Notes on Academic Staff Retreat

Present 25 academics. (Not Jayasooriah, Arcot Sowmya, John Plaice, Daniel Woo, Arthur Ramer, Paramesh, Xuemin Lin, Mahbub Hassan, Amir Michail, Boualem Benatalla)

This document consists of four parts. The parts are separated by lines.

1. This introduction
2. Summary of Outcomes. This list represents the consensus of the meeting. It was typed up by me, but seen and agreed to by everybody.
3. List of topics not, or inadequately covered, at the retreat. This is my personal opinion and is provided in order to point out what, IMHO, still needs to be resolved.
4. My notes of the discussions. These were written in real time. They are NOT verbatim transcripts, nor are they summaries of verbatim transcripts. They represent as such, by definition, my personal interpretation of what happened. I have tried to fairly represent the points made, in a somewhat condensed fashion. Some personal bias is, however, unavoidable. I provide these notes in the hope that they may be useful to others. I make no claims as to their accuracy, other that they represent what I understood at the time.

• AKS: number of half-day retreats to work out details of specific issues.

Research Goals

• Need centres of excellence, to be used to exploit/pool existing strengths. Reject BC-ism. Be mindful of cost.
• Will not try to become bean-counters wrt publications. Will continue to use simple system as a base of funding.
• School encourages commercialisation of research (and other ways to impact the world, e.g. open source)

Teaching Profile

• Try streaming (advanced/standard) in 1st year and 2011 based on maths model.
• Consider setting up COMP1002/COMP1041? following on from COMP1001 for servicing.
• Will (have to) allow service teaching to expand.
• Define and demarcate computing territory.
• Reduce number of 4 year majors (to bolster UAIs and reduce sizes of 3rd/4th year classes). Use service teaching to ensure budget health.
• New joint programmes (like Bio-Informatics) are ok as long as they don't require new courses. Needs to work both ways: allow our students to take electives from other units.
• Distance/web-based delivery may be investigated.

Teaching Content and Administration

• Create new positions of "teaching systems managers" (nominal CSOs) for teaching support, starting with 1st year. Target own graduates (help desk/lab managers model). Assign them to special projects during summer as available (give enhancement, web-based delivery etc.) Sub-committee (AT, RB, WHW, AN) to look at job-description, report by 31/7/00.
• Ok to pay tutors extra hours for extra work (e.g. answer e-mailed questions). LiC to manage, WHW to approve. Suggest "class-ss" scheme with daily changing person in charge.
• "Teaching Teams" to look after groups of related courses. Details to be worked out under fearless leadership of WHW. "Team speaker" responsible to WHW on matters of teaching allocations within team realm. Teams to provide (teaching-related) mentoring of new staff.
• Committee consisting of WHW, AT, GW, KAR, PC to propose new load balancing formula by 31/8/00.

General

• Provide admin/secretarial assistance for academics.

Research Management

• Establish Research Management Committee
• Need to establish division of responsibilities between RMC and Research Student Coordinator (RvdM, AN, HEG, JX, JPo, PC) and RMC ToR by 31/8/00.
• RMC to work out rules for ME->PhD upgrades.
• Research Student Coordinator to be "default supervisor" for research students admitted from outside?
• Half day retreat to form research groups for 2001, after Olympics.
• Need to revisit committee structures separately and sign off on research groups. Needs YAR (yet-another-retreat) within month after Olympics.

A School structure discussion was only started:

• We talked about research groups but left it to a future retreat to work out what they are supposed to be.
• We agreed that Teaching Teams were a good idea but left it to WHW to lead their formation.
• We did not talk about committees. No discussion whatsoever on the performance/suitability of the present committees. This was left to another future retreat. I volunteered to initiate the discussion by proposing improvements to the FC constitution, which I see necessary to improve committee participation.
• We did not talk about school meetings. This means that the School Meetings Chair, once the only elected position in the School, will remain, pro tempore, a HoS appointment.
• There was no talk of the role and constitution of HoSAC. (Obviously this can only be resolved once the other issues, in particular groups, have been resolved.)

