Last changed
25 Apr 2003 ............... Length about 9,000 words (57,000 bytes).
(This document started on 17 Apr 2003.)
This is a WWW document maintained by Steve Draper, installed at http://www.psy.gla.ac.uk/~steve/chi03/report.html.
You may copy it. How to refer to it.
Web site logical path:
[www.psy.gla.ac.uk]
[~steve]
[chi03]
[this page]
Report on the CHI 03 Doctoral Consortium
By
Steve Draper,
Department of Psychology,
University of Glasgow.
This is a complete second draft of the report, and no more feedback has come
in after April 2003.
This is a report on running the Doctoral Consortium at CHI 2003
(held April 2003 in Fort Lauderdale, Florida, USA).
The panel was:
Clarisse de Souza
PUC-Rio (Pontificia Universidade Catolica), Rio de Janeiro, Brazil
Steve Draper (chair)
University of Glasgow, UK
Anthony Hornof
University of Oregon, USA
Lisa Neal
EDS, Boston, USA
This report is written most directly for next year's CHI doctoral consortium chairs, Jean
Vanderdonckt and Liz Churchill. However it may be of some interest to other
years and other conferences. It also has some points about student posters as
well as the doctoral consortium.
This report, while to some extent informed by feedback and experience, is only
my opinion, not policy nor evidence-supported fact.
From the viewpoint of someone running part of a conference, the worst
situation would be to have compulsory constraints and instructions without
recorded reasons, whereas the ideal would be to have:
- No dictated rules or policy
- A record of what past office holders did with details down to the levels of
model emails and announcements ready for re-use. This would allow the lazy to
re-execute with little thought something that worked once, and equally allow
the diligent to start their improvements from something that worked, without
exhausting their energy on re-invention.
- A design rationale of those actions. This is important for recording the
point of each action (you can't execute them intelligently without that), to
spot when changing context requires changed actions, and to construct
justifications for alternative and improved actions.
CHI is nearer the worst position, although better than many conferences.
There is:
- A policy document that is supposed to be binding, is out of date, is hard
to get hold of, and includes very little rationale.
- But also a substantial effort at an oral tradition handing down pointers
from one office holder to the next. No-one has constructed a list of past doctoral consortium
organisers, though it might not be hard to do for someone with a complete
shelf-full of past CHI proceedings. I found Gilbert Cockton very helpful, and
on the basis of word of mouth I now think that Gary and Judy Olson would be two
other especially good sources of help.
In this report based on my experiences as (sole) Doctoral Consortium chair at
CHI 2003, I will try to move the organisational knowledge on this a bit further
up the spectrum. However I won't provide much detailed actions and model
emails and text, as I don't think I did those all that well. The biggest
source of my own practice at the larger scale decisions was Gilbert Cockton,
who was both conference co-chair and had been a doctoral consortium chair at CHI recently,
constrained by the policy document; while at the shorter scales (how to do the
minute by minute face to face stuff) I had some previous experience in doctoral consortiums at
other conferences.
I've done doctoral consortiums at other conferences, and I wasn't particularly
keen on the CHI format for doctoral consortiums when I took the job, but I now
think this is a particularly good recipe and other conferences should consider
adopting it. This report is in part about expressing reasons for this recipe
which were not apparent to me in advance.
The recipe, in summary, now is:
- 15 students over two full working days.
- Four members of the faculty panel.
- Closed workshop, not conference format: with all 19 expected to be
participating for the whole time including meals.
- Additional invited "notables" at the meals, to allow students to meet
people in the field.
- Funding (from an NSF grant) that provides more or less complete funding
for each student to attend both the doctoral consortium and the whole
conference.
- Students all also present a poster, integrating them with the wider
conference.
- Students' extended abstracts are published in the conference proceedings.
There was no big evaluation of the doctoral consortium, although I did collect comments by
email from the student participants, collected here.
However I now believe for these reasons that it was very successful:
- Several people, not in the doctoral consortium, told me they had talked to a participant
who had told them they were very pleased with it.
- One or two people who sat in on part of it said what a good atmosphere it
had.
- By the end, at least, students were contributing comments on each others'
talks, and were keen not to wait for the panel members' contributions.
- Afterwards, and throughout the rest of the conference, the participants
looked happy and relaxed: more so than many other conference delegates.
- The participants told me it was very useful to
them.
- They still wanted to talk to and spend time with each other and with
panel members long after the end of the doctoral consortium itself.
- However, see the feedback
for some criticisms, and support for some of the points made in this report.
