算法生活：大数据时代的计算设备与影响

需积分: 9 185 浏览量更新于2024-07-21 收藏 3.16MB PDF 举报

"Algorithmic.Life.Calculative.Devices.in.the.Age.of.Big.Data.1138852" 本书深入探讨了随着数字计算技术的发展而出现的各种计算形式和技术及其影响。作者们指出，数字计算设备不仅仅局限于其特定功能，它们在逐渐塑造、转变并治理我们生活的各个方面中发挥着重要作用。以下是一些关键知识点： 1. 大数据驱动的计算创新：大量多样化的数据处理需求如何推动新型计算设备、逻辑和技巧的开发？大数据时代，我们看到对数据的利用和理解成为创新的动力，这涉及到数据挖掘、机器学习和人工智能等领域。 2. 决策与行动能力的影响：这些计算设备、逻辑和技巧如何改变我们的决策能力和行动能力？大数据和算法可以提供更高效、精准的决策支持，但也可能削弱人类的主观判断，甚至导致算法偏见。 3. 数据化生活：我们的物理和虚拟存在如何转变为可分析和重组的数据？随着物联网、社交媒体等技术的发展，个人行为、偏好甚至情感都可以被转化为数据进行分析。 4. 公共与私密、个体与群体的界限：大数据如何重塑公共与私人的概念，以及个体与群体的关系？通过数据分析，我们可以更好地理解集体行为，但这也可能导致隐私问题的挑战。 5. 确定性与概率的新定义：寻找“隐藏”连接和模式如何改变我们对社会关系和关联生活理解？这种寻找模式的过程可能导致新的社会理论和预测模型的形成。 6. 计算可能性的边界：当代计算模式是否创造出新的可计算性和可计算性的阈值，使得原本不可能或仅仅可能的事物变得可行？这涉及到了未来预测和风险评估的领域。 7. 预测未来的治理方式：面对不确定的未来，新的计算和决策方法如何被重新引入？大数据和算法预测在政策制定和社会管理中扮演越来越重要的角色。本书分为四个部分，分别探讨算法生活、大数据时代的计算、信号可视化和计算，以及情感设备。每个章节都提供了理论深度和实证研究的案例，涵盖软件研究、大数据分析、信息决策、社会关系和未来预测等多个学术领域，是学生和学者的重要参考资料。

Introduction

LouiseAmooreandVolhaPiotukh

IfwegivethemachineaprogrammewhichresultsinitsdoingsomethinginterestingwhichwehadnotanticipatedI

should beinclined tosay that the machinehad originated something, ratherthanto claim that its behaviourwas

implicitintheprogramme,andthereforethattheoriginalityliesentirelywithus.

(AlanTuring,1951)

Introduction

OnacolddayinNovember2014,IBMexplaintoanassembledgrouphowtheirWatson

cognitive analytics engine learns about the relationships between things. Described as a

“technology that processes information more like a human than a computer”, Watson is

taughtwhattherelationsamongdatamightmean,ratherlikeachildistaughttoreadby

associating symbols and sounds (IBM, 2014a). “A subject specialist is required”, explain

IBM, in order to “teach Watson the possible relationships between entities”. The subject

specialistscouldbepolicingauthoritieswithknowledgeofcriminalbehaviours,orrevenue

and customs authorities with knowledge of patterns of fraud, or they could be medical

scientistssearchingforlinksbetweenexistingdrugsandnewapplications(IBM,2014b).It

can take around four months of what IBM call “nurturing” for Watson to learn these

subject-specificrelationshipsbetweenthedataelementsitingests.Oncethelearningfrom

atestdatasethastakenplace,however,Watsonisabletocontinuetolearnasnewitemsof

informationareaddedtothecorpusofdata.AsAlanTuringspeculatedsomesixty-three

years ago in his discussion of whether automated calculating machines could think, the

machine that is Watson does result in something interesting that had not been fully

anticipated inthe programme,andthus theoriginality doesnotlie entirelywith human

creativity.

