Communications of the ACM
Home/Magazine Archive/January 2000 (Vol. 43, No. 1)/Diary as Dialogue in Papermill Process Control/Full Text
Communications of the ACM
Diary as Dialogue in Papermill Process Control
Papermills are gigantic and complex machines, incorporating state-of-the-art technology. As part of a larger project a papermill in Finland was used to provide support for factory floor workers in process industries [1, 5]. Paper diaries were substituted with electronic diaries on one production line employing 35 workers. After a year of voluntary use, 3,500 entries were logged, leading us to believe the e-diary is a valuable tool. Entries constitute dialogues within and between work shifts, and partially with other organizational levels. Using the e-diary is like talking-out-loud or having a radio in the background. It does not call for specific responses, though it can evoke them.
This form of coordination provides an alternative to workflow in situations where efficiency depends on discretion with respect to prioritiesthe ability not to respond when there are other things to do. The usefulness of the e-diary on the factory floor led to its informal diffusion to more than 100 workers, including managers.
Papermills are tremendously expensive, extremely noisy, mind-blowingly complicated, and each one is unique in the way it's structured. Paper and papermills are a cornerstone of the Finnish economy. Let us expand the description.
Size varies, and production lines can be as long as 500 feet, especially if the workspace includes the paper rolls. Paper rolls are up to 30 feet wide and 20 feet high. It takes time to walk around them; other workers are often out of sight.
Company capital is measured in hundreds of millions of dollars. Shift work, divided into 24-hour production, is needed to repay investment. Downtime cost can be $20,000 per hour so there is a major incentive to keep the machine running at all times. It is cost-effective to wait for breakdowns, rather than stop the lines for regular servicing. Days are divided into three shifts. The usual pattern for each worker is three morning shifts, three day shifts, three night shifts, then five days off.
Noise, sheer size, and difficulty in seeing others limit communication between workers on the same shift. Noise is partly a result of the dramatic speed of production. A 16-foot-wide stream of paper runs through the machine at 4,250 feet per minute or about 50 mph. Speeds on the newest machines can be as fast as 8,200 feet per minute. Paper breaks, the main topic of [1], can be correspondingly dramatic. The paper on the machine is a torrent of sludge with decreasing degrees of thickness and wetness. Holes and other defects can start to appear in the paperstream, and fast actions are necessary to prevent a possible loss of control.
Papermills are fundamentally a set of interlinked machines using bearings, rods, connectors, pumps, gaskets, cylinders, drive belts, and so forth, that make up most mechanical devices of this size (see Figure 1). The most important elementthe conveyor belt that carries the paperstreamis termed "the wire." In addition, much of the machine is controlled, monitored, calibrated, and adjusted by a multitude of electronic devices and computers. A paper machine (depending on age and design) may incorporate up to 50 logic and process controls; it may have as many as 10 online video cameras. The images and tapes are used for post mortems on challenging or special situations.
Each machine is configured differently, and has its own history of "updates," for example, more sophisticated control computers and new sensors. Experience is critical for fluent control. Machine experience, such as length of service and intuition are highly valued by workers and managers. The importance of this experience and intuition is greater than, say, oil refining because the papermill production process can be directly seen, heard, smelled, even tasted. Experience and intuition resist a priori categories. Informality and worker-generated "local categories" were designed into the e-diary with this in mind.
Papermill management is essentially hierarchical, as are most management models, with reporting going upward. An important tool of this papermill's management is the "morning meeting," in which challenges and special situations are reviewed, production levels monitored, and new activity and purchases initiated. The original paper diary was an information instrument for these morning meetings; the e-dairy continued this role, and was one reason why it diffused from the factory floor to management levels.
The main production issue is for shift crews to keep the mill running, working around problems as they arise. Often, people or particular information are not available. The objective of the e-diary was to augment organizational memory,1 (OM) especially between shifts.
Transition from a Paper Dairy to the E-Diary
The e-diary design was a cooperative effort between workers in an oil refinery, a superintendent and foreman from the papermill, and the research group. It was piloted in the oil refinery, and a refined prototype was used in the papermill. All 35 production line staff workers received about four hours training on the diary,2 including its underlying Microsoft Windows and Lotus Notes applications. Feedback was positive and was reflected in a quick transition to actual use of the application. Soon after, the paper diary was abandoned.
The main features of the shift foreman's paper diary were:
- Only the shift foreman and superintendent made entries;
- Other workers were limited to viewing it in the shift foreman's office;
- Free-form information about interesting events, plus a lot of factual and numerical data were incorporated; and
- The left page was reserved for memos to others, the right page for diary entries.