A big issue in organising anything is the balance between pre-planning
and spontaneity (or "just in time decisions"!). I skimped on timely advanced
organisation, mainly due to character defects. On the other hand it worked,
and it is possible to reduce flexibility by doing too much advance planning,
and to make it all feel burdensome by telling people your plans in advance when
they don't really need to know until the moment.
In writing out reasons in this report, I am trying to provide future organisers
with material for thinking about and planning doctoral consortiums. It doesn't follow that they
should adopt all my actions, nor that they should make detailed plans in
advance. Better understanding of the issues allows good on the spot decisions
just as much as better advanced planning.
This is a summary: for some rationale, see the main report below.
- Good quality discussion of each student's work in turn.
Hence: the time per student devoted to this is an important metric (about 30
minutes at CHI 03), you may need to rein in the panel if their comments show
more quantity than quality, may need to further encourage student comments but
only if they are good quality, consider inviting visitors to join in here and
there, get everyone to read the papers in advance, keep the talks short.
- Students getting to know each other well.
This is the foundation of a peer group of new HCI researchers,
introduces them to the great breadth of topics in HCI, and lays the foundation
for good discussion. Organise socialising together, support student comments
on each other's work.
- Students having prolonged access to the panel faculty.
Not just their formal comments in the sessions, but chatting at meals.
- Students making contacts with "notable" researchers.
Programme of inviting them to meals with the doctoral consortium,
encourage students to approach them elsewhere, perhaps use your own contacts
to prompt their visiting the student posters.
- Students present a poster as well.
Good for the students, good for promoting further contact-making, good
for integrating the doctoral consortium with the rest of the conference.
- Talks/discussions within the doctoral consortium on other aspects of being
a doctoral student.
- Subsidising students by getting funding for them.
- Making it a truly international group.
Promoted by subsidy, by mentoring, and by careful selection of the panel.
- Giving them practice (and feedback) on giving talks.
Talking to 20 strangers is obviously better practice than doing it in their
home department, but not as scary as to a big conference audience.
- Giving them practice at contributing to discussion.
I will deal with points here in order of their importance to the overall
success of the doctoral consortium (not in chronological order of the action
required, nor in some logical structure about the type of issue).
Devoting substantial time to discussing each student's research.
The CHI formula of 15 students over two working days allows 45 minute slots
for each student, of which only 10 minutes need be given to their talk (not
least, because their documents can be precirculated and read by all in
advance). This is strongly supported by student feedback as a key feature.
One student said that this was the only time they had had such prolonged
attention to their work apart from one on one sessions with their advisor, and
how useful it was to get other perspectives. Another student said they had
been in the doctoral consortium at CSCW, which had the same number of students
in half the time, and how much better this format was. Note how sensitive this
is to the overall time allocated: if the student talks take 15 minutes, then a
extending each slot from 30 to 45 minutes doubles the time on this top
priority activity of discussion for the cost of only 50% increase in time
overall. The policy document might be tightened in this respect to stress that
the full two days for 15 students is important.
The focus on discussion is because this is what you need the face to face
situation for, while learning about their work can largely be done by reading
in advance.
In addition, a significant benefit is (as one of the students remarked in
feedback) hearing the comments made to other students. The time other than
when their own work is being discussed is not dead time, nor useful only if
they are making comments themselves: significant "incidental" learning goes on
when they are only acting as bystanders.
The students get to know each other well.
This is important for several reasons: (a) founding contacts in the field
that may remain important to them for years; (b) hence experiencing the great
range of kinds of research done in HCI; (c) making them feel very relaxed
during the consortium and for the rest of the conference as part of a group
they feel at home with. (One of them actually said this, but it was obvious in
their relaxed demeanour after the doctoral consortium, and in the way they
continued to seek each other out for evening socialising, sit next to each
other in talks, etc.) This is an important part of a good conference
experience, and is particularly missed by most first time attenders.
(d) As a consequence, feeling able to make contributions to discussion, and to
receive comments without getting too anxious.
While no special action needs to be taken for this, it is actually
probably one of the most important benefits for the participating students.
It is also experienced by student volunteers, but possibly something more
might be done for student poster presenters. For the doctoral consortium, it
will usually happen automatically given two days in a nearly closed session,
with most meals held together: another reason for these aspects of the
standard formula.
The third of the big three aspects is inviting other researchers to meet
the students at meals during the doctoral consortium. There is no doubt this
is a big hit with the students: at one end one student this time went round
saying it had changed her life (after lunch with the person's whose published
framework she was trying to use in her PhD), and at the other a student saying
he wished there had been more such guests so he could have met someone more
directly relevant to his own work.