How might we begin to think about the new forms of calculation that emerge with

digitalcomputation?Ofcourse,in onesenseunderstandingtherelationshipbetweenthe

algorithmandformsofcalculationisnotanovelproblematall.Understoodasadecision

procedure that predates the digital era, the origins of algorithmic thought have been

variouslylocatedinLeibniz’snotebooksoftheseventeenthcentury(Berlinski,2000:5)and

in the twentieth century mathematicians’ disputes on decidable and undecidable

propositions (see Hilbert, 1930; Gödel, 1965; Turing, 1936). Yet, with the twenty-first

centuryriseofbigdataandadvancedanalytics,thehistoricalquestionofcalculatingwith

algorithmicdecisionproceduresappearstobeposedanew.Indeed,the‘4Vs’of‘bigdata’–

increased volume, variety, velocity, and veracity of data elements (Boyd and Crawford,

2012;Mayer-SchönbergerandCukier, 2013)–demandnewkindsofcalculationandnew

kinds of human and machine interaction to make these possible. But what happens to

calculation, with the emergence of new ways of enumerating and modelling human

behaviour? How do new digital calculative devices, logics and techniques affect our

capacity to decide and act, and what are the implications for the governing of society,

economyandpolitics?Inaworldofchangingdatalandscapes,howdomundaneelements

of our existence become data to be analysed and rearranged in complex ensembles of

people and things? When the amount of available data is such that it exceeds human

capacitiestoreadandmakesenseofit,docontemporarymodesofcalculation,basedon

constantincorporationofheterogeneouselements,producenewthresholdsofcalculability

andcomputability,allowingfortheimprobableorthemerelypossibletobeembracedand

actedupon?Doessomethingoriginalemergeoutofthesecalculations,aswemightinquire

withTuring,somethingofinterest,whichhadnotbeenanticipatedintheprogramme?

Theaimofthisbook

istocriticallyexaminealgorithmiccalculativedevices,logicsand

techniquesthatemergeinaworldcharacterisedbyavastproliferationofstructuredand

unstructureddata.Thepredominantscholarlyandpublicemphasisonthe‘big’inbigdata

has tended to obscure what we call the ‘little analytics’, the arguably smaller and less

visiblecalculativedeviceswithoutwhichthisworldofbigdatawouldnotbeperceptibleat

all. As others have argued, the apparently vast array of contemporary data forms are

rendered“moreorlesstractable”viathealgorithmsthatmakethemamenabletoanalysis

(Hayles,2012:230;Hansen,2015).Ifthemetaphorofbigdataistocontinuetocaptureour

understandingofdigitallife,thenitcannothavemeaningwithoutalgorithmiccalculative

devices.Fromthefinancialsubject’sonlineaccesstosub-primelending(Devilleandvan

derVeldeninthisvolume)to thebiometricallyenabled battlefield(Nisa inthisvolume),

andfrompotentialpartnersandlovers(Mackinnoninthisvolume)topersonalisedurban

locations (Widmer in this volume), we are increasingly intertwined with algorithmic

calculativedevicesasweconsumeinformation,inhabitspaceandrelatetoothersandto

theworldaroundus.Yet,justasbeinghumanmayalsobecloselyenmeshedwithbeing

algorithmic,thesecalculativedevicesalsoalterperception,filteringwhatonecanseeofbig

datalandscapes,howonemakessenseofwhatcanbeperceived.AsEvelynRuppert,John

LawandMikeSavage(2013:24–25;originalemphasis)suggest,thereisaprofoundneedfor

“a conceptual understanding of the specificities of digital devices and the data they

generate”.

Inthisbook,adiverserangeofspecificalgorithmiccalculativedevicesandapplication

contextsarediscussed(e.g.,frominsurancetocounter-insurgency,fromfireandrescueto

addressing obesity, and from credit-rating to on-line dating). Beginning from a

commitmenttoexaminealgorithmicdevicesinsitu,thebookalsodevelopsanalyticaland

methodologicaltoolsforunderstandingcalculativelogicsandtechniquesthatreachacross

thediversedomains.

