+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+Enabling Supervision
A two way process
+James Brusey
5 March 2024
+Set ground rules at the start
+
+
+Ground rules for you
+-
+
- Meet regularly +
- Public praise, private criticism +
- Rapid turnaround for drafts +
- Be part of the team, not just the boss +
- Allow recording + +
Ground rules for the student
+-
+
- Organise, run meetings, write minutes +
- Criticise supervisor privately first +
- Rapid turnaround for drafts +
- Show up (like a normal job) +
- Take ownership of the PhD +
- Understand and obey the Vancouver protocol +
- Understand and avoid plagiarism +
- Learn the tools +
- Backup your work +
- Yes - you can buy stuff +
- Be part of the team + +
Here is the ground rules document that I use
+ + +How to run a meeting (advice for student)
+-
+
- Send documents in advance +
- Don't expect technical feedback on something just presented in the meeting +
- If you get feedback, act, and show that you acted +
- If there is nothing to discuss, don't take an hour +
- Write minutes on the same day + +
Possible agenda
+-
+
- Review previous minutes, including plan +
- What you actually did +
- Issues, problems +
- Plan for next period + +
Why the supervisor needs an agenda
+ +
+
+
+
+
Project management
+
+
+Use a gantt chart, right?
+ +
+
+
+
+
+#+comment 
+
Wrong!
+-
+
- To produce a nice looking gantt chart, you'll need hours with project management software (or a spreadsheet) +
- Everyone uses different software anyway +
- Either too high-level or too low-level +
- Too hard to update (so it won't be updated!) +
- It doesn't necessarily keep your project on track +
- Research is lumpy
+
-
+
- hard to estimate how long something will take +
- new tasks appear frequently + +
+
+
Alternative – use Scrum
+-
+
- Scrum is designed about a product backlog or list of features +
- A backlog is not about doing things on time but more about rate of progress +
- Supervisor and student agree on a priority order for the backlog +
- Group features into a sprint (these 3 items will be done by the next meeting) + +
Example backlog
+| Feature | +
|---|
| Lit review on thermal comfort | +
| Clean data | +
| Try linear regression | +
| Try Neural network | +
| … | +
Can also add a difficulty estimate (story points)
+| Feature | +Points | +
|---|---|
| Lit review on thermal comfort | +100 | +
| Clean data | +50 | +
| Try linear regression | +20 | +
| Try Neural network | +20 | +
| … | ++ |
Burn chart
+ +
+
+
+
+
Also consider word count trackers
+ + +Match your approach to the student
+
+
+Give flexibility to stronger students
+-
+
- They get to choose the topic more +
- Meet when they have a result +
- Meeting agenda flexes to what they think is most important +
- Attendance not as important as progress + +
Give discipline to weaker ones
+-
+
- 10am latest starting time +
- Always review with them the previous minutes to pick up what they are avoiding +
- Bring them back to the agenda when they go off on a tangent +
- Give them a topic that you know will work and produce a viable thesis +
- Keep in mind:
+
-
+
- you are not punishing for poor performance - just adjusting your style to their needs + +
+
+
Engender self-respect and ownership
+
+
+Good students care about every aspect of the quality of their work
+-
+
- Software should have unit-tests +
- Experiments should be reproducible +
- Outputs, graphs, tables well-proportioned, graceful, aesthetically pleasing, informative +
- Written work should be readable by a lay audience + +
How do we get there?
+-
+
- Have high standards and diligently point out issues +
- Make sure it is not just about the supervisor - enable other feedback too
+
-
+
- journal / conference reviewers +
- wider community (e.g., stackoverflow, open source developers) +
- personal reward systems (e.g., chocolate 😋) + +
+ - Show them great examples +
- Also see Writing for Computer Science (Zobel) + +
Conclusion
+-
+
+
+