It is also rather easy to arrange, since everyone I invited, whether they
could come or not, said they were delighted to be asked. Furthermore, those
who came all seemed to enjoy it: after all, it is a free meal with company
that is non-threatening but very pleased to meet them. Given that, you should
consider asking anyone: keynote speakers, the very famous, the more modest but
whom you actually admire the most. A good tip is to ask, where possible,
couples (whose plans are obviously often linked). If you organised this well
in advance (which I did not) you might be able to persuade some to travel
earlier than they otherwise would have. If not, you will be largely limited
to those on the conference committee plus those giving tutorials or workshops:
but clearly this is actually a very rich pool for these purposes. Finally the
people running the following year's doctoral consortium should probably be
invited for one meal, and will probably want to sit in for part of the
doctoral consortium to see how it feels.
Given you have arranged the guests, don't forget that the students probably
don't know who they are, however famous. So brief the students on them before
the meal, and explicitly pick out the students who would get the most out of
them and tell them to sit next to the guests. You can do this by a) reading
the references cited by each student; b) looking at the short summary of each
student's research in your notes. Perhaps I should additionally have got
people to change places after each course. Remember the bad thing about
conference meals is that you can only talk to the 2 or 3 people next to you
and may have nothing to say to them; and the next worst thing is that old
hands all stick together and never meet new people. Students are obviously
the most vulnerable to these traps. A little work on active introductions can
turn it into the best conference experiences for the doctoral consortium
students: meeting someone new who works on something very similar to you.
In 2003, worries about budget meant the conference provided a cheap lunch in
our room, and booked the dinner at the main hotel. This was extremely
convenient for me: less organisation and still more importantly, no time and
trouble spent in travelling to a restaurant.
In the feedback, one of the students suggests taking the "guests" idea
further by reserving an hour (say) for having multiple guests in, and dividing
up the students in much smaller groups between them.
As the conference wore on, I found myself extending this principle. I went
over the students' work again and thought of people at the conference who might
be interested in their work, and did some work suggesting they visited that
student's poster at the designated poster times. When I got this right, this
was obviously great for the student, but also in fact welcomed by the contacts
too, few of whom think systematically about checking out student posters, but
who are in fact interested in new work in their area.
With hindsight I would recommend thinking this out and doing some of it in
advance by email and by a bunch of notes on the personal messages board:
giving them advance notice raises the chance of them doing it. The absence of
a delegate list at CHI is an obstruction to this: lobby for a delegate list at
least for you. A list of paper presenters would do almost as well for this
purpose.
Ask the rest of the panel faculty to help with this. You could suggest this
long before the conference to alert them. Then near, but still before, the
conference, pass them what lists of attenders you have been able to get, and
ask them to consider using their contacts a little.
Give the students a 5 minute pep talk during the doctoral consortium on
seeking out contacts in their area of research. I did this and pointed out:
- You actually already are now members of the community: you've done
research, and have published (your abstracts are in the proceedings). Other
researchers are happy, indeed quite interested, to talk to you.
- If you open the conversation by saying you've read their work, you
will have their attention.
- If you open the conversation by saying you've cited their work, you
will have their attention big time. Researchers are quite interested in sex
and money, more interested in getting research grants, but very interested in
being cited.
- If your work is actually applying some theory they published, you will
probably have their attention for a long time. Nothing is more flattering.
One of the doctoral consortium students immediately went off and applied this,
and came back to report success.
Another consequence of being more active in connecting students to other
researchers and contacts outside the consortium is that you will find yourself
needing a list of short summaries of their research areas, to help you carry
out planning introductions etc. I recommend, therefore (I wish I'd thought of
this in advance), making them submit a 30 word (2-3 line) summary of
their topic and what their contribution to it is. You will find
yourself using this repeatedly. At other conferences this has been tried
anyway for all submissions as a help to assigning referees, as a handy summary
for all attenders in selecting talks to go to, and as a way of making the
author think about and be explicit about what value their paper has as a
contribution to research. Here you will want to use it a) for assigning
reviewers from the panel, b) in choosing an order for the talks, c) perhaps in
considering what guests to invite, d) in considering approaching contacts to
come see their posters. Of course you may well edit your copy of these
summaries, but getting authors to provide the first version is still helpful.
I recommend that you lobby the conference chairs to make all conference
applicants do it for all categories. If that doesn't get accepted require it
anyway in the call for doctoral consortium submissions. If you don't do that,
do it by email for those you accept. (I believe William Newman got it done
for one CHI (2001?). It was done after submission but for the conference
programme for HCI 2002, and I think Janet Finlay has agreed to do it for HCI
2004.)