We did not talk about academic staff profiles, and the future hiring policy of the School. We did not, except for the "teaching systems managers", talk seriously about teaching-only staff. We did not really address the issue of how we will be able to handle a first year which may yet double in size. (Streaming mainly addresses other issues, it will only help marginally with dealing with sheer size.)

Session 1 Tue 9:00.

Arun's intro:

• UNSW aligns itself with Universitas 21. Strive to reach Michigan/UCLA level. Good research _and_ teaching. How do we fit in.
• List of issues for later sessions.

Personal Goals & School Vision

Personal Goals:

• research excellence (GH, SKJ, NF, AH, JX, RvdM, RW, AKS, CAS, HEG, JPo, JAS)
• paper in Nature (AT)
• teaching excellence (TL, RB, WSW)
• get average students to learn more (AT, JAS)
• internet teaching (HEG)
• establish computing as an engineering discipline (KAR)
• better personal links with students (PH)
• find time to do research (MC, TL, RB)
• minimise teaching (CAS)
• less (PC)
• develop/restructure courses to increase relevance (RW)
• technology transfer to industry (RW)
• trying to work organise himself/establish niche (MC, KE, OD)
• slabs of uninterrupted time to do research (AN)
• able to achieve more (JPo)

School Goals:

• UCLA level within 5 years, many int'l confs/ws' (NF)
• more collaboration among staff (JX)
• change student/teaching attitudes to reduce spoon feeding (KE)
• "research school" with research-only staff, opportunity via research commercialisation (RvdM)
• teaching infrastructure (AM) low maintenance teaching (JAS)
• "the place which produces exciting things" (PC)
• more industry collaboration (PH)
• commitment to distance education (PH)
• more support (KAR, AN)
• more planning (teaching alloc) (AN) teaching stability (HEG)
• do more for the really good students (JAS)
• more modern admin (JAS)

What Kind of School?

• more informal contacts with other academics, graduates positive about CS&E experience (TL, RB)
• place where you have interesting conversations with interesting people (RB) "scholarship" (WSM), collegiality (OD)
• maintain (social) life (KE, PC, PH)
• 72h weeks an option, not necessity (WSW)
• people should enjoy teaching, not treat it as a burden (HEG)

Tidbits:

• Achim: likes to increase student quality
• Kai: interested in concurrency
• CAS: "teaching is the price you have to pay to do research, ambition is to get out of it"

Session 2: Research Goals (Tue 11:55)

Industrial Research Collaboration

• PC:
  • research mgmt paper going through UNSW, contains lots of bonkers on increasing PhD student numbers
  • need to position ourselves by creating exciting output, aimed less at academia and more at industry

Quality and Quantity of Research Students

• GH job prospects are no longer the issue, its the fact that they make so much more money in industry
• NF, AKS et al: SPIRTs for funding
• GH: good PhD students are produced by good teaching. Get students into research projects early and they'll stay.
• AKS: white paper will help to tap into international market for students. Changed funding mix makes it attractive to create scholarships for to cover fees.
• RvdM: mobility is important, US unis expect people to move after U/G degree. Ensures influx of new ideas.
• NF: target top students from 1st year, give them projects in place of some courses. Address mobility via temporary placements in other institutions. E.g. Toronto has offered exchanges.
• AT: nationally advertise scholarships for 4th year students.
• MC: bait them by getting them to write inter'l papers and sending them to conferences.
• JPo: MRI got good students by offering higher scholarships
• AH: Much of the cool work is commercialisable but not publishable. Should accept commercial outputs as well as publications for PhDs.
• RvdM/AN: maintain database of industry contacts to make better use of contacts already established by some people in the School (unknown to others)

Centres of Excellence (externally funded research centres)

• AKS: could establish centres to provide environment with research-only staff and funding fro PhD students
• research centres take a lot of work to set up and run, questionable benefit.
• CAS: best way to do it is do the work and establish the reputation
• PC: present UNSW thinking is to show ARC cold shoulder on centres
• PC: really against centres, but Chem Eng created centre pulling in lots of people with immense level of industry collaboration
• HEG: School must commit to provide support for people to do research. Need to support those who have potential rather than to those who already have it.
• NF: Should use centres to exploit/pool existing strengths. Focus on areas where we are capable to compete at international level.

