Computers, Education, and Public Policy
`
`Computerizing Information:
`Lessons of a Videbtext Trial
`
`by William Paisley
`
`Taking into account both content and audience,
`the objectives of one videotext endeavor
`could have been met more simply and reliably by
`a broadcast-based teletext delivery system.
`
`“Green Thumb” was indeed a videotext “trial” to the Kentucky farmers.
`Sometimes the system worked; sometimes it flashed “abort”; sometimes
`it didn’t work at all. Farmers had to enter their requests before learning
`whether the system was “up” or “down.”
`The Green Thumb electronic text system was designed to provide
`better weather and farm-management information to the farmers. Yet, at
`the mid-point of the trial, weather information failed to update about 60
`percent of the time. (The record subsequently improved to about 20
`percent of the time.) Late in the trial, because of a computer change
`hundreds of miles away, commodity market updates ceased for seven
`weeks.
`But this is not a cautionary tale of how not to run a videotext trial.
`Despite the technological and logistical mishaps that can be expected in
`a first-generation service, Green Thumb, the U.S. Department of Agri-
`culture’s rural videotext service, provided valuable information to Ken-
`
`William Paisley is Associate Professor of Communication at Stanford University. The
`Green Thumb evaluations were supported by the Extension Service, U.S. Department of
`Agriculture. Stanford evaluators included Donald Case, Milton Chen, Hugh Daley, Joung-
`Im Kim, Nalini Mishra, William Paisley, Ronald Rice, and Everett Rogers. Evaluators at
`the University of Kentucky were Paul Warner and Frank Clearfield, Department of
`Sociology. The Green Thumb trial was supervised by John Ragland, director of the
`Kentucky Cooperative Extension Service. The author alone is responsible for conclusions
`drawn in this article.
`
`153
`
`Exhibit 2004
`IPR2015-01077
`
`

`

`Journal of Communication, Winter 1983
`
`tucky farmers and yielded rich data for the Kentucky and Stanford
`researchers charged with evaluating it. After describing some of the
`available electronic text systems, this article reviews the findings of the
`Green Thumb evaluation with the goal of suggesting how content,
`audience, and delivey system could be matched in future electronic text
`systems.
`During the brief history of videotex in North America, Europe, and
`Japan (see 1, 2, 3, 4, 5), the terms videotex (or videotext),’ teletext (or
`broadcast videotex), and viewdata (or interactive videotex) have been
`used inconsistently. Conventionally, videotex is understood to encom-
`pass both broadcast and interactive electronic text transmission. In its
`generic usage, videotex is distinguished from broadcast or cable trans-
`mission of camera images, which are not textual, and from microcomput-
`er text displays, which are not transmitted?
`Teletext refers to electronic text transmitted during the vertical
`blanking interval of a broadcast television signal, although teletext can
`also be transmitted in dedicated channels, such as FM subcarriers.
`Teletext “frames” are broadcast in a constantly repeated sequence; the
`number of frames is limited by the length of time the sequence takes to
`repeat, since the user must wait for a given frame to be broadcast before
`it can be “captured” by the teletext decoder and displayed on the screen.
`At a transmission rate of about five frames per second (high-resolution
`frames take longer), one thousand frames comprise a maximum se-
`quence. In order to minimize waiting time for popular frames, these are
`broadcast two or more times in each sequence.
`The typical content of a teletext system, such as England’s Ceefax/
`Oracle, reflects the premium placed on sequence capacity as well as the
`content resources that are available to broadcasters as system operators.
`News, including sports, weather, and financial information, timetables
`and event calendars, directories, advertisements, and shopping catalogs
`are the preferred content of a teletext system because they are concise,
`widely used, and in some cases profitable.3
`Znteractive videotex, which is simply called videotext in describing
`systems like Green Thumb, refers to text transmitted via cable, tele-
`phone, or other non-broadcast channels. Typically, videotext is also
`
`There are several explanations of why videotext lost its final “t.” The International
`Telephone and Telegraph Consultative Committee, whose cachet makes the term official,
`may have wanted a multilingual term. Tandy, a Texas-based pioneer of American videotex,
`may also have had something to do with it.
`2There are hybrid systems, such as HI-OVIS in Japan and QUBE in the United States,
`that allow camera images and text to be intermixed in transmission.
`Some frames can be shared by seldom-requested content that is not broadcast in
`every sequence. A user group (e.g., physicians, real estate agents, antique collectors) can
`retrieve its frames for a certain number of minutes each quarter hour at a given starting
`time and digital address in the sequence.
`
`154
`
`