Additionally (just for consortium submissions) you might suggest to them that
if their work was particularly inspired by one paper, then they say
what the paper that inspired them was
as an addendum to the 30 words. This may apply to, say, a third of
submissions: for instance a thesis trying to apply Dey's framework for
context, or Julian Orr's paper on "war stories". It's important to prompt
comments on whether the thesis work does or does not seem to capture the
spirit of the paper; but also because research may be inspired by something,
yet the actual work turns out to lead to somewhere quite different, which is
important but painful to recognise.
Function of the panel of faculty.
The most important thing is to be able to cover all the research areas that
the selected students have. In my estimate, I can find something of some kind
to say about any talk, but something worthwhile about only 75% of them. (I
estimated that in advance, and in fact found that to be true on the day as
well.) So two panelists can probably cover it, but four is much more
comfortable and certain. In 2003, the student topics ranged from low-level
input device design up to highly social issues. (They might have ranged from
actually constructing new physical input devices, out to pure
ethnomethodological approaches.) I was a bit worried in case there were
software engineering methods, or design in the art school sense to cover: but
not this time it turned out.
So topic coverage is the most important issue. Complete coverage is
impossible, but a good range is feasible. In fact even if no panel member had
any deep comments, it would still probably be satisfactory for the student,
since simply dealing with a new audience and explaining why their work is
important to the general HCI audience, is valuable. If one panellist has deep
comments that is good. If two do, that is very good. In most cases this
year, all four panellists had something to say, but you should probably point
out that there is no need, and perhaps it isn't even desirable, for all
panellists to say something to all students.
Selecting the panel.
Since only a day or two's expenses are paid, they pretty much have to be going
to that CHI anyway. Ideally, the panel of four should be balanced in all
these respects:
- Topic area
- University and private sector
- Gender
- Age: some very experienced, some recent PhDs. The value of the recent
PhDs is they have a much more immediate memory of the process and problems,
and are much more directly useful as role models. I hope there are also good
reasons for some more experienced panel members: not least, confidence that
the doctoral consortium is going to work well and be both useful and
enjoyable.
- Past alumni of CHI doctoral consortia, and not. At least one recent
alumnus is a good idea.
- Some with previous experience of being a doctoral consortium panellist.
- Continent / country. Not all USA, but definitely some. In 2003 we
had 2 from USA, one UK, one from Brazil. The most important thing is to
represent the fact that there are very different PhD processes round the
world, not just the USA model.
Selecting the students.
The most important principle (stated in the call and in the policy document)
is that students should be at a stage to benefit: not too near completion to
change anything, not too early to have much to talk about. Next, the policy
is to pick the best (as opposed to those who most need help). In 2003 we had
41 applicants, of which about 35 would have been satisfactory to accept, but
only 15 places. Consequently you may well have many of similar merit, and
could choose to balance some other aspects: gender (you might have to avoid
selecting too many females if you want equal numbers), research area, country
of origin. There is no one way to define the latter, since we had many
students born in one country, educated in another, and doing their PhD in yet
another.
Mentoring.
One of the students we accepted, from Athens, told me he would never have been
accepted without a lot of help in writing his extended abstract from a CHI
mentor. You may want to make sure this is in place, as it is probably
important to achieving genuinely international equality of access.
Students all also present a poster.
This is important to integrate them with the wider conference: it lets others
see what the doctoral consortium students are working on, and conversely it
also gives a structure for other delegates to meet them, if they want to. The
students didn't much enjoy having to stand by their poster, but in fact many
did make further contacts by doing so. (A pep talk on making contacts should
perhaps talk about this as well.)
It would be a good idea to improve the support for presenting posters:
perhaps get together with the student poster committee member for this, but it
may be particularly important for doctoral consortium students, as they mostly
aren't thinking of the poster as a primary concern. I didn't pay attention to
this at the time, but noticed several small problems all of which would have
to be addressed before the conference:
- Several students took the specified poster board size as a spec. for their
poster size. However the boards came with metal rims, so posters have to be
smaller than the board size by at least an inch all round.
- The second of the two scheduled poster presentation times was a waste of
time, as coffee was 10 minutes walk away in a different place, preceding a
plenary nearly everyone wanted to go to, and additionally the organisers wanted
the posters gone by the end of the 30 minute slot. This short-changed all the
student posters.
- However the basic recipe of having the posters in the commons where most tea
and coffee was provided, along with other stands, is a very good one.
- Having posters grouped by topic, not by status (doctoral consortium,
student, interactive) was excellent: after all, "customers" are almost
certainly searching by topic, and this rightly integrates student work with
other research.
- The doctoral consortium students, at least, could have done with some
advice (before they set out for the conference) about what else can usefully
and appropriately be provided at the poster: business cards, printed copies
of their abstracts (not available to delegates since these were only on CD),
perhaps other handouts (e.g. fuller papers), notices about whether and when
they were looking for a job.