Quality of Research Publications

• White paper simply counts papers, without respect for quality
• CAS/NF: need quality standards, otherwise encourage recycling
• citations don't work well in some areas (lot of relevant work not indexed) and tend to be more of an indication of the activity of the field, evaluate citors as well as citees
• KAR: publishing just for the sake of numbers publishing is bad, morally.
• RvdM: can for an area draw up ranking of publications, and in order to establish excellence need to aim for top ones.
• JPo: Works within area, but very dubious for using this to compare across disciplines. Not usable for performance comparisons
• GH: In some areas (like OS) you'll find that people will readily agree about what top outlets are, but if you look at them will find that they are extremely insular, dominated by a small number of people.

AKS: Need some way to establish excellence

Session 3, Tue 16:30

School Teaching Profile

service teaching

• PC: will have to do ST in some areas, because otherwise others will do it, do it badly, and develop offerings which muddle the waters
• AKS: do we need separate CS intro for non-majors? Do we get back to 1081/1082?
• AN: need staff
• GH: staffing is secondary. Need to decide what to do with non-majors, whether merge them in with our students or give them different courses. Problem is how to control who's doing our courses.
• KAR: need to decide what part of computing real estate we own. Then need to decide what courses to offer as service, do we set up separate ones or not.
• AH: should limit service teaching and leave it to others to pick it up. RvdM: same issue at e.g. Cornell, almost everybody needs
• computing. Don't want to hire extra people to do service teaching, as they would dilute their teaching profile. Went to set up Faculty of Computing, with other departments responsible for service teaching.
• GH: we can only keep monopoly on "computing" if we are prepared to do all the service teaching (expect majority of UNSW students to do a computing course). Will only be possible by a drastically different approach to first year teaching. Need to discuss model of creating first year teaching unit, staffed (partially?) with teaching-only staff.
• NF: several US unis done something along these lines. Let (almost) everyone in, let the good ones proceed into our programs.
• CAS: To do this need separate streams for non-majors.
• NF: US unis use teaching assistants to teach first years. AKS: won't work here, and is leading to problems in US.
• PC: use 1st year teaching as a money maker, allowing us to reduce number of majors, AND at the same time pick out top students from other programs.
• WSM: don't have to be scared of classes of 1000s of students, has done it (using 100s of tutors in distance teaching setting). Can be done, but requires effort to be put into developing good teaching materials. Let's look at adopting Maths' model as a start.
• KAR: Cannot see the need for every UNSW student to do programming course. If get serious about 1st year service teaching need to look at what the market really needs.
• WHW: Maths also has service teaching in second year. Maybe need CS-Lite in higher years.
• NF: Sydney Maths has 4 levels of maths, least is "maths for everybody".
• CAS: cannot take analogy with maths too far. They don't need large labs. WSM: only need modem.
• AH: Should ask other schools what they need. (Others: it isn't the other Schools, its the students themselves.)
• PC: Can't do it ourselves. Need to work with IS, and need to have them find out what their role is. Could be CS-Lite.
• AKS: IS can't deliver, they can't even satisfy AGSM students.
• AM: lots of COMP1001 students wanted more. Case for more CS-Lite.
• Several people making cases for others (Bio-Informatics, lawyers eg) needing far more than "scripting". Argument against CS-Lite.
• MC: domain-specific programmers. Do real CS stuff, but in a setting CS people won't have the background to understand.
• SKJ: Why not let IS do the CS-Lite stuff.
• NF: Summarising: Should consider service teaching for everybody, but not in isolation, need to do it with IS. Orthogonal to implementation (teaching-only staff or not). Also need to address issue of specialised joint degree, is a simpler issue because they can slot into existing offerings.

AKS (summarising): Multiple streams for COMP1011&COMP1021 (expecting that most of ours would take advanced stream). In Maths streams have different number but same name and contents.

Want to reduce numbers of majors, can use service teaching to bolster budget.