`

`Computers, Education, and Public Policy I A Videotert Trial
`
`received on ordinary television sets fitted with decoders. Unlike tele-
`text, which merely “captures” a signal that is being broadcast anyway,
`videotext requires a request channel from the user to the distribution
`center. Requested frames are sent to the individual user alone. Because
`it is not necessary to broadcast videotext frames in a frequently repeated
`sequence, the content of a videotext system can be orders of magnitude
`larger than the content of a teletext system. For example, one early U.S.
`videotext system, Channel 2000 in Columbus, Ohio, held a 32,000-page
`encyclopedia and the card catalog of a large public library, as well as
`other extensive resources.
`Discussions of teletext and videotext have focused upon the transmis-
`sion difference. However, a videotext system like Green Thumb was
`functionally equivalent to a teletext system like CeefadOracle and
`functionally different from videotext systems like Prestel or Channel
`2000. These differences can be clarified by the concept of inteructiuity.
`Interactivity is defined as the ratio of user activity to system activity.
`At one extreme, known to cable viewers, text scrolls on the screen
`without any user control. Since the text display doesn’t “need” the user
`to complete its programmed sequence, interactivity is zero. At the other
`extreme, exemplified by the “Chat” or “CB” services of The Source and
`CompuServe videotext systems, there is near parity between user
`activity and system activity; the interactivity ratio is about one to one.
`For the modal videotex function of retrieving information, neither
`extreme of interactivity is optimal. In the middle range of interactivity,
`the user commands maximum system response with minimum input.
`The “90/50 rule” of information retrieval applies here: in a well-
`designed interactive system, a user should be able to access 90 percent
`of the system’s resources with 50 percent of the effort that would be
`required to access 100 percent of the resources.
`
`The case of “Green Thumb” illustrates how a videotex
`system, designed to serve a broad range of needs,
`was reshaped b y circumstance for limited uses.
`
`The concept of a rural electronic text system (not necessarily video-
`text) was developed in 1976 by Howard Lehnert of the U.S. Department
`of Agriculture and Harold Scott of the National Weather Service. Scott
`had pioneered the all-weather FM radio stations for the National
`Oceanic and Atmospheric Administration and was aware of their limita-
`tions. Using voice-only weather information, it might take more than two
`hours to cycle through the weather information of potential interest to
`farmers in one locale, while an individual farmer might be interested in
`only a few minutes of that information.
`The Lehnert-Scott proposal came to the attention of Senator Walter
`Huddleston of Kentucky. Huddleston’s legislative aide, William Seale,
`
`155
`
`

`

`Journal of Communication, Winter 1983
`
`began to work with Lehnert and Scott to locate funds for a trial. Lehnert
`and Scott had named their system “AGWEX” to denote its weather-
`information function. However, a secretary in Huddleston’s office began
`calling it the “Green Thumb,” and the name stuck.
`By late 1977, configuration for the system had been decided. A low-
`cost decoder and storage unit, the Green Thumb Box, would use the
`farmer’s existing telephone connection and television set. Information
`would be downloaded4 into the Green Thumb Box to minimize tele-
`phone charges and to free the telephone for other use. The main
`computer for Green Thumb (a Hewlett-Packard 3000) would be located
`in Lexington (in 1978, Kentucky had been chosen as the trial state) at the
`University of Kentucky’s Agricultural Data Center, and microcomputers
`would be located in each county served by Green Thumb. After
`information was updated in the main computer, it would be transferred
`to the county computers, which were within farmers’ local dialing area.
`Certain “feeds” to the Green Thumb system were intended to be
`semi-automatic. Computerized weather information was received in
`Lexington and processed into the Green Thumb data base. Computer-
`ized market information was received from two sources: cash prices from
`the Department of Agriculture’s Agricultural Marketing Service, and
`futures prices from the Chicago commodity boards via the Commodity
`News Service.
`In addition to weather and market information, Green Thumb was
`supplied with advisories concerning agricultural production and man-
`agement, home economics, 4-H/youth, and community development.
`Most of this information was entered into the data base by extension
`specialists in Lexington, although county agents could add their news
`and advisories to it. The total number of frames in the system fluctuated
`from month to month, but averaged about 250 in a typical month.
`Weather maps and forecasts, along with market prices, were displayed in
`30 to 35 frames each, as was information about home economics and
`plant diseases. Horticulture, agronomy, and county affairs would usually
`be covered in 15 to 25 frames and such topics as 4-H activities, rural
`sociology, and community development would receive at least five
`frames each. A table of contents and an overview of the data base were
`also included.
`Two rural Kentucky counties, Shelby and Todd, were chosen for the
`trial. Committees composed of farmers and extension agents selected
`100 farm families to receive Green Thumb Boxes in each county.
`
`“Downloading,” otherwise known as “dump and disconnect,” is the videotext
`procedure of transmitting all of the user’s requested frames at once into the decoder’s
`memory, then terminating the telephone or cable connection so that other users can access
`the videotext system over the same lines. The downloaded frames can be paged backward
`and forward for reading; they can be held indefinitely in the decoder’s memory. The
`“Green Thumb Box” held about twelve frames, depending on the amount of text in each
`frame.
`
`156
`
`

