1994 SLD Shift Policy Committee Report

Marty Breidenbach and Charlie Baltay
The 1993/94 SLD Shift Policy Committee:
Philip Burrows, Stan Hertzbach, Mary King, Steve Manly, Tom Markiewicz (chair), J.J. Russell, Tracy Usher
Recommendations of the Committee
10 March 1994


This policy has now been superceeded by the 1996 shift policy.

Summary of Charge to the Committee

The committee was asked to formulate a shift policy recommendation and, after review, to implement that policy. Specific points to be addressed included: fairness, training, the role of on-site and off-site collaborators, the role of the Run Coordinator, the technical and sociological interface to SLC, and the relationship of on-line to off-line shifts. We were also asked to address the mechanics of shift assignment and accounting, and to look at possible changes to our publication policy regarding collaborators who do not take shifts.

In this memo we present the proposed policy and the mechanics for implementing it. The last section contains a discussion of the motivation behind the policy.

Proposed Shift Structure:

We will continue to staff all shifts with: As in the 1993 run we will staff the Liaison shift during the day and swing shifts, and provide one person per day for a (typically off-site) off-line shift.

Following suggestions from the SLD community, the Expert and Non- Expert shifts will be offset with respect to each other by 4 hours. The three Expert shifts and two Liaison shifts will be:

          Owl:      00:00 to 08:00
          Day:      08:00 to 16:00
          Swing     16:00 to 24:00.
The three Non-Expert shifts will be:
          Owl:      04:00 to 12:00
          Day:      12:00 to 20:00
          Swing     20:00 to 04:00.
All shifts will be taken in consecutive blocks of 3 or 4 days: The Expert weekday block will begin with the Monday day shift (08:00- 16:00), while the weekend block will end with the Monday owl shift (00:00-08:00).
The Non-Expert weekday block will begin with the Monday day shift (12:00-20:00), while the weekend block will end with the Monday owl shift (04:00 - 12:00).

Collaborators will be assigned to one or more of the shift types (Non- Expert, Expert, Liaison, Run Coordinator, Off-line). In particular, all graduate students will be labeled Experts. It is their responsibility to insure their level of competence is consistent with the role.

In addition, each collaborator will be assigned a weighting factor that will increase or decrease the number of shifts assigned. The default weighting factor is 1.0. All graduate students will be assigned a weight of 2.0 until they are in the process of writing their dissertation. Individuals may ask to be assigned a higher weighting factor if they wish; the advisors of graduate students may request a higher weight for their students if they feel it is desirable. The SLD Final Focus Background Team will be de- weighted to 0.2.

Shifts will be assigned through a scheduling program that uses as its input availability lists provided by members of the collaboration who expect to sign physics papers that include data from the run in question. Details of the procedure for specifying availability will be provided in a future memo.

Some relevant features of the availability lists are as follows:

It is recognized that financial constraints may require an individual to serve his/her shifts for the entire run in a block. The sign-up program will have a field where the user may indicate if they want their shifts packed together as densely as possible or not. Collaborators desiring maximum packing would so indicate. Collaborators not desiring their shifts packed together will leave the field blank. Detailed instructions will be included when the sign-up program is made available.

The program will enforce the block structure and attempt to EQUALIZE per collaborator:

subject to the constraints of the availability list, personal weight factors, and packing factor.

After shifts are assigned, it is the PERSONAL RESPONSIBILITY of the shiftee to trade shifts he/she cannot attend. Off loading assigned shifts is not allowed. A master list will be maintained in the CEH and all changes must be made on it. The SCP record of who actually took the shift will be used when the program calculates the assignments for the next block of shifts.

We are planning for a nine month run, nominally beginning June 1, 1994, and will assign shifts in three blocks of three months each one month before each block. The program will have look ahead and look back features.


