CS 584, Sample Discussion Questions

Most of the homeworks assigned in this course will be in the form of discussion questions that you'll submit to the cs584 list at least 24 hours before the class that covers the topic.

Contribute questions that are reasonable for the presenters to answer, given suitable thought and perhaps a little digging into other references if they are so inclined.

Do not ask for the presenter's opinion about something or to speculate about design decisions -- UNLESS there is a good chance they can answer using a scientific basis, in which case these can be excellent questions.

Contribute questions that promote thought/ insight/discussion about the topic the paper was chosen for. eg, if the paper is a case study about how to apply particular HCI techniques, "how/when/if to apply" is the ideal territory for questions.

Discussion questions are to be questions that help us understand the research issues of humans doing software development. So, raise these questions from the perspective of a researcher, not from the perspective of a programmer. Here are some examples of good discussion questions that have been raised in the past:

• (About a language called Pursuit): The author states ... While this provides the benefit of both a single language and concreteness, doesn't it potentially decrease the liveness of the language? This is because...

• (About an experiment that was reported): I'm not sure if that was a really fair test. The specially cooked-up "text-based" language really looked pretty nasty to me. What sorts of requirements can we make, from the scientific point of view, to make sure that 2 languages being compared are actually comparable?

• The static representation of (some VPL) forces the user to use and understand wild cards and variable-like abstractions. Is this a reasonable requirement for a language directed at end users? If not, what could replace those 2 devices?

• In section ..., the authors conjecture that the reason the study came out so favorably for the graphical version was because of the particular way control flow was represented in the comic-strip language. I think this is a very good possibility. I wonder how we could project a similar device onto the object-oriented concepts used in Prograph.

• (Starts with an introductory argument of several sentences.) So the question is: Does (some highly-touted feature of a language) actually harm program comprehension?

In general then, your questions should attempt to dig deeply into some issue that arises from the paper we are studying.

Sample grading comments:

Consistent with what I've said above, here are some sample grading comments I've given in the past:

Question 1 and 3 are reasonable questions, but off the point.
Question 2 based on a mistaken understanding.

None of these questions do much to advance discussion
of the role of the paper in this class, a case study of
using HCI techniques to ...
--------
Qu 1/2/3 = B+/B/A+

Insights in isolation are reasonable (but this does not make for
an A, which means excellent).
Insights that bring in substance we've learned are better ideas.
----------
1: reasonable, but off topic.  (Answer is yes, foreseeing a problem
that doesn't exist.)
2: ok at beginning, trivial toward end, regarding _this_
class's focus.
3: not advancing discussion/understanding of this course.
4: reasonable, but off topic.

None of these questions do much to advance discussion
of the role of the paper in this class, a case study of
using HCI techniques to ...
---------
I especially like question 3, which is just the kind
of thing I'm looking for to promote discussion/learning
of the course topic.
----------
Most of
these questions don't do much to advance discussion
of the role of the paper in this class, a case study of
using HCI techniques to ...

1. Doesn't advance the course discussion/topic.
2. why would this be something we'd care about for the
   intended audience?
3. same comment as with 2.
4. good question, this is a good one for the course goals.
----------
These are the kinds of questions I'm looking for!
Very nicely aligned with the kinds of discussion
and learning the class is trying to promote.
----------
Question 1 is reasonable, and question 4 is extremely nice.
Those two are nicely aligned with the course goals,
namely to study _how_ to use HCI in the design of
programming languages/environments.

Question 2 is based on a flaw in understanding and is
not the sort of thing the presenters could profitably
spend time to answer.  Question 3 even more so.
------------
Question 1 is good, 2 is somewhat adequate,
and 3 is somewhat adequate.

In general, you want questions that can promote
learning/discussion of the course theme, namely
_how_ to use HCI techniques to design better
languages/environments for forms of programming.
--------------
Question 1 is an excellent question, and just the
kind of thing I'm looking for, to promote discussion/learning
of the course theme: _how_ to use HCI techniques to
design better languages/environments for humans to
program with.

Question 2 is ok.  The answer is "individual differences",
which one always finds in human data.

Question 3 is ok.
-------------