Disadvantages of the paper diary were:
- Entries were hard to read because handwriting varied enormously;
- Indexing depended on the person. Often entries were recorded under such headlines as date, shift or hard data such as machine speed. The date appeared in different places. Sometimes only the shift was recorded;
- Searching and combining entries to give different perspectives was almost impossible; and
- Hard data, including many quasi-tables of numbers and times, was unsystematic, and often repeated.
E-dairy production design principles are as follows:
- All workers in any shift can read and make entries from convenient locations (access points were provided in the wet and dry sections, as well as in the control room);
- Emphasize recording of "soft" information, such as interesting events and experiences;
- Direct hard data to more appropriate reports;
- Offer an overview of what is happening and allow different perspectives, such as showing the views of different operators);
- Establish more structure such as specific roles and headlines) than the paper diary;
- Offer a report of the last 24 hours (for morning meetings); and
- Provide full-text search facilities to support the use of "local categories" (see Figure 3a/b).
Actual Use of the E-Diary
In paper diaries, people's names appear often. For example: "Will X please do Y." In the e-dairy this is rare. Unaddressed notes are the norm. Addressing is completed by the respondent, not by the author. This reflects the use of the e-diary as multiperson dialogues, rather than a command or workflow mechanism. The exact person to whom the e-diary entry was addressed is not known to the author. It may be someone on the same shift, the following one, or a distant shift. It may be addressed to a manager. The relevant person is defined if and when they respond to, or take some other action regarding the entry. A fairly typical small dialogue follows, in which the process of addressing is completed by a person's response working the next shift.
- 17.5.97 morning, shift foreman, finishing: Sensor problem in PL72 ... Night shift will make the next trim. Then we will see if the fault occurs again.
- 17.5.97 evening, shift foreman, finishing: Making trim on PL72 went fine.
This completion process led us to reinterpretating of the meaning of entries. They appear to be factual records, but in fact are not.3 Unlike databases, where generality and completeness are important, factual entries and comment are structured by salience. Their meaning comes from being embedded in a context and is not available outside this frame. E-diary entries are part of a dialogue in which the workers do not have a clear idea of who in particular they are addressing 4 (though obviously they know the general viewers when making entries).
This drew our attention to a precedent in the work of Heath and Luff [3, 6]: "Talking-out-loud" and "overhearing." These are phenomenona where workers will, apparently eccentrically, talk to themselves while performing their tasks. They describe their actions much the same as in train control roomswhich train has been diverted; possible consequential problems (another train on the wrong schedule); who the worker is trying to phone. Workers may describe their own actions with screens, keyboards or other local equipment. Speakers may be left on so conversations with others can be overheard. Talking-out-loud is not specifically addressed to others, but Heath and Luff suggest it is designed to be overheard. Colleagues who overhear may take supportive, collaborative action without specifically being asked or, depending on their own pressures, they may not. Either way, they have peripheral awareness of the overall working context because of talking-out-loud. Such informal processes are essential for collaboration and have been found in other ethnographic studies.
E-diary entries are a text version of talking-out-loudwith extended spatial and temporal scope. Entries reflect other features of talking-out-loud, appropriately modified for the text medium. Emotions and humor are used freely, reinventing such email forms as 'CAPS', multiple '!!!!!' and '???' characters, truncated and incomplete sentences, and slang.
We also observed that making a production change without recording it in the e-diary is discouraged. The system is voluntary, but there is social pressure to use it. More likely the pressure arises because e-diary use results in smoother and fewer working breaks, and there is a bonus that depends on production levels.
E-Diary as Overview
Mill workers' shifts involve nine days on and five days off. With the e-diary it is possible for workers returning from days off to "catch up" by viewing its entries. This did not occur with the paper diary and is attributed to the salience of the entries highlighted and to the increased ease of displaying the last few days. The practice of gaining an overview seems confined only to factory floor workers, and is not followed by managers.
Brief dialogues within and between shifts5 constitute the vast majority of entries, and illustrate conversational exchanges. Dialogues within shifts are in part a way of overcoming noise and distance between workers. For example:
- 10.11.96 evening, shift foreman, mass sector: Some yellow rolls have been wrecked.
- Mass operator, mass section: The lock on the yellow ink drum has been removed (it will run until its empty) ...
Roll person: It is rattling along here too
Dialogues between shifts usually have a similar brief form:
- 8.6 night, wet section: The retention pump for X3 was ... blocked. Shift maintenance were busy working in other departments. Repair for morning.
