Computers in Love

Stanford and the First Trials
of Computer Date Matching

by C. Stewart Gillmor

Time magazine erred greatly in 1963 when it attributed
the first computerized dating service to Iowa State Univerity.
Four years earlier, two Stanford students - intent on a good party-
devised a "Happy Families Planning Service" as a final math class project,
pairing up 49 men and 49 women. The results may have been mixed,
but the experience reveals yet another side to romance,
and to engineers at Stanford
[1].


The idea of mixing computers and love began in the fall of 1959 when two electrical engineering students, Jim Harvey and Phil Fialer, decided to take Professor Jack Herriot’s computer course. Math 139, Theory and Operation of Computing Machines, was a popular course among math majors, electrical engineers, and graduate students in various science and engineering programs. Using Stanford’s relatively new IBM 650, they could gain proficiency in large-scale data analysis techniques using machine language assembly coding (or the relatively new IBM FORTRAN—FORmula TRANslation —computer language). The course required a term project, which gave students the rare opportunity to use Stanford’s otherwise closely guarded mainframe computer.

The virtually complete files of Harvey and Fialer’s project, the "Happy Families Planning Service", document this creative undergraduate project as the earliest known computer date matching service in the United States[2].

The Hardy Boys Meet Computer Science: Harvey and Fialer

Harvey and Fialer wanted to learn programming but they were, no doubt, thinking of the project from another angle as well. After all, the pair had been among the stalwarts of student radio station KZSU who had been thrown off the air the year before for causing illegal radio interference, operating with excessive transmitting power, and broadcasting less-than-tasteful radio programs—especially the steamy potboiler, the “Stanford Sadie” show[3]. Both men had career records of fooling around with scientific and technical hobbies.

From junior high school through college, Jim Harvey had more than one encounter with the telephone company while inventing, and utilizing, his own "toll-free" telephone connection boxes and "automatic" electronic dialing system. Telephone companies in Oregon and California may have appreciated Harvey’s youthful enthusiasm and creativity, but they apparently felt that experimenting with the Bell System was best left to Western Electric and Bell Labs.

Phil Fialer was a youthful adventurer who had moved from color photography development and primitive stereophonic music experiments to inventive radio applications. Within a week of settling into Wilbur Hall in fall 1956, Fialer and a pal were summoned to Wilbur dormitory mistress Mrs. Summers. The pair had installed a ham radio-transmitting antenna in the attic eaves and succeeded in blotting out TV and radio reception for most Wilbur Hall residents. Later, the same duo built a ham transmitter from a table model AM radio and strung an "invisible" (very thin) magnet wire antenna from the dorm window to a tree on the Wilbur lawn. This caper, however, proved ineffective --- the grounds maintenance crew removed the "invisible" antenna.

Enter Stanford’s Mainframe

By 1959, the older and wiser Fialer and Harvey had both enrolled in fall quarter’s Math 139 to learn how to program computers - in particular, Stanford’s most powerful computer, the IBM 650.

In the 1950s, mainframe computers were largely unavailable to students and the public (machines such as the Illiac at Illinois, Whirlwind at Princeton, and NORC, were built by IBM for governmental and military use)[4]. Following IBM’s delivery in 1954 of the first 650 to John Hancock Mutual Life Insurance of Boston, nearly 2,000 machines would be sold by 1962, making it the most popular computer by the late 1950s[5].

Universities wanted in on the act, and Stanford got its first mainframe computer, an IBM CPC (Card- Programmed electronic Calculator), in 1953. Stanford’s IBM 650, rented for $3,750 per month, was installed in the Hewlett-Packard teaching wing of the Electronics Research Laboratory (ERL), then located near the center of what is now the Science and Engineering Quad, or SEQ), in January 1956. To complete its reading, adding, subtracting, and printing instructions, the IBM 650 machine used 2,200 vacuum tubes and 4,000 diodes. The vacuum tubes generated so much heat that the Computer Center of the ERL was the first (and, in 1959, the only) portion of the building to be air-conditioned.

Though not classified as a large computer, the IBM 650 certainly seemed large to students in 1959. It was nearly three times more powerful than the IBM CPC. The IBM 650 was considered the “biggest” of the small computers—its closest challenger was the ElectroData Corporation’s Datatron. (The IBM 650 required 16 Kilowatts of power—100 amperes at 220 Volts AC—and the Datatron 13 Kilowatts.)