Students' extended abstracts are published in the conference proceedings.
This is important in getting them published, and accustomed to being so; in
allowing others to look up their work during the conference; and for a wider
audience to be able to find them by search afterwards.
Student talks within the doctoral consortium:
The main value is in getting discussion of each student's work, not in them
presenting. Partly because they have written a 2-page abstract, and all have
read it in advance, it is feasible for them to give only a short talk. I
instructed them to talk for only 10 minutes and that if they took longer, it
would reduce the discussion time in which they could receive comments. In the
event, some kept to 10 minutes, many went to 15 minutes, one went to 20
minutes. I didn't pressure them to keep to 10 minutes beyond that general
warning: again, the 45 minute slot takes the pressure off. A particular issue
here is that some students had made a significant shift in their work since
applying, and at least two turned out to have a depth of further relevant work
they hadn't even mentioned in their abstracts.
I also told them to be explicit about what issues they would particularly
like us to discuss afterwards, and they all did that. Some started their talk
with this, so we would bear it in mind during their presentation.
This is the text of two advance emails I sent about giving the talks:
Aim to speak for only about 10 minutes, in order to allow lots of time for
discussion from both the panel and the other consortium members.
You can assume everyone will have read your extended abstract at least.
There are basically two possible aims for a talk (or a spectrum):
a) Showing off, making people think you and your work are wonderful
b) Getting new ideas from the audience.
If you want to look good (e.g. in a job talk) then you hide all
uncertainties, and talk only of accomplishments. If you want to learn from the
audience, then you use the talk to state your problem, discuss your
uncertainties (e.g. the arguments both for and against things).
Since the consortium is closed, you don't have to worry much about showing
off, and should think about what you most want to learn from the audience, and
what you need to say in your talk to get them interested and in a good posiiton
to comment on the issues you most want discussed.
If your thesis is well developed and you are using the consortium to check
that outsiders don't see flaws in it you hadn't anticipated, then you will
want to give an overview of all your work. I think my advice then would be to
concentrate on presenting the logical skeleton argument at the core of your
thesis rather than details e.g. "The literature suggests these 3 ideas, and I
conducted an observational study that also suggests similar points plus this
additional idea; from that I want to argue that my design of my software
should centre on these 4 features. Is that a reasonable leap from observation
to design ideas, or have I included untested assumptions in that leap?"
However it is up to you to use your talk to promote the discussion you most
want to hear; and that will probably mean ruthlessly pruning your talk,
leaving out a lot of the detail that in other kinds of talk would be
important.
I sent a long message a while ago about the talks.
To paraphrase the main point:
In this consortium, you should not think about how to interest an audience
that isn't sure it wants to be there or listen to you, but on how to start a
discussion that will be of the most use to you yourself.
In my experience this might well include:
a) A quick overview to warm the audience up and remind them of what they
should have read.
b) Say what you feel is the most interesting aspect of it all.
I.e. what is really neat, what your personal enthusiasm in the project is.
c) What it is you would most like comments on: and you may want to spend some
time presenting details of this aspect in order to get those comments.
This will be very different for different people. It might be "I'm confident
about my implementation but am scared someone will be nasty about my
statistics" or "I'm happy about all the technical details but worried it could
all seem boring to an outsider: what are the exciting implications I might
suggest my work leads to?" or "I've got a literature review and an
implementation and a user study, but do these all fit together into a single
argument i.e. a single thesis?" (I've had students that should have worried
about this one a lot more.)
The talks seemed to me to work fine.
Having said that, practice at giving a talk is a worthwhile aim in itself, and
these talks are roughly the length of a conference short paper. This is
practice at giving a talk to a group of strangers (not in the student's home
department), although in a relatively friendly and supportive atmosphere. One
of the panellists persuaded us to have a discussion of presentation issues, and
this might be a worthwhile standard feature. If you are particularly keen on
this, you could reserve the last 5 (or 2?) minutes of each slot for comments on
the presentation style (as opposed to content), and address issues such as how
to handle difficult questions, whether to rehearse and how else to improve, and
not appearing too self-doubting.
Discussion constituents.