- 9.6 morning, other: Gasket of retention pump for X3 is broken. Ordered new, will be changed tomorrow.
- 10.6 morning, wet section: Gasket of retention pump for X3 was changed and pump works.
Hanging Entries and Extended Dialogues
Hanging entries are those to which there is no response when one might be expected. These form a small minority of the e-diary entries, but have theoretical significance. Three examples:
- 5.5.97 night, wet section: Problems in making PL 72 trim, system displayed a transfer problem, ...sensor is defective, to do in morning shift.
- 1.6.97 morning, finishing: Inner part of PL 72 safety gate falls too quickly. Whipping for night shift ('whipping' translates approximately as 'beat the bushes and see what comes out')
- 12.5.97 night, mass section: ...rising over two (about 2.3%) but no foam in wirewell. Foam blocking solid 58 -> 60 ml/min. Is it necessary to check calibration?
Extended dialogues are also a minority, but, like hanging entries, they indicate important aspects of e-diary use. Extended dialogues are created by workers use of the search facility. This gives a different perspective from the adjacent shift perspective normally used for making entries, or when "catching up." Entries about the same issue are seen together. In the example (see accompanying box), "Reject Carrier" is the search term. This term (many more obscure ones are used by workers) is a "local category"one not anticipated in formal reporting sheets. Extended dialogues are part conversation, part OM. Later comments can be made in the light of earlier onesas in the irony of "(29.6) morning-other" in the example. The most noticeable aspect of extended dialogues is the repetition of the same issue. We have conceptualized this as redundancy, not inefficiency. Repetition does not indicate failure when dealing with a challenge, but rather a continuing opportunity to deal with an irritating issue not critical for production.
Similar extended dialogues occurred over problems with the packing-machine glue pump, brake disintegration, relay, and jumping ram over six months. These diary entries are dialogue: each entry demonstrably responds to others, amplifying, repeating or completing them; they use humor, slang, and agrammatical emphases such as 'CAPS'.
The example shows entries are aimed generally at others as notes, questions, or requests. The key property is that they are not "answered." It is usual to regard such unfulfilled requests as a "bad thing"to regard computer systems that support this level of repetition as unnecessary, inefficient, and time wasted. A down-to-earth manager might say the same. Many workflow systems are designed precisely to avoid such unanswered requests. Requests should be met with promises (or at least with counteroffers or rejections), promises should be monitored for completion within specific time frames. At some levels (legal, contractual, schematics of project management) these views are sensible. However, we believe it is a mistakeand counterproductiveto try to discipline all work in these terms. One size does not fit all. Hanging entries and extended dialogues illustrate one class of work process where effective working depends on the ability not to respond.
We have discussed e-diary dialogues as analogous to talking-out-loud and overhearing in face-to-face work situations demanding tight coordination (similar to plane or train control rooms). Talking-out-loud maintains colleagues' awareness of occurrences otherwise invisible. They can orientate their actions, directly or indirectly, in a wider context. It is the nature of talking-out-loud that actions are unaddressed. Specific responses are not called upon, although they are frequently evoked. E-diary entries generally follow this pattern. There are two differences. First, as noted, entries radically extend the spatio-temporal boundaries of overhearing. Second, they are more laid back. Papermill work can be as pressured and time-critical as train control, but textual overhearing is not used for tight, live coordination. It is used appropriately for things that are troublesome or important but not critical at a particular moment.
The talking-out-loud paradigm provides an alternative to workflow with its implicit assumption that interactions are role-specific and will be channeled through a specific medium. Against this, researchers observed that many real tasks are not role-specific (hence the importance of unaddressed entries), and that work involves fluid transitions between various media (phone, fax, face to face, and computer) as convenient and appropriate. Discussions with papermill workers reveal that many (but by no means all) hanging entries are part of a dialogue that has been resolved by other means. Completions (where there are completions) may be made in conversations, by workers simply seeing that something has been done, in morning meetings and in other ways.
Discretion makes the difference between jobs carried out by people, and tasks that can be automated or subject to routine commitments. Discretion involves at least a sense of priority and intuitions about what can be safely ignored or even forgotten. Complaining outloud via the e-diary about the irritating delays and sporadic malfunctioning of the reject carrier fits generally with priorities of uninterrupted, maximum production. Ignoring, or not making a special effort to respond via the e-diary, fits with the same priorities: there are other more important matters to be getting on with at this particular moment.
To sum, such entries reflect the effectiveness (not the inefficiency or lack of discipline) of the moment-to-moment management of the production process. They are a useful form of coordination, allowing both discretion and completion through other media. The use of the e-diary draws our attention to an application area that has not been thoroughly investigated, although exemplified in Internet discussionsthe role of unaddressed comment in extended, spatio-temporal contexts.