Session 4 (Tue 15:50)

Teaching Content and Administration

• AKS: Provide admin/secretarial assistance for academics
  

Increase teaching admin support

• AT: Can't run 1st year with part timers. KAR: strong support
• CAS: Need to consider alternative delivery mode
• PC: introduce pool of programmers to write auto-marking scripts
• GH: biggest overhead of large classes comes from large number of students asking questions by e-mail (from lecturer and tutors)
• PC: hard to get good F/T course administrators. Need career path.
• AT: need to split roles of administration and scripting - organising tutors, dealing with extensions, maintaining web pages, maintaining SMS, course FAQ...
• PC: take lab managers/help desk model, get our own graduates
• lot of supporting noise, discussion on how-to...
• AKS: need technical component, CS background
• AM: weekly tutors meeting is very valuable
• Generally accepted policy of paying tutors extra hours for e-mail answers.

Session 5 (Wed 9:00)

Teaching quality control

• HEG: attending each others classes to give feedback should be encouraged
• RB: two types of QC: summative (assesses satisfaction, used for promotions etc) and formative (aimed at improvement)
• Consensus that we should talk about formative
• (several) official surveys are mostly summative
• PC: can use official forms for formative WSM: only very sketchy, self-selected group
• AT allocate a slight component of the mark for filling in surveys, gets almost 100% return rate ("1% trick")
• lots of anecdotal stuff
• PC: Civil's HoS examines summative reports at end of session, congratulates good ones, advises poor performers. WSM: different story, their courses are very stable
• (several) need to assess various components of what makes up teaching: course development, lecturing, assessment, exams, fairness, ...
• AT: free form comments in JAS' survey system gives excellent feedback on what's considered good and bad.
• SKJ: what can be done when teaching surveys come out poor?
• AKS: how can we ensure that people take feedback seriously and seek help if necessary?
• MC: informal mentoring by colleagues may be the best (non-scary) way to improve
• GH: course surveys should be looked at by TC. CAS: they are bad surveys and shouldn't be used. AT: they are carefully designed to distinguished between components. PC: promotion committees look at things in context.
• AKS: Should TC use course surveys as initial indications of potential problems, to be followed up?
• AM: Should be able to talk about problems in non-threatening way.
• AN: Calls for vote whether surveys should be used by TC. Majority (but not overwhelming) is in favour
• KAR: TC isn't the right forum, too many vested interests and conflict. PC: HoS would be better
• AKS: Need some mechanism to improve teaching quality. WHW: HoS and A/HoS looked at course surveys in S2/99. Three courses stuck out as having problems, in two cases there were reasons to believe they were one off, in one case they talked to LiC.

Teaching clusters

• AKS: Informal groups (like database) work out their course contents. Could use this model in other areas, and could form orthogonal structures to Research Groups.
• KAR: must maintain connections between courses. Presently have no structures to deal with aspects like allocation, course coordination within programmes, etc.
• AN: TCs as alternative clusters generate additional overhead (meetings etc)
• JAS: no big deal, get together in week 10 and work things out. It's easy in this case because the courses are essentially in place.
• NF/PC: sort-of-voluntary membership: if you want to teach in an area you'll have to join cluster.
• AT: Agree with KAR model, but shouldn't formalise. Lots of overlap, and frequent change.
• HEG: Need continuity, and clusters may be way to achieve this. Speaks in favour of institutionalisation.
• WSM: Already have de-facto clusters. But cannot be committee-ised because it's inherently fluctuating. Will fall over if try to formalise.
• KAR: talking of two different models: formal and informal. Formal ones would have to be needs based.
• GH: KAR's "formal" clusters seem to relate to programs, "informal" ones to small sets of courses. Who helps people whose teaching areas puts them into a cluster of size one?
• JPo: doesn't like the matrix-management approach. But talks about not liking informal structures. [I guess I'm confused about what he said.]
• PH: Clusters would provide resources and quality assurance mechanism.
• CAS: want to get rid of all formal structures.
• JPo: at present LiC can do with a course what they want. Structures could prevent this.
• NF: Can survive with informal structures if they are strong and stable enough.
• WHW asks PC about abolishing all structures: PC: structures always have danger of becoming territorial, but are to a degree needed.
• AN: Biggest problem of school is lack of structures.
• CAS: need to differentiate between structures and procedures. Have enough structures, need more procedures.
• AT: not much formalisation needed to get what we want.
• AKS: Want research groups focussed on research, teaching clusters on teaching.

