لطفا منتظر باشید ...
The Automation of Tabulation: Counting and Cross-Referencing the People
The word census comes from the Roman censor, an official responsible for the registration of citizens, evaluation of property, the spending of that revenue, and the guarding of public morals. The U.S. Bureau of the Census does the first task, and something of the second. But it is specifically prohibited from aiding in collecting taxes, and instead of guarding public morals census takers dutifully count households of unmarried couples—with no pointing of fingers. (Halacy 1980, 10)
In the above passage, law historian Dan Halacy implicitly champions the objectivity of the American census as a counting process whose workers make no moral assumptions as they ‘‘dutifully count households.’’ However, the first American federal census was defined in article 1, section 2, of the U.S. Constitution as ‘‘the whole Number of free Persons, including those bound to Service for a Term of Years, and excluding Indian not taxed, three fifths of all other Persons.’’ When President George Washington signed the census bill in March 1790, citizens therefore understood that the process would not be a objective count of heads that represented all inhabitants of the United States but, rather, that the population would be categorized along sexual and racial lines (Halacy 1980, 202).
The institutional and technological automation of such discriminatory practices was largely limited, however, until the ninth American census of 1870, when the chief clerk of the census, Colonel Charles W. Seaton, designed a simple tabulating machine (Alterman 1969, 225). Seaton’s system counted large numbers of census returns, but the age of diagnosable demography (meaning the study of data that can be efficiently counted and cross-referenced) was not fully realized until Herman Hollerith’s punch-card tabulating machine (1890). Hollerith, later one of the founders of International Business Machines (IBM), constructed a system that had three main advantages over previous census counting machines. Hollerith’s machine was efficient: it tabulated the results of the 1890 census approximately eight to ten times faster and twice as accurately as human beings had counted the results of the 1880 census. The punch-card tabulating machine also saved the government approximately $5 million (Alterman 1969, 225). As previously noted, though, the most important technological aspect of the tabulating machine was its ability to efficiently cross-tabulate items, making demographic profiles of particular groups based on relationships such as dwelling and income. Thus, according to an article published in The Electrical Engineer in 1891 (Martin 1891),
[Hollerith] saw that with increasing population and increasing complexity in data, the difficulties were becoming such that unless improved means of compilation were devised, the work must be abandoned in despair or become more incomplete and unsatisfactory each decade. On the other hand, with the aid of new facilities, not only might time and money be saved, but the data could be thrown into combinations full of suggestion and teaching, but which had been utterly beyond reach before. Such facilities Mr. Hollerith has furnished in his electric tabulating machine. The fundamental idea is . . . to punch holes in cards so that the positions of these holes will correspond to certain data, and then to pass these cards through presses by which the perforations in the cards are made to control the operation of electromagnetics or groups of magnets, which in turn energize counting mechanism or sorting boxes, or will bring both into play at once.
As a forerunner to the modern computer, Hollerith’s tabulating machine used technological logic that can be traced through census-taking and -tabulating technologies to the present day. For the 1940 census, for example, cross-referenced data from previous years were used to construct sample demographic groups that gave a statistically accurate profile of the entire nation. The technique of sampling allowed the census bureau to question only portions of the nation, saving a good deal of time and money in the process. According to Hyman Alterman (1969, 236), the 1940 census used sampling primarily to count and determine the domicile of army and navy veterans, the possession of Social Security numbers, and general details about regional occupations and industry.Hollerith’s techniques were further institutionalized within the U.S. Bureau of the Census with the acquisition of first-generation or UNIVAC I computers in 1951. Surprisingly, perhaps, the basic punch-card system devised by Hollerith was not replaced until the 1960 census, when the Film Optical Sensing Device for Input into Computers (FOSDIC) systems were operationalized. With the introduction of FOSDIC, enumerators merely transferred answers from individuals to a worksheet by filling in circles with pencils (Alterman 1969, 244).With the introduction of third-generation computers in the early 1970s that were some 600 times more efficient than UNIVAC, the demographic information compiled by the U.S. government literally took on a life of its own in the Census Bureau’s Data User Services Division (DUSD) (Halacy 1980, 56). Initially conceived as the marketing arm of the Census Bureau, DUSD received requests for data services from government officials, civic groups, charitable organizations, educators, students, and, significantly, market researchers (Halacy 1969, 181).
In addition to the census bureau’s mandate to collect demographic data, though, overall American governmental support—in the form of new technologies and data services—for the business sector first emerged as a distinct and coherent policy in the early 1920s. Commerce Secretary Herbert Hoover was one of the first government officials to address the lack of coordination or even knowledge of the relationship among the spheres of production, distribution, and consumption. Hoover recognized that ‘‘No one had tried to properly trace the movement of a single commodity from manufacturer to consumer’’ (Leach 1993, 354). In the years preceding his presidential bid, Hoover successfully spearheaded the formation of the Bureau of Foreign and Domestic Commerce (BFDC). In conjunction with the Bureau of the Census and with partial funding from the U.S. Chamber of Commerce, the BFDC oversaw the first ever ‘‘consumption’’ census from 1926 to 1928. Reports from the census point to the topographical utility of such cross-referenced pieces of information, which provided the business sector with information on ‘‘how best to deliver goods, widen streets, construct parking lots and underground transportation, employ colored lights, foster store circulation, and present merchandise in ‘tempting ways’ ’’ (Leach 1993, 365–366).Philip M. Hauser’s Government Statistics for Business Use (coedited with William R. Leonard, 1946) likewise offered over four hundred pages of in-depth analysis of government statistics and data related to manufacturing, agriculture, transportation, banking, housing, and national demographics. An early indicator of buying power, beyond land use and simple population count, could be found in the ‘‘Families: Income and Rent’’ index, which according to the index’s creators, greatly assisted in the ‘‘Determination of Sales Areas and Quotas’’ (Hauser & Leonard 1946, 355). Hauser and Leonard also detailed the importance of sex, age, race, place of birth, and education in determining the market for products and services.