The main ingredients are the panellists' comments, comments from other
students, and whether the speaker manages to retain anything from the
discussion. Anthony Hornof made a crucial suggestion about the last point at
the start, which we adopted: that students should find a partner (who wasn't
presenting in an adjacent slot) to take notes on the whole of the discussion,
since obviously the presenter might be too anxious to remember it all
accurately. I in effect chaired the discussions and, without this being a
planned policy, I looked round the panel to see who wanted to start this time,
usually the other panel members went next, then me, then other students
commented. On the second day I kept the slots to 45 minutes fairly strictly,
but on the first day I let them run on until discussion waned. I regret this
reduced time for other things, but it may still have had the advantage that a
good way to encourage discussion from all is not to cut anyone off, and allow
discussion to proceed until they are sure they can say something as
interesting as the current discussant. With 45 minute slots, there is almost,
but not quite, enough time for discussion to run until no-one has anything to
say. Consequently there is not a big problem here, but we could ask ourselves
what would further optimise the quality of contributions.
Panellists' comments.
There was some prior email discussion among the panel about how to organise
our comments e.g. should we divide up the students and agree who should lead
for each one. In fact this was unnecessary. In most cases all four had
something to say, and never was there a student who had everyone floundering
for a comment. With hindsight I wish I'd reminded the panel in advance that
it wasn't necessary to make a comment for every student, though that is nice
if they found that easy, and that they should feel free to concentrate on the
ones they felt most interested in. Having four on the panel provides a very
comfortable cushion: not everyone needs a comment on every talk, and if one
has a crisis and misses part of the session, and another does no advance
homework it will still all work smoothly.
A potential problem is having panel members who talk too much, regardless of
the quality of what they say or time constraints. I was clearly the biggest
threat of this kind on this panel, but putting myself last to comment and
being the one responsible for time-keeping as well, largely contained this
problem (I think!).
Content of comments.
An issue raised in the feedback, but for which I'm not sure of a settled view,
is whether the comments should include suggestions about major changes of
method etc. On the one hand, a student with less than a year to go cannot
consider learning a new method and running new studies. On the other hand,
exposure to contrasting methodological attitudes is one of the more valuable
things that can be offered, not least in order to give the student practice in
defending their work to those who don't share their assumptions.
There is some discussion of this in the feedback.
Student comments.
I didn't plan specially for this in advance. However, again because of the 45
minute slots, there was plenty of time, and there were some student comments
from the first talk onwards. On the second day, for other reasons, I kept
calling on a student to make the first comment, and that set a precedent that
had other students pitching in comments before the panel made any. Maybe I
was lucky, or maybe that would be a good deliberate plan: to be lax with time
keeping for the first talk or two, to encourage hesitant students to make
comments and set a precedent; then later to call on students before the panel.
Another possibility would be to assign students in turn to lead the
questions. On the other hand, since in my view even experienced panellists
may not have anything worthwhile to say about every talk, simple assignment
may not lead to good quality questions. Another variant would be to instruct
them in advance to prepare comments on at least three papers of their choice.
Alternatively, I suppose a more systematic chairman would keep a note of who
had asked questions, and begin to call explicitly on those who hadn't ever
asked one rather than only waiting for volunteers.
On day two, next year's doctoral consortium chairs sat in on a few talks.
I had ordered the talks so that the ones most likely to be of interest to one
of them, Liz Churchill, were scheduled then. Liz joined the discussion, right
in her area of expertise, and both students and panellists later told me how
impressed they were with this contribution. In fact this was not the first
contribution by outsiders: two different student volunteers, attending the
session for occasional chores, got interested and contributed (useful)
comments. This raises the possibility of including guests more deliberately.
With 15 + 4 people in the core doctoral consortium group, a couple of extras
in no way felt like an intrusion on the group's cohesion: certainly not after
the first morning. It might be possible to suggest that meal guests, or
others you happen to know are nearby, sit in for a talk or two. I had
suggested to the student volunteer organiser that it might be good to allocate
student volunteers who had applied but been rejected by the doctoral
consortium, but I'm not sure if this was done systematically.
This of course interacts with ordering the talks.
Most of the students would prefer to speak early, to get the anxiety over
with, and to make sure they get a fair share of the time. However someone has
to go last. I did pick a student I thought better able to cope with this to
go last, and I did take care not to erode the 45 minutes for the last talks.
I made up an order at breakfast on the first day, and so didn't give them
advance notice of the order (apart from the first speaker whom I'd warned the
night before). After all, this is not like a conference with people rushing
off to other talks after giving one. I roughly ordered them with similar
topics next to each other (although I wondered if the opposite policy would be
better, so that someone with specialist interests wouldn't have a long period
with nothing near their own topic), and with a view to the interests of the
two scheduled visitors. Having the 30 word descriptions would have made this
easier, and I would now consider the possibility of inviting more visitors to
the discussions.
Other talks/activities.
Most found it hard to sit through continuous sessions, and wanted frequent
breaks. Since they usually talked a lot to each other during the breaks,
these were in any case useful in their own way. It does however reduce the
time left for other things.