`

`Computers, Education, and Public Policy I A Videotert Trial
`
`Although operators of large farms were over-represented on the list of
`volunteers, the selection committees assigned the Green Thumb Boxes
`to a cross-section of large and small farm operators in each county.
`The experimental phase, beginning in March of 1980, was consid-
`ered to be over when the evaluation data were collected in June of 1981.
`The system is still available to farmers.
`The Stanford evaluation team, with the help of local interviewers,
`gathered data from families who had used Green Thumb as well as from
`a comparison sample of farmers on the volunteer list who had not been
`selected to receive Green Thumb Boxes. Interviews were completed
`with 194 of the 200 Green Thumb families and with 76 families in the
`comparison sample (hereafter referred to as “non-users”). In addition,
`interviews were conducted with 25 “key observers” from agriculture,
`business, education, media, and local government in each county.
`
`The responses from the farmers and other
`key observers were undoubtedly aflected
`by problems of system reliability.
`
`Just before the evaluation period, the Shelby County extension agent
`wrote a letter to Western Union, manufacturer of the microcomputer that
`sat in his office, complaining that the machine’s chronic failures were
`undermining the Green Thumb trial. A Western Union technician
`advised that the humidity in the extension office, located in the base-
`ment of the old Shelby County Court House, was probably causing the
`failures. In fact, the Western Union microcomputer in Todd County wus
`more reliable; it was located in a dry, modern office.
`The main computer in Lexington was inundated by the flood of
`Green Thumb information, which required up to 50 percent of the
`computer’s capacity during the daytime. Because the Lexington comput-
`er was not dedicated to Green Thumb, events conspired against the trial.
`For example, if updates arrived from other computers during scheduled
`or unscheduled interruptions in the main computer’s service, they were
`lost.
`Surprisingly enough, the Green Thumb Boxes themselves were very
`reliable. A few were incapacitated by lightning before surge arrestors
`were installed. Green Thumb users had no complaints about their boxes.
`Users reported accessing Green Thumb an average of 19 times per
`month. According to a monitoring program in the computer, these
`reports were too high; the computer could confirm only about 10 calls
`per month. Some of the discrepancy may be explained by simple
`overestimation and some by unsuccessful efforts to use the system when
`it was not working.
`Use of Green Thumb declined sharply with time. In the first quarter
`of operation, 11,945 calls were made. Only 70 percent as many calls, or
`8,347, were made in the second quarter, followed by 4,512 calls in the
`
`157
`
`