The Shift Policy Committee will assign collaborators to the Expert, Non- Expert, Run Coordinator, Liaison, and Off-line shift categories. The committee will assign weighting factors to each person. These assignments will be publicly available and open to amendment by consultation with the Committee. With aid from the appropriate SLD experts, it will provide the scheduling program, user interface, and instructions for availability sign up.

The Shift Czar will be the Head Commissioner (Steve Manly until 8/94). The Shift Czar will operate the program and serve as a contact person. He will NOT help collaborators trade shifts.

The Shift Policy Committee will disband itself when the run begins. Thereafter, the Run Coordinators, under the leadership of the Shift Czar, will take responsibility for the success of the run. This committee suggests that the Shift Czar and Run Coordinators hold fortnightly meetings to review procedures, developments, changes of shift type assignment, etc. We suggest that the SLD member of the SLC steering committee attend these meetings. Our ideal is that the RC group, as the pool of SLC Program Deputies after which RCs are modeled, provide continuity and guidance to respond to changes in the run conditions.

Run Coordinators:

The Run Coordinator pool will be limited to about 20 people; only two weeks of RC duty will be required of each during the nine month run. To make sure that each RC is up-to-date on the current state of the experiment, he/she will be scheduled for one block of shifts in the week preceding his/her RC tour of duty.

The "8:00 AM Daily Report" to SLC aspect of the RC's job should be de- emphasized. Counts of Zs on tape from STATUS are adequate for the daily report with a weekly reporting of hard numbers. We urge that the RC role stress monitoring of data quality, education of less-trained shiftees, and expertise for times of problems.


Turning Non-Experts into Experts and Experts into "True Experts" is very difficult. While deemed to be of limited usefulness, classes for all shiftees will be organized before the run by the shift czar. Hands-on training is the best way to learn. To that end during the first month of running two Experts and no Non-Experts will man the shifts. In addition, out-of-date documentation will be periodically reviewed and updated, preferably in context of commissioner's meeting. It remains the responsibility of each individual, especially new Experts, to ensure that he/she is capable of fulfilling his/her shift role. This includes spending several days prior to the first shift in CEH with an expert learning shift duties.

Reality Check

As a rough guide to the anticipated shift load for the 1994-95 run, consider the following. A nine month run will require 810 shifts of each category. In 1993 SLD had approximately:
     20 Run Coordinators,
     25 Non-RC Experts,
     18 Liaisons, and
     75 Non-Experts
For the reasons given earlier we will limit the RC pool in 1994 to 20. Making graduate students Experts will result in about 15 new people running double shifts. The effective Non-RC Expert pool size should then be about 55. Transferring an additional 10 qualified individuals from the Non-Expert to the Liaison pool augments its size to 28 and leaves a total of 50 Non-Experts. Again, while the numbers below are only estimates, we believe the number of assigned shifts should be sufficiently small so as not to be a burden.

RCs (~20):

Each RC will take two shifts over a nine month (38 week) run. Giving each RC a block of shifts before their tour of duty will eat up 140 of 810 Expert shifts. This will saturate the shift availability of the RC pool. Each will serve two long weeks, two shift blocks, and a couple of training shifts in the first month.

Non-RC Experts (~55 effective):

The remaining 670 Expert shifts (=810-140) will be divided among 55 effective (40 real) Non-RC Experts. Each non-grad student Expert will serve about 670/55=12 shifts over 9 months in 3-4 shift blocks; double that for the 15 graduate students means that they will serve 24 shifts in 7-8 shift blocks. Additionally each of the 40 experts and 20 RCs will need to serve 90/60=~2 "Expert Training" shifts in the first month.

Non-Experts (~50):

There are (810-90)/50=~14 Non-Expert shifts per person. This translates into a very light load of about 4 shift blocks over a nine month period.

Liaison (~28):

There are 540 shifts of this category. If we augment the pool as described, each shiftee would serve 540/28=19 shifts per person or about 6 shift blocks over the entire run.