Beyondprobabilities:calculativedevicesofknowledge

discovery

Theuseofstatisticalcalculativedevicesforenumeratingpopulation–whatIanHacking

hascalled“themakingupofpeople”bythestate–layattheheartofnineteenthcentury

knowledgeofsociety(Hacking,1986;seealsoBowkerandStar,1999).Theriseofmethods

for population sampling and statistical analysis witnessed the emergence of profiles for

whatAdolpheQueteletcalled“l’hommetypique”,ortheaverageman,aprobabilisticfigure

whose attributes could be known and acted upon (Daston, 1995). Just as the nineteenth

century“avalancheofprintednumbers”(Hacking,1982)wastwinnedwithdevicessuchas

punchcardmachinestomakesenseofthenewlyavailabledata,sothetwenty-firstcentury

riseofdigitalbigdataisparalleledbyinnovationintheanalyticaldevicesrequiredtoread,

processandanalyseit.

Yet,wherethemanagementoftheavalancheofprintedandtabulateddataobservedby

Hacking was concerned with the capacity to index data in structured and directly

retrievableforms,theproliferationofdigitaldatatraceshasbroughtaboutvastquantities

ofunstructured,incompleteandfragmentaryelements.AsVictorMayer-Schönbergerand

KennethCukier(2013)observe,theriseofbigdatawitnessestwoparallelphenomena:an

expansion in what can be rendered as data, or “datafication”, and an extension of the

capacitytoanalyseacrossheterogeneousdataforms,suchasacrosstext,imagefiles,voice

orvideo.Inthisway,bigdatacanbeseenassimultaneouslyaproductof,andimpetusfor,

newdigitalcalculativedevices.Thecontributionsinthisvolumeprovidemanyexamplesof

thisdoubletransformation:fromonlinebehaviourturnedintodatathroughtracking(e.g.,

DevilleandvanderVelden)tobiometrics,includingvoiceandgait(e.g.,Nisa),andfrom

attitudes, opinions and interests, datafied as ‘likes’, ‘check-ins’, status updates (e.g., van

Otterlo; Widmer), to affects, emotions and feelings (attraction and love in Mackinnon;

anxietiesinColeman,butalsoinNisa,Belcher,O’Grady).

The twinned processes of data expansion and analysability are also significantly

challenging conventional social science understandings of what it means to draw a

‘sample’ of data from a population. The twenty-first century claim that “n=all”, or that

everything can now constitute the sample, extends the limit of sampling to an infinite

spatialhorizon(Gruhletal.,2004;Chiao-Fe,2005).Indeed,forsomecommercialpurveyors

of data analytics,thecore of the issueisto dispense with the notionofthesample and

samplingaltogether,sothatonecanworkwiththepatternsandcorrelationsofanygiven

dataset:

Datascienceisinherentlydiminishedifyoucontinuetomakethecompromiseofsamplingwhenyoucouldactually

processallofthedata.…InaworldofHadoop,commodityhardware,reallysmartsoftware,there’snoreason[notto

dothis].Thereweregoodeconomicreasonsforitinthepast,[and]priortothat,thereweregoodtechnical[reasons].

Today,noneof[thosereasons]exists.[Sampling]isanartefactofpastbestpractices;Ithinkit’stimehaspassed.