`

`Journal of Communication, Winter 1983
`
`third quarter and 3,595 in the fourth quarter. Reasons for the decline,
`apart from the unreliability of the system, will be discussed below.
`According to the farmers’ reports, large farm operators used Green
`Thumb more often than small farm operators. Forty-six percent of those
`who farmed more than 750 acres, versus 37 percent of those who farmed
`251-750 acres and 21 percent of those who farmed 250 acres or less,
`reported using Green Thumb more than 20 times monthly. However,
`multivariate analyses controlling for type as well as size of farm opera-
`tion “washed out” most of the relationship between farm size and Green
`Thumb use. Large farm operators also tended to have a “marketing
`orientation” and to be more innovative in their farming; these variables
`were related to Green Thumb use independent of farm size.
`Both the farmers’ reports and the monitoring tape confirmed that
`weather and marketing information were the most valued services of
`Green Thumb. Farmers in Shelby County ranked weather information,
`particularly in the form of three- to five-day forecasts and state radar
`maps, as somewhat more important than marketing information. The
`converse was true of farmers in Todd County, who found marketing
`information like “futures” for soybeans, corn, and wheat the most
`important. These responses parallel agricultural differences between the
`two counties, in that Shelby County has a larger variety of weather-
`sensitive farm activities and Todd County has a larger concentration of
`grains and other market crops.
`
`During the period of the trial, Green
`Thumb neoer rose to a dominant position
`among weather and marketing information sources.
`
`Users ranked two conventional sources of weather information-
`radio and television-and
`three sources of marketing information-
`newspapers, radio, and buyers-above Green Thumb. However, Green
`Thumb outranked other sources on each list.
`Green Thumb frames concerned with 4-H/youth activities, home
`economics, and other topics not directly related to farm production and
`management were seldom used, for reasons that will be discussed
`below.
`Notwithstanding users’ narrow interest in the weather, marketing,
`and other farm production and management frames, Green Thumb use
`had both family and social dimensions. Farmers were the principal
`users, but their spouses and children also placed many calls. In some
`cases, teenage children were the “delegated users”; they checked the
`weather and marketing frames on behalf of their parents. Green Thumb
`also stimulated communication in the farmers’ social networks. Users
`reported that they demonstrated their Green Thumb Boxes to an average
`of 16 persons outside their families and that they passed along Green
`Thumb information to an equivalent number of persons.
`
`158
`
`

`

`Computers, Education, and Public Policy I A Videotext Trial
`
`Users ascribed various benefits, including savings of time and mon-
`ey, to their use of Green Thumb. Users said that they would be willing to
`pay an average of $8.75 per month for the same Green Thumb services
`and an average of $17.50 per month for expanded services (including, for
`example, farm business accounting). Non-users, who lacked direct
`experience with Green Thumb, would be willing to pay only 66 percent
`as much for the same services and 54 percent as much for expanded
`services.
`In summary, Green Thumb was technologically faulty in several
`respects. Only extraordinary efforts of extension personnel kept the
`system running as well as it did. However, users were generally satisfied
`with the service. They used it quite often, if not daily, and ascribed
`farming benefits to it. While it did not become a dominant source of
`information during the first year of operation, it took its place among the
`several sources of information reported by each farmer. It became a
`factor in both family and social network communication.
`
`The decision to provide Green Thumb service over
`telephone lines, which was reasonable in 1977,
`entailed other decisions that led to Green
`Thumb’s reliability and utilization problems.
`
`Unlike a broadcast-based teletext service, the telephone-based vi-
`deotext service was almost entirely under the control of the Extension
`Service. In theory, at least, service improvement or expansion that the
`trial might suggest could be implemented unilaterally by the Extension
`Service.
`But, in reality, even if many more Green Thumb Boxes were
`available, the relatively few computer ports available for a telephone-
`based service meant that the number of Green Thumb families would be
`limited to the hundreds rather than the thousands. Furthermore, in order
`to reduce the load on the computer and to free the farmer’s telephone for
`other uses, the downloading protocol was adopted. The problems of
`operating a downloaded, telephone-based service soon became appar-
`ent.
`First, on the “supply” side, the extension specialists responsible for
`entering and updating information could not ignore the numerical
`disparity between the 200 Green Thumb farmers and the thousands of
`farmers who used their services of other kinds. Except for weather and
`marketing, there was no automatic procedure for converting information
`into frames. Time spent designing and entering frames was time lost
`from other activities that were at least as essential to the Extension
`Service. As a consequence, extension specialists could not enter as many
`frames as they wished, nor could they keep the frames updated.
`On the “demand” side, users soon learned that downloading was a
`slow process. Each frame they requested took more than 15 seconds to
`
`159
`
`