For technophiles, the IBM 650 had a 2,000-word memory drum, and an average access time of 2.4 milliseconds, using a decimal system with 10 digits per word. Its top speed was 22 to 33 operations per second. Its cardpunch and printer functioned at exactly half that speed. The IBM 650 was the only “small” computer to have double precision arithmetic capability, very useful for statistical problems. The computer-processing unit was coupled to a large Type 655 power unit and a Type 533 electromechanical card read-punch unit. Its card reader could process 200 punched cards per minute, or 265 digits per second. For printed reports it could also be connected to a Type 407 accounting machine output printer. Input was largely via 80- column Hollerith cards. Data on each card was prepared using a typewriter keypunch unit that punched rectangular holes in the cards[6].

As the course progressed, the Math 139 class learned to express various functions in computer code and to prepare the logical flow charts and language instructions —Enter, IF NOT-THEN, Add, Subtract, Sum, Do Loop, and so forth. Weekly homework assignments kept the students in line. Using the IBM 650 itself, however, was a treat saved for the end of the course.

The Great-Date-Matching Party: The "Happy Families Planning Service"

The notion of melding Math 139 with the Great- Date-Matching Party surfaced early in the quarter when Fialer and Harvey needed to come up with a term project.

For some time, Fialer and Harvey had hosted parties in houses that they rented with several electrical engineering and KZSU buddies at 1203 and 1215 Los Trancos Woods Road in Portola Valley. Student nurses from the Veterans Administration psychiatric hospital on Willow Road in Menlo Park were often invited. The boys represented themselves to the nurses as the “SRI Junior Engineers Social Club”—which was at least partially true, since one Los Trancos housemate worked summers and part-time as a junior engineer at Stanford Research Institute (SRI). KZSU radio station parties also were held in Los Trancos, featuring the KZSU musical band marching around the Los Trancos circle loop road at midnight. (This somewhat impromptu band was the basis for the current Los Trancos Woods Community Marching Band, officially organized at a KZSU party on New Year’s Eve in 1960.)

Fialer and Harvey figured a KZSU-Los Trancos type party could emerge as a positive by-product of Math 139, using the computer to match “a given number of items of one class to the same number of items of another class.” The classes would be male and female subjects, and the population would be Stanford students, with a few miscellaneous Los Trancos Woods residents thrown in.

The pair wrote a program to measure the differences in respondents’ answers to a questionnaire. A “difference” score was then computed for each possible male-female pair.

The program compared one member of a “class”—one man—with all members of the other class—women—and then repeated this for all members of the first class. The couple—a member from each class—with the lowest difference score was then matched, and the process repeated for the remaining members of each class. Thus, the first couple selected was the “best” match. As fewer couples remained in the pool, the matched couples had larger and larger difference scores.

Given the limitations of computer time available and the requirements of the course, Fialer and Harvey did not use a “best-fit” algorithm, so the last remaining pairs were indeed truly “odd” couples. Two of the women in the sample, not Stanford students, were single mothers with two or three children. One of them, age 30, ended up paired with a frosh member of the Stanford Marching Band.

The Personal Questionnaire

The experiment would match 50 couples, one of which was a “dummy” couple added to the project for test-checking purposes. The sample was hardly random, as it included many friends of the duo, and other KZSU regulars. Questionnaires also were distributed in dormitories.

The social-psychological profile the two enterprising scientists prepared was, to put it mildly, not the most sophisticated of social science tools. As Fialer recalls, “Well, Harvey had taken Psych 1....” Citing no scientific literature, they simply modeled the matched pairs by similarity of answers—certainly not the adage “opposites attract.” Thirty questions asked for such data as age, height, weight, religion, and hobbies, as well as personal habits, personality traits, and number of children the respondents wished for in a marriage. They were asked to note their drinking and smoking habits, and those habits they hoped for in their preferred mate.

Some questions were graded across several levels —from most introverted to most extroverted, passionate to frigid, aggressive to shy, lethargic to ambitious. Respondents were asked to indicate from a list their three most favorite and three most disliked hobbies or pastimes. From another selected list, they checked off their political stance (ranging from Communist, Socialist, Democrat to "New-Deal Republican", "Old-Guard Republican", and Fascist).

