelfTortureandGroupBeneficence’ Erkenntnis 47(1997)129144.HereI remainneutralonthepoint.Seconditdoesnotaffectthe No Minimum argumentwhethertheinfiniteseriesiscountableoruncountable.

8Thereisastrongerversionofthe No Minimum Thesis thatvanInwagen mighthavehadin.Itisnotclearfromthetextoftheargument.The strongerversiongiveswidescopetothequantificationoverincrements.

Thereissome i( i >0)suchthatforanyamountofevil kn,( k>kn>0)

andif knissuchthatanygreateramountofevilisunnecessaryfor

divinepurposesthen kn- i issuchthatanygreateramountofevilis unnecessaryfordivinepurposes. 9Supposeinsteaditisurgedthat50turpsisnecessaryfordivinepurposes andanyamountofevilgreaterthan50turpsisunnecessaryfordivine purposes.Inthatcasethereisaminimumamountofevilsuchthat any greateramountofevilisunnecessaryfordivinepurposes(viz.,50turps).A perfectbeingmaythereforeallowthetotalamountofevilnecessaryfor 43

divinepurposes(viz.,50turps).Andthatisperfectlyconsistentwiththe standardpositiononevil.

10 ‘TheMagnitude,Duration,andDistributionofEvil:ATheodicy’ op. cit .p.

104.

11 ‘TheProblemofEvil,theProblemofAir,andtheProblemofSilence’ op. cit . note11,page77.

12 ‘TheMagnitude,Duration,andDistributionofEvil:ATheodicy’ op. cit .p.

103

13 The No Minimum Thesis entailsthe Vague Minimum Thesis butthe conversedoesnothold.

NMT.Thereisnoamountofevil kn,( k>kn>0)in Ssuchthatfor some

admissibleprecisification,anyamountofevilgreaterthan knis unnecessaryfordivinepurposesandsomeamountofevilgreater

than kn- i isnecessaryfordivinepurposes. VMT.Thereisnoamountofevil kn,( k>kn>0)in Ssuchthatfor every

admissibleprecisification, knisunnecessaryfordivinepurposesand

kn- i isnecessaryfordivinepurposes. Prove NMTentailsVMT:SupposeVMTisfalse.Thenthereissome knand kn–isuchthatoneveryadmissibleprecisification knisunnecessaryfordivine purposesand kn – i isnecessaryfordivinepurposes.Itistrueof kn – i thatany greateramountofevilisunnecessaryfordivinepurposes.Butthenitfollows fromNMTthatnoamountofevilgreaterthan kn – 2i isnecessaryfordivine 44

purposes.Butthatisfalsesincetheamountofevilin kn – i isnecessaryfor divinepurposes.ThereforeNMTentailsVMT.

Prove VMTdoesnotentailNMT:AssumeVMTandthat,forsomeadmissible precisification, knisunnecessaryfordivinepurposesand kn – i isnecessaryfor divinepurposes.ContrarytoNMTitfollowsthat,forsomeadmissible precisification,anyamountofevilgreaterthan kn – i isunnecessaryfordivine purposesandsomeamountofevilgreaterthan kn – 2i isnecessaryfordivine purposes.SoVMTdoesnotentailNMT.TheweakerthesisinVMTistrue but(aswe’veseen)thestrongerthesisinNMTisfalse.

14 ‘TheMagnitude,DurationandDistributionofEvil:ATheodicy’ op. cit .p.

103

15 JeffJordan,‘EvilandvanInwagen’, op. cit. .Jordanproposesthisversion ofthe No Minimum thesisandfinallyrejectsit.

16 ThereissomereasontobelievethatvanInwagenendorsessomedegree theoreticaccountofvagueness.Seeforinstancehis Material Beings (Ithaca:

CornellUniversityPress,1990)Seeesp.pp.213234.

17 Iamnotsuggestingthatanyepistemicistwouldordoesendorse(2f).Iam suggestingthat(2f)mightadvancevanInwagen’s No Minimum argument.

ThefollowingargumenthasobviousaffinitieswithTimothyWilliamson’s antiluminosityargumentinhis Knowledge and its Limits (Oxford:Oxford

UniversityPress,2000),esp.sections4.2,4.3and4.6. 45

18 Ifthedefinitein‘ontheborderofevilunnecessaryfordivine purposes’isrestrictedtothatamountofevilthatisdefinitelyindefinitely unnecessaryfordivinepurposes,thenusetheindefinitedescriptionin‘ona borderofevilunnecessaryfordivinepurposes’foramountsofevilthatare indefinitelyindefinitelyunnecessaryfordivinepurposes.

19 SeeJohnHawthorne,‘VaguenessandtheMindofGod’ Philosophical

Studies 122(2005)125.InaveryinterestingdiscussionHawthorne considersthreedefinitionsofomniscienceincludingwhatIrefertoasO 1.Itis

O1 thatHawthornefindsmostplausible.

20 Cf.CianDorr,‘VaguenessWithoutIgnorance’, Philosophical Perspectives

Vol.17(2003)83114.Dorrurgesthateveryomniscientbeingknowsa propositionPifPistrue.Butaswehavenotedthisentailsthatomniscient beingsknowthat knisunnecessaryfordivinepurposesevenwhenitis superdefinitelytruethat knontheborderlinebetweenamountsofevilthat arenecessaryfordivinepurposesandamountsofevilthatareunnecessary fordivinepurposes.Andtomyear—thoughcertainlynottoDorr’s—this soundsawful.

21 ThisaccountissimilartoonediscussedinTimothyWilliamson’sin

Vagueness (Routledge:London,1994).Seethediscussionofoperator

‘definite*’p.160ff.