`

`Journal of Communication, Winter 1983
`
`load into the Green Thumb Box, and they could not view the first frame
`until all frames were loaded. As the novelty of sampling frames wore off,
`users began to limit their requests to the weather and marketing
`information they most wanted to see. Furthermore, when they requested
`other frames, they often discovered that the frames had not been
`updated for weeks or months. Thus, the system was reshaped by
`circumstance into a special-purpose weather and marketing information
`system.
`The decision to provide local access to microcomputers rather than
`long-line access to the main computer in Lexington was a reasonable
`corollary of the original decision to transmit Green Thumb information
`over telephone lines rather than airwaves. The system designers could
`not have anticipated that the microcomputers would fail as often as they
`did (and one county’s microcomputer did perform better than the other).
`A set of INWATS lines direct to the state computer might have provided
`more reliable service without adding to user costs, if the state computer
`could have been equipped with a sufficient number of ports.
`
`However, the most compelling conclusion of this
`analysis of Green Thumb is that a broadcast-based
`teletext system would have met all of the
`designers’ objectives more simply and reliably.
`
`First, downloaded videotext is no more interactive than teletext.
`Second, the Green Thumb data base was only a fraction of the size of a
`manageable teletext data base. Teletext wait times for a 250-frame data
`base would average only 25 seconds, even without frame prioritization.
`Since the user would be saved the additional delay of placing a
`telephone call for each set of requested frames, the total wait time would
`generally be less for teletext than for downloaded videotext, given a data
`base of this size.
`Third, the Green Thumb counties had adequate VHF signal recep-
`tion from Louisville (Shelby County) and Nashville (Todd County).
`Arrangements with only two VHF stations would have sufficed, not only
`for the trial but for Green Thumb’s expansion to several other counties as
`well. Of course, update feeds to the television stations from the Lexing-
`ton computer would have been required, but this linkage is no more
`complicated than the linkage between the state computer and the county
`microcomputers.
`Fourth, and most important, teletext has the same provision cost
`irrespective of the number of users. If the Green Thumb Boxes had been
`teletext decoders, which, like videotext decoders, are cheap to produce,
`the number of Green Thumb families could have been much larger. And
`with 2,000 or 10,000 Green Thumb families rather than 200, the
`
`160
`
`

`

`Computers, Education, and Public Policy I A Videotext Trial
`
`extension specialists would almost certainly have allocated more time in
`their busy schedules to the preparation of Green Thumb frames.
`Teletext is not a suitable medium for all electronic text transmission.
`In fact, it is a very limited technology in interactivity and in data base
`capacity. The future of videotex around the world probably belongs to
`continuous-connect videotext via cable, both coaxial cable systems (e.g.,
`QUBE) and fiber-optic cable systems (e.g., HI-OVIS). However, when
`relatively small data bases are to be transmitted to geographically
`concentrated audiences without need for a high level of interactivity,
`teletext may be the medium of choice.
`Such rethinking of both “micro” and “macro” system decisions
`should be the goal of videotex trials now underway. The Green Thumb
`trial proved that a dedicated staff could offset the mischief caused by an
`undedicated computer. In the future, users should be treated to the kind
`of service that the same staff could provide when the machines work for
`them rather than against them.
`There are lessons for communication research in this study as well.
`Although sweeping assessments of “the effects of videotex” are infeasi-
`ble, it seems clear that videotex is an information medium that is
`particularly suitable for problem-solving and decision-making tasks with
`specific information requirements. In contrast to the attitudinal effects
`paradigm that has arisen in research on entertainment media like
`television, research on videotex is likely to foster a new paradigm
`focusing on the tasks of everyday life.
`The methodology of communication research will also be affected by
`the unique properties of videotex. No previous medium has been
`capable of monitoring its own use in such staggering detail-by user, by
`frames of information used, by search sequence or strategy, etc. While
`user surveys will still be needed to assess certain social effects of
`videotex use, the monitoring data may lead to new understanding of the
`interplay between information input and cognitive tasks like problem-
`solving and decision-making.
`
`REFERENCES
`1. Bloom, L. R., A. G. Hanson, R. F. Linfield, and D. R. Wortendyke. Videotex Systems and
`Services. Boulder, Colo.: U.S. Department of Commerce, National Telecommunica-
`tions and Information Administration, 1980.
`2. Carey, J. “Videotex: The Past as Prologue.” Journal of Communication 32(2), Spring
`1982, pp. 80-87.
`3. Case, D., M. Chen, H. Daley, J. Kim, N. Mishra, W. Paisley, R. Rice, and E. Rogers.
`Stanford Eoaluation of the Green Thumb Box Experimental Videotext Project.
`Stanford, Cal.: Institute for Communication Research, Stanford University, 1981.
`4. Sigel, E. (Ed.) Videotext. White Plains, N.Y.: Knowledge Industry Publications, 1980.
`5. Warner, P. and F. Clearfield. An Eoaluation of a Computer-Bused Videotext Informa-
`tion Delioery System f o r Farmers: The Green Thumb Project. Lexington, Ky.:
`Department of Sociology, University of Kentucky, 1981.
`
`161
`
`