On religion the testers gathered the responses into six rather unusual groupings: “Mystery Cults”; Jews; Quakers and Mormons; “Protestants and Non-Religious”; “Christians, etc.”; and "Catholics".

The Answers

Most of the population sampled was between 18 and 22. Several graduate students were as old as 25, and the few non-students included three up to the age of 30[7].

As would be expected from the Stanford student body of 1959, the predominant religious preferences were Christian, totaling 70 of the 98 responses. Fifty of these were Protestant and 10 were Roman Catholic. Among Protestant denominations, Presbyterians led with 10 and Episcopalians with 9 respondents. There were 2 Jews, 1 Christian Scientist, and 1 Quaker. The only significant difference by sex was the number of men (16) versus women (8) who entered their religious preference as “non-religious” or “agnostic.”

As to politics, not one respondent checked Communist or Fascist, although 2 men and 2 women bravely checked Socialist; 2 of these were non-students. Republicans (51 “New-Deal” and 17 “Old-Guard”) significantly outnumbered the 24 Democrats. Men were very slightly more likely to be Democrats.

As for smoking and drinking, both men and women wanted a mate whose habits were similar. Women were slightly more permissive, willing to accept their future mate’s desire to smoke or to drink. For example, a woman not smoking might check that her preference was for a mate who smoked occasionally; several noted the desirability of a man who “smoked a pipe.” Most of those who smoked or drank did so moderately or occasionally, at least as reported. Interestingly, women were slightly more likely than men to admit to smoking and drinking. (Of the 49 males responding, 20 smoked, 42 drank and 7 did neither. Of the 49 female respondents, 21 smoked, 46 drank and 3 did neither.) In all cases, if a respondent smoked, then that person drank alcohol at least occasionally.

Women preferred larger families. The mean number of children desired in a marriage was 3.5 for women, 3.1 for men. (The mode was 4, followed by 3 and 2 children desired.) No one in the sample wanted just one child, and only one male hoped for a childless marriage. On the other hand, women outnumbered men 6 to 3 among those who wished for 5 or more children.

Without performing a more complete statistical analysis, it is difficult to determine whether the women had a significantly different profile than the men on selfperceived personality traits. For example, a number of women as well as men viewed themselves as “very ambitious.” It would be uncharitable to suggest that the number of men who checked “very passionate” as a self-attribute, or those few non-drinking men who checked “heavy” drinker as desirable in a female mate, were fantasizing about their chances for drawing a “hot date.”

Running the Computer Program

The students in Math 139 had been required to submit their punched cards to a computer operator who would then run their program. That fall, the course had an enrollment of 150 students. The official total logged-in usage of the IBM 650 for the class was a little more than seven hours for the entire quarter—that is, each student team got to use the mainframe for be- 6 tween five and ten minutes[8].

Since user time on the IBM 650 was carefully restricted, Fialer and Harvey explained in their report to Professor Herriot that it would take 90 minutes to do the complete sort on the 50 couples, and they were thus limited to running the model through for just 10 couples.

Working late at night, however, and unknown to Professor Herriot (or to the computer center staff), the boys went ahead and ran the program for all couples. A number of KZSU students were expert with lock-picking sets and carried them for "emergency use"; this skill aided their entrance to the computer lab after hours. Harvey and Fialer ran the program three times, fixing minor errors along the way, and using as much time as the rest of the class combined.

Fialer recalls that the ice cream machine in ERL that night required no money to give out ice cream bars, and thus the boys were sustained throughout their lateevening efforts.

Field Research in Computer Science: The Party Site

At the end of the fall quarter, the Great-Date-Matching party was planned for Harvey’s residence at 1215 Los Trancos Woods Road. Among his many research interests, Harvey had been practicing the art of malt brewing, and several cases of home-brew rested in his dirt-floor basement. Root beer malt rather than regular beer malt had been used and occasionally a bottle blew off its cap, shooting alcoholic root beer up to the floorboards of the main portion of the house. As a result, the residence had an unusual smell of a mixture of Anchor Steam and A&W.

In addition to the homemade beer and other potables, the party provided a chance to hear the KZSU musical regulars, typically consisting of piano, banjo, guitar, clarinets, cornet, trombone, and several kazoos.