There was certainly an appetite for some
other kinds of discussion or short talks by the panel. My timekeeping wasn't
strict enough to do as much of this as might have been good. I was uncertain
in advance what these students would like to hear (after all, they are better
than the average PhD student, and they come from different places, so I didn't
expect my standard spiels to PhD students to be necessarily appropriate). At
the start, I got them to suggest a list of topics they would like to hear
about ("what are the general issues about being a doctoral student that you
find difficult?"), and we covered some (*) of them:
- (*) What this doctoral consortium is about
- How to write up a PhD
- (*) How to get an academic job
- What other kinds of job are possible?
- (*) Don't hesitate to approach researchers you are interested in at the
conference
- What else to do at this conference
- (*) Presentation style (a veiled reaction to the varied quality of the talks
given in the doctoral consortium).
- How PhD programmes vary around the world
- (*) Being a woman as well as an HCI researcher
- Map different areas in HCI: see interactions between different interests.
The panel hadn't overtly agreed what we might talk about, but I had alerted
them to the general possibility and we had probably each thought a bit about
it. This turned out well, with each topic we did talk about being contributed
to by several panel members rather than a single prepared talk. Particularly
useful, even though brief, were Anthony Hornof's remarks as part of his
self-introduction on how the doctoral consortium is a form of the kind of peer discussion that
is so central and essential to any researcher's work.
In addition, and following some previous CHI doctoral consortium practice,
there was a women-only subgroup at one of the lunches on being a woman in this
(or any) research field. It seems clear this is greatly appreciated by the
women participants (though the need for it is less extreme than it used to be,
with strong female applications to the doctoral consortium easily equalling
the male ones, and finding good female panel members being little problem).
Unsurprisingly the men felt a bit left out. Since I was warned about this
well in advance, I should have dreamt up a suitable alternative (it must be
too long since I was in a men's group). In fact it would have been
interesting to discuss whether we men had picked academic life as more easily
compatible with being an active father, whether we'd move to suit a partner's
career, and whether we can cope with the present strong female competition in
HCI. There is also now a group of women CHI participants (contact
Judy Olson)
who probably want to make contact with the doctoral consortium participants.
This year they had a birds-of-a-feather session, next year perhaps it will be
still more organised.
Introductions.
Unlike many meetings, in this case everyone already knew everyone else's
official face: by reading the students' abstracts etc., and the web pages of
the panel members. In fact, I now realise it is possible, easy, and useful to
start the community in advance by digital means. I should have asked the
students too for their web page URLs and/or photos, and put up a simple web
page to collect these for the doctoral consortium. It is unlikely that
people, especially the panel, will feel they have time for email communication
with strangers, but allowing everyone mutual pre-familiarisation in their own
time seems worthwhile. (I only partially thought of this at all because of a
request in advance from a student, and it is reinforced by some of the
feedback afterwards.)
The original recipe called for starting with dinner the night before, but
since I couldn't find a flight that got me there in time, this was changed to
drinks in the hotel bar. In fact not everyone turned up for that (only 2/5ths
of the students and half the panel). Thus I regard a prior social meeting as
both not really necessary and apparently not wanted by a substantial subset
(who undoubtedly were already in town); although offering it is certainly the
friendly thing to do. On the other hand (see feedback), it is
highly valued by another, perhaps larger, subset.
I began the first day's session by short self-introductions. I had everyone
write a nameplate to put on the table in front of them. I went round the
panel asking them to cover: their previous doctoral consortium experience,
their CHI experience, their own research topic, their area of expertise, in
what ways they were interdisciplinary. Then round the students, asking them
to cover: how to pronounce their name, the countries they came from, how many
conferences of any kind they have been to, and something / anything about
themselves that is not in their (generally very professional) applications to
the doctoral consortium. Thirdly, I asked everyone to suggest possible topics
(put up on the flipchart) they would like discussed ("what are the general
issues about being a doctoral student that you find difficult?"), starting
with a couple of example topics I'd elicited in the bar the night before. My
real aim in these introductions was to get everyone's voice heard briefly, in
order to set an unconscious precedent; while the content was just to
complement the more serious content in what we'd read on the web about each
other. Anthony took the opportunity to make two important contributions:
suggesting students get a colleague to take independent notes of their
discussions, and putting his view on what the doctoral consortium was about
(see above). This all took about an hour.
Changing seats.
At the start of the second day, the students switched some of the name plates
around to get people to sit in different places. A further suggestion in the
feedback was to do this every half day at least.
Refereeing process.
You have to invent one. I had about 42 applications for 15 places. I
pre-screened out about 5 who seemed clearly not at the right stage of their
PhD. You have to decide how much reviewing your panel will do: after all, if
they don't like reviewing they'll want to avoid it, and if they do like it
they will already have done a bunch for other parts of CHI.
