Deep Root Analytics是创建并维护这个泄露数据仓库的共和党数据公司,2013年由亚历克斯·朗德里(Alex Lundry)与他人共同创办,朗德里是名共和党竞选数据科学家,曾在米特·罗姆尼失利的2012年总统竞选活动中担任数据总监。该公司自称是“共和党政坛中最有经验的精准定位者”,为力求覆盖目标群体的众多公司、游说团体以及共和党政治竞选班子提供媒体分析服务。Deep Root声称能够通过使用大数据分析工具实现“精准定位”(microtargeting),因而能够更有效地覆盖这些预期的群体,从而让客户在购买广告时得以做出更明智的决策。 这种出身背景让朗德里得以成为2016年前佛罗里达州州长杰布·布什(Jeb Bush)的共和党总统竞选班子的“首席分析官”。这边厢,布什即使在拉起一支资深的数据团队后还是未能获得提名;那边厢,特朗普的境况截然相反,他没有在竞选班子内部设立一支强大的数据团队,却获得了提名。RNC迅速将其数据团队的工作与特朗普竞选班子的工作协调起来,共同对付大选对手:希拉里·克林顿。 为了赢得选举,RNC需要大量运用几家专门从事数据分析的私人公司的资源。这些私人咨询公司当中就有Data Trust,总部位于华盛顿的这家公司声称“通过收集、开发和完善选民档案,不断地开发共和党及保守党数据生态系统。” 作为“共和党的独家数据提供商”,Data Trust由RNC于2011年创办,“承担构建和管理共和党选民档案的成本”――这个存储库含有详细的选民信息,这些信息对任何成功的选举广告和动员投票工作而言至关重要。据Slate报道,Data Trust作为RNC的私营附属企业来运作――“这是一家由政党首脑创办,但无法正式运营的混合型私人公司。” 在Deep Root Analytics数据库中,“data_trust”文件夹似乎正好包含这个RNC/ Data Trust合作项目的全部成果,这个全面而详细的数据库含有2016年潜在选民的信息。 “data_trust”里面有两个庞大的个人信息库,总共代表1.98亿潜在选民。每个信息库主要包括两个文件存储库,一个是2008年总统大选的256 GB文件夹,另一个是2012年总统大选的233 GB文件夹,每个文件夹又有51个文件(每个州一个文件,加上哥伦比亚特区)。每个文件的格式是逗号分隔的值(.csv)文件,列出了内部的32个字符长的字母数字“RNC ID”,用来识别数据库中每个潜在选民的独特身份,比如530C2598-6EF4-4A56-9A7X-2FCA466FX2E2。这些RNC ID以独特的方式将不同的数据集连接起来,结合了众多敏感的个人身份数据点,因而可以获得关于每个美国选民的大量详细信息。 维克里查看了这些电子表格后,确认这些文件含有准确、敏感的个人信息。下面列出了.csv类别: "RNCID", "RNC_RegID", "State", "SOURCEID", "Juriscode", "Jurisname", "CountyFIPS", "MCD", "CNTY", "Town", "Ward", "Precinct", "Ballotbox", "PrecinctName", "CD_Current", "CD_NextElection", "SD_Current", "SDProper_Current", "SD_NextElection", "SDProper_NextElection", "LD_Current", "LDS_Current", "LDProper_Current", "LD_NextElection", "LDS_NextElection", "LDProper_NextElection", "NamePrefix", "FirstName", "MiddleName", "LastName", "NameSuffix", "Sex", "BirthYear", "BirthMonth", "BirthDay", "OfficialParty", "StateCalcParty", "RNCCalcParty", "StateVoterID", "JurisdictionVoterID", "AffidavitID", "LegacyID", "LastActiveDate", "RegistrationDate", "VoterStatus", "PermAbs", "SelfReportedDemographic", "ModeledEthnicity", "ModeledReligion", "ModeledEthnicGroup", "HHSEQ", "HTSEQ", "RegistrationAddr1", "RegistrationAddr2", "RegHouseNum", "RegHouseSfx", "RegStPrefix", "RegStName", "RegStType", "RegstPost", "RegUnitType", "RegUnitNumber", "RegCity", "RegSta", "RegZip5", "RegZip4", "RegLatitude", "RegLongitude", "RegGeocodeLevel", "RADR_LastCleanse", "RADR_LastGeoCode", "RADR_LastCOA", "ChangeOfAddress", "COADate", "COAType", "MailingAddr1", "MailingAddr2", "MailHouseNum", "MailHouseSfx", "MailStPrefix", "MailStName", "MailStType", "MailStPost", "MailUnitType", "MailUnitNumber", "MailCity", "MailSta", "MailZip5", "MailZip4", "MailSortCodeRoute", "MailDeliveryPt", "MailDeliveryPtChkDigit", "MailLineOfTravel", "MailLineOfTravelOrder", "MailDPVStatus", "MADR_LastCleanse", "MADR_LastCOA", "AreaCode", "TelephoneNUm", "TelSourceCode", "TelMatchLevel", "TelReliability", "FTC_DoNotCall", "PhoneAppendDate", "VH12G", "VH12P", "VH12PP", "VH11G", "VH11P", "VH10G", "VH10P", "VH09G", "VH09P", "VH08G", "VH08P", "VH08PP", "VH07G", "VH07P", "VH06G", "VH06P", "VH05G", "VH05P", "VH04G", "VH04P", "VH04PP", "VH03G", "VH03P", "VH02G", "VH02P", "MT10_Party", "MT10_GenericBallot", "MT10_Turnout", "MT10_ObamaDisapproval", "MT10_Jobs", "MT10_Healthcare", "MT10_SoCo", "PG01", "PG02", "PG03", "PG04", "PG05", "PG06", "PG07", "PG08", "PG09", "PG10", "PG11", "PG12", "PG13", "PG14", "PG15", "PG16", "PG17", "PG18", "PG19", "PG20", "PG21", "PG22", "PG23", "PG24", "PG25", "PG26", "PG27", "PG28", "PG29", "PG30", "PG31", "PG32", "PG33", "PG34", "PG35", "PG36", "PG37", "PG38", "PG39" (责任编辑:本港台直播) |