All of the 98 respondents were invited to the party. Some couldn’t make it: One young woman got engaged during the trials (not to her prospective "date"), and one young man, a radio science graduate student, callously decided to go skiing instead.

Field Research: Test Case

It was assumed that the closer the matching scores of a given couple, the greater the chance for computer-aided romance.

Harvey and Fialer quietly slipped the news to male cronies whose matching scores were low, and who therefore were liable to be lucky in love. Scores ranged from under 100, for two lucky couples, to just over 100 for several others. The worst mismatches produced scores above 600.

This author’s score was fourth best (lowest) of the 49 couples! The young lady with whom he was paired, Miss McD., was an 18-year-old frosh from Roble, and, like the author, bore a Scots-Irish name. She was 5’ 7" and 130 lbs., had reddish-hair and freckles. Though listing her religion as “uncertain,” in fact, she said her parents were Presbyterian. Miss McD. indicated that she was neither “passionate” nor “frigid,” and was of average attractiveness. She managed simultaneously to be, according to her questionnaire, somewhat aggressive, stubborn, and irritable, yet slightly affectionate and understanding.

Her lucky date was then 21 years old, 5’ 10” and 160 lbs. While he listed his religion as “Universalist,” he, too, was a Presbyterian. As did a number of the young men, he had listed himself as very passionate. He also saw himself as amiable, ambitious, and of average attractiveness.

Miss McD. liked reading, camping and dancing, and disliked boating, dinner parties, and cooking. The author also liked reading, camping, and concerts, but disliked dancing, dinner parties, and home/shop projects. So, except for their views of dancing, the two were a fairly close match.

One must note that the partygoers in 1959 were not privy to these raw data scores or specific answers, and the author did not learn of the recorded self-perceptions of Miss McD. until 42 years later.

On the big Saturday night, the adventuresome youth picked up Miss McD. from Roble Hall early in the evening. He turned his Austin-Healey 100-6 sports car up Alpine Road toward Los Trancos Woods. Their relationship hit an initial bump when Miss McD. informed the driver that he was preceding at somewhat too fast a pace. Upon arrival at the party, the young lady expressed the thoughts that the house in the woods was odd, the band too loud, the party guests rather strange, and the homemade alcoholic root beer not to her liking. While things at the party overall seemed to be going rather well for some of the couples, by 10 p.m. the youthful male asked Miss McD. whether she would prefer to return to Roble. She replied definitely in the affirmative, and the duo motored back to campus, with Miss McD. constantly keeping the driver fully aware of his weaknesses as an operator of a motor vehicle.

The Fate of "Happy Families Planning Service"

This author did not return to the Los Trancos party that evening, so he cannot report on the overall success of the computer date matching experiment. Nor was any follow-up on the matched couples attempted by the research team. Among the 30 or so persons at the party known to this author, no marriages seem to have resulted. Certainly the 30-year-old single mother and the frosh Stanford Band member did not pursue their relationship.

Interestingly, in the Iowa State faculty-designed computer matching experiment of 1963, similar answers on matters of opinion were favored for matching, while on personality traits, the theory was that “opposites attract.” Perhaps Iowa applied more advanced psychological theory.

Whatever the social and psychological merits of the computer dating project, the “Happy Families Planning Service” received an “A” from Professor Herriot. In an attempt to raise his Stanford grade point average and, of course, to refine his Happy Families experiment, Harvey enrolled in the succeeding quarter (winter 1960) in an Education School course titled “Marriage and the Family.” Harvey proposed as his project for that course to re-tune the computer model and again conduct a date matching experiment with 50 couples. The Education professor strongly endorsed Harvey’s proposal.

The second time around, the party was held at the end of the quarter, again in Los Trancos Woods. Although there was virtually no difference in the questionnaire itself (or in the denouement), there had been a significant difference in participation. Without being too sexist, the second survey included students enrolled in the Education Department course itself, including some of the real campus beauties and Quad yearbook queens—Sue G., Maurine M., Dixie M., Carol P., to name a few—as well as Willa G. Wonka, cousin of the ubiquitous Warren G. Wonka.

This author, always willing to help in scientific research for the common good, again was a participant in the computer date matching survey. Again, the overall low (best) matching score was under 100 and the worst over 600.