Session 6 (Wed 11:00)

More on Teaching

Teaching Clusters (cont'd)

• AKS: Suggests clusters for maintaining group of courses, doing initial cut at teaching allocations and provide mentoring
• PC: Needs a defined person to be responsible to A/HoS for suggesting teaching allocations. Otherwise formal groupings.
• JPo: Need to ensure that all courses fall into a cluster, hard to do with totally informal groupings.
• JPo: Problem is overall load balancing
• GH: Have given up even pretense of balancing teaching load.
• AT: Danger that no-one will look after medium-term averaging of load.

Load Balancing

• WHW: GW's load balancing scheme fell into disuse due to the difficulty of getting information (and justification) out of academics
• AKS: Will never get fully balanced load, but can identify extreme cases and try to rectify. EE&T have very sophisticated scheme counting everything, Mechanical completely ignores research and supervision. Need scheme which includes supervision, class sizes, administration. Should get group of 3 or so respected people to draw up formula.
• WSM: Should start with GW's formula. Has been used successfully.
• GH/AT: discount research output, count supervision.
• PC: should at least have some component for papers
• JAS: GW's formula had the problem that you could never cash in your credits.
• AKS: Nominate members of load allocation formula task force WHW, AT, GW, KAR

Research Management

Research Management Committee

• needed

Research Student Progression

• White paper provides big completion bonus, decays after 3.5 years
• NF: 3.5 years is politicians' invention, have no clue what it takes. Let them do a ME first. Use to effectively extend time.
• GH: will be difficult to sell to students. CAS: ME theses need to be externally assessed - a pain. PC: ME cannot really be used as the first step of a PhD, coursework master's maybe
• WSM: Industry salaries will make it increasingly difficult to get people to complete.
• AKS: Need to distinguish between failure and getting out due to successful commercialisation
• PC: Should AN keep data on progression track record of supervisors, monitor personality types to help students pick matching supervisor?
• WHW/AKS: maybe allocate real supervisor after first year only. GH: don't want middle man for students I know well. AKS: only for externals.
• GH/CAS: need mechanism to weed out unsuitable students, particular problem for people from O/S. Can combine with interim supervisor idea?
• PC: change of supervisor is generally difficult. Maybe have AN as initial supervisor for all PhD students will help to find real supervisor.
• discussion on initial Masters' enrolment. GH/AKS: difficult to sell to students
• HEG/NF: initial ME enrolment, change retroactively to PhD if successful
• AN: official upgrade rule is very strict. Requires international paper. GH: unreasonable.
• left for RMC to sort out policy

Research Groups

• CAS: RGs should be scrapped. Confuse several issues. PhD progression is a lot of admin, should be done centrally. Don't do mentoring, which should be done. RGs are mostly administrative, should be about actually doing research. Propose nominated "supervisors" to mentor. Can decentralise funding.
• GH: agreed that they are too big/inhomogeneous to foster research. Do a lot of teaching administration. Mentoring doesn't take place mostly because we have never developed a policy about it. Real research groups should be smaller and may be more informal. Best mentors would be people doing related teaching and research. Must prevent setting up structures that are used for empire building.
• KAR: sort of agreed that these groups weren't just about research, mostly administrative units.
• PC: Groups (ex depts) were introduced as admin units, and it's natural to use to distribute funding.
• AN: Need more than informal ad-hoc grouping to support research. Centralising funding becomes unwieldy. There has to be a very clear hierarchy.
• AKS: Prefer to have small, focussed research groups, don't want to have to decide at central level whether to fund a student's conference trip.
• WHW: Need someone to sign travel forms and approve absences during exams.
• CAS: Wants to ensure that School stays whole, not a collection of groups.
• AKS: "Informal groups with a few formal responsibility"
• AT: Geographics is about to change, segment building into floors.
• JX: Like informal groups, used as platforms to provide visibility, recruit students, go fro industry funding etc.

Session 7 (Wed 13:30)

Occupational Health and Safety

• AKS: we are required by law to set up OHS Committee. People need to be cooperative.
• GH: OOS has been a significant problem in the past and need to take preventive measures.
• PC: Had several cases of students with mental illnesses.