(Inbar,inSwoyer,2012)

Yet, although the rise of big data has extended the availability of data sets, the

completenesssuggestedby“n=all”isanillusion,accordingtoHildebrandt(2013).Oneof

theimportantreasonswhy‘n’cannevertrulyequal‘all’isbecause,asHildebrandtputs

theproblem:“thefluxoflifecanbetranslatedintomachinereadabledatainanumberof

ways and whichever way is chosen has a major impact on the outcome of data mining

operations”(2013:6;alsoKitchin,2014).Inthissenseitisinsufficienttomakeclaimsabout

theinfiniteavailabilityofdatawithoutcarefulattentiontohowitisanalysed,andtowhat

canbesaidaboutthedataonthebasisofthatanalysis.AsDanahBoydandKateCrawford

point out in this respect, there are many reasons why “Twitter does not represent ‘all

people’” (2012: 669), and so analyses of vast quantities of Twitter data cannot provide

insightsthatcanbemeaningfullysaidtorefertothepopulationasawhole.

In this book, we are concerned with the new calculative devices that have begun to

shape, transform and govern all aspects of contemporary life algorithmically. As Michel

CallonandFabianMuniesa(2003:190)haveproposed,

Calculatingdoesnotnecessarilymeanperformingmathematicalorevennumericaloperations…Calculationstartsby

establishingdistinctions between thingsor statesof the world,and byimagining and estimatingcourses ofaction

associatedwiththingsorwiththosestatesaswellastheirconsequences.

Though the work of contemporary algorithms does involve the performance of

mathematical functions, at least at the level of the machinic code (Dodge and Kitchen,

2011;Berry,2011),italsoactivelyimaginesandestimatescoursesofactionassociatedwith

things or states of the world. In this sense, and following others who have understood

marketcalculativedevicesasthingsthatdotheworkofmakingthemarketitself(Callon

andMuniesa,2003;MacKenzie,2006),forusalgorithmiccalculativedevicesarere-making

ourworldinimportantways.Indeed,asDavidBerry(2014:2)hasargued,“weareentering

a post-digital world in which the digital has become completely bound up with and

constitutiveofeverydaylifeandtheso-calleddigitaleconomy”.Whilethechaptersinthis

volumeexploretheworkofalgorithmiccalculativedevicesacrossmultipledomains,here

wewishtohighlightfouraspectsofalgorithmiclifethatsurfaceacrossthesepluralspaces.

First,calculativedevicesintheageofbigdataareengagedinthefilteringofwhatcanbe

seen, so that they create novel ways of perceiving the world and new visibilities and

invisibilities.InLauraPoitras’sAcademyawardwinningdocumentaryfilm‘Citizenfour’,

for example, Edward Snowden refers to the “ingestion by default” of “bulk”

communications data by the US National Security Agency (NSA). The vocabulary of

ingestioniscentraltodataminingpractices,wheretheprogrammeabsorbsthatwhichis

considered valuable, while filtering out that which is not of interest.

 The idea of data

ingestion suggests a qualitatively different process of “bringing something to attention”

from the traditional forms of data collection one might associate with social statistics

(Crary, 2013).From theLatin“in-generere”,tocarry into,to ingestsuggestsa processof

drawing in quantities of matter into an engine or body, such that the contents can be

filtered,someofthemabsorbedandothersexpelledordiscarded.Thecalculativedevices

剩余265页未读，继续阅读

ramissue

粉丝: 354
资源: 1487

算法生活：大数据时代的计算设备与影响

LaTex算法库——algorithmic.sty

Data.Structure.and.Algorithmic.Thinking.with.Python

LetMeRead.net__Data.Structures.and.Algorithmic.Thinking.with.Go.B08FT5V4B9.pdf

PHP.and.Algorithmic.Thinking.for.the.Complete.Beginner.1503015

Distributed.Systems.An.Algorithmic.Approach.2nd.Edition

Distributed.Systems.An.Algorithmic.Approach.2nd.Edition.zip

Discrete.Algorithmic.Mathematics.3rd.Edition.1568811667.epub

Algorithmic.Thinking.2020.11.pdf

CRC.Algorithmic.Cryptanalysis.Jun.2009.eBook-ELOHiM

算法设计技巧与分析课件(英文版)：ch1 Basic Concepts in Algorithmic Analysis.ppt

最新资源