The author was matched this time with a Miss W., a junior and member of the Stanford Rifle and Pistol Club (!). They were the sixth best couple with a score of 102, only two points out of third place. And again, hopes were high for computer-driven love. A slightly muted but similar outcome resulted. The author’s dating experience with Miss W. did not extend beyond that one occasion.

Looking Back

Stanford’s computer activities had been initiated in the early 1950s by such prominent faculty as Fred Terman, Albert Bowker, Jack Herriot, Allen Peterson, and others. By 1958, a computer science academic program was well on its way with the arrival of computer scientist George Forsythe and access to Stanford’s IBM 650. The Computer Science Department became one of the country’s truly outstanding programs, but such serious endeavors cannot hide the fact that it also fostered the nation’s first foray into Computers in Love.

Shortly after working on the computer date matching experiments, both Harvey and Fialer graduated from Stanford, Harvey in April and Fialer in June of 1960. James A. Harvey joined Lockheed Missiles and Space Corporation and later worked for Philco Corporation, currently known as Loral Corporation. He retired in 2000 as Aerospace Program Manager of Loral. Philip A. Fialer also joined Lockheed soon after graduation, returning to earn a master’s degree in electrical engineering from Stanford in 1964 and Ph.D. in 1970. He subsequently worked for the university and for Stanford Research Institute (now SRI-International), leaving there to co-found Mirage Systems, where he has served as president. He is currently chief technical officer of Mirage.

Neither has experimented further, however, with mixing computers and love.


C. Stewart Gillmor, ’60, is professor of history and science at Wesleyan University. For the years 1998- 2001 he was visiting professor of electrical engineering at Stanford while writing a biography of Frederick E. Terman. A KZSU jazz disk jockey and cornet player, and more than simply an eyewitness to the above events, his grade-point average suffered mightily from his escapades.


ENDNOTES

  1. An Iowa State University administration operation (a date-matching student dance using computers) was erroneously reported in Time as the first such experiment. “216 Meets 14,” Time, 25 October 1963, p.102. This oversight was soon corrected in a letter to the editor by Stew Gillmor, Time, 8 November 1963.
  2. Philip A. Fialer and James A. Harvey, Math 139 Happy Families Planning Service Project file. The original 1959 computer program, flow charts, data input questionnaires and output results are preserved in the private collection of James Harvey.
  3. C. Stewart Gillmor, "Stanford Sadie and the Early Years of KZSU Radio Broadcasting", Sandstone & Tile 23, no. 1 (Winter 1999): 11-20.
  4. Other large computers available for public use in the period 1956-59 included the IBM 701, 702, and 704 series, and the Univac I and II. In 1953, IBM had announced the IBM 650 "Magnetic Drum Calculator" to handle diversified accounting and scientific computations. It was widely used by universities and businesses. For a description of U.S. computers in 1956, see John W. Carr, “Solving Scientific Problems,” Control Engineering 3 (January 1956): 63-70; Carr and Alan J. Perlis, “A Comparison of Large-Scale Calculators,” Control Engineering (February 1956): 84-92; and Carr and Perlis “Small-Scale Computers as Scientific Calculators,” Control Engineering (March 1956): 99-104.
  5. See Charles J. Bashe, Lyle R. Johnson, John H. Palmer and Emerson W. Pugh, IBM’s Early Computers (Cambridge: MIT Press, 1986).
  6. IBM announced in 1956 that the 650 would have available as an option in the near future magnetic tape reading, alphanumeric data handling, floatingpoint arithmetic, and magnetic-core storage. This was in addition to its thenavailable numerical data handling, magnetic drum storage, and fixed-point arithmetic abilities.
  7. "No one ever analyzed the student questionnaire,” the author recently wrote the editor, “so I did so in my seat on airplane flights.”
  8. SC160/III/61/2, George E. Forsythe, Report of Ad Hoc Committee for the Stanford Computation Center, Appendix by John Herriot, February 19, 1960.


This material is covered by copyrights by Stanford Historical Society (2002), C. Stewart Gillmor (2002), and Stanford University (2002)

The material is published in "Sandstone&Tile", Summer/Fall 2002 Stanford Historical Society, Volume 26, No. 2-3 "Computers in Love: Date Matching at Stanford"

(HTML file updated July 16, 2003)