Informatics 2C - Introduction to Software Engineering (INF2C)

Response to 2015/16 survey feedback


General remarks

As highlighted by the multiple choice answers, there were serious issues with the course organisation. These principally concerned the running of the three courseworks and delays in providing feedback on these, though there were some issues too with the lectures and tutorials. I discuss these issues in detail below.

A large challenge I faced in putting on the autumn 2015 instance of this course was working with the under-resourcing of teaching support. Despite repeated efforts to find suitable staff, the course ended up having to run without a TA assisting with coursework setup and missing almost half the budgeted coursework markers. At the time of writing in autumn 2016, the course unfortunately faces similar under-resourcing. I will be trying my best in the coming days to find further support staff.


Delay of coursework issue from initial announcements

The 3 coursework issue dates were 3, 6 and 4 days later than initially advertised, partly compensated for by setting due dates 1, 4 and 0 days later.

A major reason for the delay was having to devise the case study running through the three courseworks completely myself, without support from a TA. I appreciate the significant disruption that this caused to students' work planning, and every effort will be made in autumn 2016 to stick to the announced issue and due dates.

Length of courseworks

The structure of the courseworks in autumn 2015 was similar to that introduced in autumn 2014, as this gave a more systematic coverage of course topics by all students than with the courseworks in earlier years. In autumn 2014 the case study that ran through the three courseworks was rather limited in size and allowed few opportunities for requirement, design and implementation decisions. For this reason, a larger case study was chosen for autumn 2015. In retrospect, it is apparent that the case study was rather larger than suitable for the nominal 40 hours of work each student is expected to spend on the courseworks.

For autumn 2016 the case study size is being significantly reduced, while still trying to give it more depth than the autumn 2014 coursework. In addition, the amount of repetitive work in the first two courseworks (e.g. preparing use cases and UML diagrams) is being reduced.

Clarity of coursework instructions

Software engineering courseworks, by their very nature, are going to be less well defined than courseworks for most other informatics and maths courses. As ambiguities and missing details arose, I did add clarifications on the discussion forum. This is entirely in keeping with the common practice of software engineering, where an ongoing dialogue with stakeholders is very helpful in quickly sorting out issues that are less than clear.

For the autumn 2016 run, I am reviewing the forum posts to see what additional kinds of information I can fruitfully add in to the coursework instructions.

Clarity of coursework marking schemes

With the first coursework there was an issue where 10% of the marks were for more open-ended criteria that I mentioned on the discussion forum, but not in the instructions or in an email to all students. The class reps raised this issue, and the marking scheme was modified so no student was disadvantaged if they missed the relevant forum postings. In both the second and third courseworks I took extra care to make the full marking schemes clear, with the second coursework by emailing the scheme to all students, and with the third coursework by including the scheme in the coursework instructions.

Coursework feedback was much later than needed

The importance of rapid feedback is appreciated, especially as each coursework built on the last.

In autumn 2015, a number of factors combined to yield slow coursework feedback times (e.g. 14 and 20 days for the first two courseworks) which resulted in students not receiving needed feedback until close to the end time of subsequent courseworks. The factors included me being slow than ideal in distributing coursework submissions to markers, there being fewer markers than needed, markers having other vital time commitments (e.g. presenting their research at conferences), and, in one case, a marker revisiting some of their marking in order to improve their feedback comments.

In autumn 2017, the course organiser will pay extra attention to the marking schedules in order to bring down these feedback times.


Availability of lecture slides before lectures

While the exact versions of lecture slides used in lectures were often not available until just before lectures, the course website pointed students to similar lecture slides from the course run two years ago.

For the first weeks of the autumn 2016 course run, the situation has been similar, with all the lecture slides from autumn 2015 readily available from the semester beginning. For the remaining weeks, I will try to make sure the exact slides used in lecture are on the course website 24 hours beforehand.


Number of tutorials

For several years this 10 credit Informatics 2 course has offered each tutorial group 4 meetings through the semester, compared to at least 8 meetings for the 20 credit Informatics 2 courses. Students asked if there could be more tutorials.

While the tutorials cover only a fraction - maybe 1/3 - of the course topics, they do cover much that has a `problem solving' or `application of knowledge' flavour and that is most suited to going through in a tutorial setting.

For the remaining course topics, the recommended readings, the courseworks themselves, and interactions with demonstrators and on the discussion forum do provide good alternative opportunities to reinforce and explore further the core topics introduced in lectures.

Timing of tutorials

The course was given tutorial slots in Weeks 2,4,6,8 and 10. There was not time for lectures to have covered enough for Week 2 tutorials to make sense, so tutorials started in Week 4. To allow time for tutorial preparation, lectures supporting a tutorial must happen at latest the Tuesday of the week before. As it happened, the supporting lectures for the 4 tutorials finished respectively 5, 7, 19 and 7 days earlier than this latest time. So in particular the delay until the 3rd tutorial was certainly larger than desirable. This delay was unfortunately unavoidable because the corresponding lecture fitted thematically earlier in the course. And the third tutorial could not be swapped with the first or second, because these solidified material needed for the first and second courseworks.

Running Inf2C-SE tutorials in Weeks 3,5,7,9 might be better, if the Inf2C-CS tutorials could be shifted to the even weeks. This will be investigated for the 17-18 academic year: it is too late to make changes now.

Paul Jackson, September 2016