Conclusion
The e-diary, although fully operational, is an experimental system. There were no protocols on entitlement to access. As the e-diary became extensively used and mentioned in conversation, many others asked for access. One aspect of this was that the workers using the e-diary became known as factory experts on the use of Lotus Notes since they were using the e-diary daily. Other groups with more formal systems began to request help. These users included 55 people from other production sections with read-only access (de-inking, cutting, and stores), automation and IT development engineers, and 13 managers. In one sense, this diffusion is a sign of success. On the other hand, it is dangerous since formal and informal diary dialogue is not designed for such a wide audience. So far, problems have not arisen. It appears there are advantages in reaching a wider audience to workers as well as managers.
E-diary use creates dialogues of a special type, best understood in terms of "talking-out-loud" and "overhearing." E-diary entry addressing and completion is provided by the respondent, not the author. The extended dialogues illustrate the way in which a search mechanism enables the use of "local" categories (not previously anticipated) to construct perspectives that are part conversation, part OM. Hanging entriesthose with no responseand extended or repetitive dialogues are analyzed as a merit rather than a defect. Specific responses are not demanded, but may be evoked. This provides an alternative to workflow in situations in which effective working depends on discretion with respect to prioritiesthe ability not to respond when there are other, more pressing tasks to accomplish.
The diffusion of factory floor dialogue to other audiences, such as managers, proved unproblematic in this case. Nevertheless, in general, there are reasons to expect unlimited access to be problematic. Future application developments to support unaddressed dialogue will need to consider issues of boundary management carefully.
References
1. Auramäki, E., Robinson, M. Aaltonen, A.,Kovalainen, Liinamaa, M.A., and Tuuna-Väiskä A.T. Paperwork at 78 k.p.h. In Proceedings of CSCW'96. Boston, 1996.
2. Bannon, L. and Kuutti, K. Shifting perspectives on organizational memory: From storage to active remembering. In Proceedings of the 29th Annual Hawaii International Conference on System Sciences (HICSS-29). IEEE Computer Press, Los Alamitos, 1996:
3. Heath, C. and Luff, P. Collaborative activity and technological design: Task coordination in London underground control rooms. In Proceedings of the Second European Conference on Computer-Supported Cooperative Work, L. Bannon, M. Robinson, and K. Schmidt, Eds. Kluwer Academic Publishers, Amsterdam (1991), pp. 6580.
4. Kim, D. The link between individual and organizational learning. Sloan Management Rev., (Fall 1993).
5. Kovalainen, Robinson, M., Auramäki., M.E. Diaries at work. In Proceedings of the CSCW'98. (Seattle, Nov.16-18), 1998.
6. Luff, P. Computers and interaction: The social organization of human-computer interaction in the workplace. University of Surrey, 1997.
7. Norros, L. Conceptual mastery of work. In XIII European Annual Conference on Human Decision Making and Manual Control. Espoo, Finland (Jun. 1314), VTT Automation & Technical Research Centre of Finland, 1994.
Footnotes
1 For a theoretical discussion of organizational memory see Ackerman (in this issue), and others [1, 2, 4, 5].
2Workers who started later learned how to use the diary from the original group (in the context of quite widespread use of, and familiarity with, Notes applications in the organization).
3Factual records and databases have great variability. Machine operators collect data on specific numerical values. This data could easily be collected automatically, and in any event is never used by anyone. Discussion revealed that the value is in the smell. The two workers visit various different points on the machine to collect the dataand would probably not make these systematic visits without the need to collect the data. While making the visits they see, hear, touch, smell and taste the machine, sensing impending problems that may be invisible to numerical monitoring or virtual surveillance.
4This was contrary to our initial expectationprobably deriving from our familiarity with emailthat production workers would address each other through the diary, probably by name, with notes, details of "challenges" and requests for action. Lack of specific addressing is, of course, commonly found in newsgroups and bulletin boards.
5There were difficulties in translating all e-diary entries from Finnish to English; orthogonal languages with neither word roots nor grammar in common made this translation difficult. Compounding this, entries mix abbreviation with jargon, technical terms, and local slang.
Figures
Figure 1. Diagram of a typical papermill.
Figure 2. A working paper machine.
Figure 3. a. Screenshots showing overview of entries (with alternative view buttons). b. The e-diary.
©2000 ACM 0002-0782/00/0100 $5.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2002 ACM, Inc.
No entries found