Timetable for Implementation

-3 months
Publish a list of all collaborators, their expertise categories and personal weighting factor, and instructions on how a collaborator may change his/her availability.
-2 months
Beta test program with currently entered (or dummy) availabilities.
-1 month
Run program. Master List posted on World- Wide-Web, VM and/or VMS; all changes to Master List on paper in CEH.
Run Begins
end of Month 2
Run program for months 4, 5, and 6 (with look- behind for shifts actually taken).
end of Month 4
Run program for months 7, 8, and 9 (with look- behind for shifts actually taken).
Next Run
We may need to overhaul our procedures in the light of the experience of the 1994 run.


We have ahead of us a long run. The combined needs of shift staffing, on going data analyses and publications, the VXD3 upgrade, etc., mean that SLD's limited resources must be used optimally. Although some people feel that the voluntary shift system used for the 1992 and 1993 runs was ideal, many collaborators felt that the system could be made better. There were complaints of unfairness countered by complaints that "no shifts were available"; there were complaints about loss of data, quality of data, and the level of expertise and interest of those who stood shifts.

We feel that insisting that shifts be served in blocks is the best way of addressing the level of competence and data quality issues. The system is constantly changing; it is better to learn it once and use it before it changes again. Previously we requested that people voluntarily sign up in blocks. This usually didn't happen. Instituting a scheduling program that enforces the block structure and eliminates claims of unfairness and unavailability seems to be the best solution. The voluntary system is partially accommodated by allowing "preferred" shift blocks. The concept of shift points was generally recognized to be ineffective in fairly distributing shift loads. In the current plan we will try to equalize the discomfort as much as possible.

Clearly, collaborator cooperation in this scheme switches from a willingness to sign up for shifts to a willingness to enter availability data. As the existing publication policy prevents SLD management from using it to achieve compliance, shift availability will be defaulted to AVAILABLE. Collaborators will be socially ostracized and hassled by the Shift Czar and management if they decide not to participate in the system.

Any scheduling program needs some flexibility in order to function. On the other hand, each collaborator's flexibility will vary according to his/her place of residence, teaching schedule, or personal requirements. These can both be accommodated by requiring each collaborator to make himself/herself available for a minimum number of shifts. After much discussion, we settled on a minimum required availability of 20% for all collaborators, resident at SLAC or not. It is hoped that physicists whose schedules are not highly constrained will make themselves available for a larger percentage of shifts and limit their use of PREFERRED availability.

A poll was taken of SLC EOICs to determine if there is a need to continue Liaison shifts. Their unanimous desire was to keep the status quo. The small size of the pool and the perceived problem of its sociological composition will be addressed by requiring that Liaison shifts be filled in the same manner as the other shifts. Various physicists labeled for the Liaison pool will be cross indexed for the Expert or Non-Expert pools and vice versa to try to address these issues.

Discussions with Richard Dubois resulted in a decision to keep the Off- line shifts organized as they were during the previous run. Physicists will be cross referenced for that pool as well so that the system may assign them. However, no points will be awarded for these shifts; they will in effect become service work.

There was a strong agreement that ALL graduate students need to be more involved with the detector that collects the data used for their dissertations. The decisions to label all graduate students as Experts and to assign them twice the nominal shift load are the steps we have taken in that direction. The Advisory Group felt that even these steps were inadequate; advisors who feel their students will profit from greater participation may apply to have their weights or responsibilities increased.

We feel that the Run Coordinator pool needs to be welded into an effective quality control team that recognizes the changing needs and conditions of the experiment. The current model, where one serves one's tour dominated by the stress of preparing for the SLC morning report, should be changed. A good model is that of the SLC Program Deputies. All RCs need to keep current of the SLD situation; the pool should be small enough and its members expert enough that they form a group that is recognizable, has the respect of shiftees and SLC operations, and can deal with any reasonable questions that might be asked. They should review the SLD status on a fortnightly basis and either institute or eliminate procedures as required.