The main issue is that rejected applicants would have liked feedback, which
I didn't provide, though I got a number of explicit requests for it. This
would require the panel to generate written feedback useful to applicants (as
opposed to useful to the selection: which will often be summaries of content,
and comments on research methods). In my view, the trouble is that applicants
want to know why they were rejected, but the main answer is: too few places
for all the acceptable applications. It is not true that writing a better
application would have made them a better person more deserving of a place.
Applying to the doctoral consortium is not a lifelong skill worth practising.
One of the features for CHI03 which I strongly approve of was putting all
the student opportunities together in a coordinated way (see the website and
call for participation): the doctoral consortium, student posters, and
student volunteers. We encouraged people to apply for them together, though
this means coordinating acceptances to student posters and the doctoral
consortium. This is clearly best for conference goers, organising it from
their point of view, and particularly for students who will not be familiar
with CHI.
Application documents.
Applicants were required to submit:
- An extended abstract. This will be a public, published document, and tells
you about their research.
- A personal research statement: to be circulated to other doctoral consortium members, and
express what they most want discussed. It would probably be best to provide
(contrasting) examples of this, as responses varied and some were very long and
formal, instead of saying realistically what would be useful to them and how.
- A letter of support from their advisor. These were often long, but not
that helpful as it's hard to know what degree of over-statement to expect from
different academic cultures. Basically, it checks whether this student is
real and supported by their institution; plus often it was the best source of
information about what stage the student was at. It might be better to ask
supervisors more explicitly to cover this question in a way that doesn't
presuppose the reader knows the structure of the PhD programme: e.g. ask when
completion is expected, how much data has been gathered, how many full-time
equivalent person-months of research (as opposed to coursework, or RA work for
other people) this person has done on their PhD. Both the policy document and
supervisors' letters show that most people aren't aware of any other PhD
programme structure than the one they are themselves engaged with: they just
know it's "different" elsewhere.
- An estimate of expenses. This opens the way to possible negotiation about
extra support above the $1500 limit for some students, though this year I
didn't do any of this as for those students who needed it, their departments
seemed able to contribute.
- A poster layout sketch. Really just to concentrate their minds on this
as a real requirement they will have to meet by the conference, plus
compatibility with applying for a student poster place. This generated the
biggest number of queries. Better advice (liase with student poster chair on
this), and above all putting some (contrasting) examples online would have
been a big help. The other issue here, was that some generated complete
posters in huge files, that wouldn't always print out, and overflowed a) the
ACM email relay; b) some PDF reader programs; c) some printer hardware
buffers. It might be best to insist on a max. file size, and require that
they themselves produce a reduced version for this "sketch" that prints on a
black and white printer.
- I would now require in addition a 30 word description of their
research topic and contribution.
Application process.
This year, at least, they sent these to me direct by email. Quite a lot of
work (about 42 applications), but I didn't have to learn some foreign software
I couldn't control. They were required to use PDF format, but in fact some
couldn't: since I could convert Word to PDF easily, this wasn't a problem.
Some supervisor letters, sent separately, were late, but not a big problem.
Acceptances to student posters and the doctoral consortium must be
coordinated, since some apply to both. This means accept/reject emails have
to be coordinated too.
Get Carol Klyver in during the consortium to cover the expenses
reimbursement process for the students. (15 minutes at least)
Get the student poster chair in to talk briefly about how to man/present
their posters at the designated times?
The web page call promised that applications would be acknowledged by
return. I would change that wording. In fact three different acknowledgements
are needed (and perhaps some are better done by web page than by 3 times 42
personal emails). a) I got some kind of email submission from you; b) I
decoded your attachments OK and they seem in the right format; c) I have
now got all your bits including your supervisor's letter of
recommendation.
I now regret not creating a web page in my own space, and having the
conference site point to it. That would allow me to announce what stage the
process was at e.g. "I have now sent acknowledgements to the 41 applicants"
(so if I was late, I wouldn't get 41 enquiries, yet applicants could tell if
they had indeed lost an email to them): this can address the issue of how
someone can detect whether an email has probably got lost. It might be a way,
if no-one objects on privacy grounds, to show each applicant's application
stage (perhaps on a separate, unadvertised, web page): whether all 5 documents
have arrived (or which are still missing).
Don't forget in sending out rejection and acceptance emails that you should
also tell the accepted who else was accepted, and who is on the panel; and
circulate to all the abstracts and personal statements of the accepted ones.
Furthermore they wanted to know how many applications there had been (i.e. how
lucky/good they were). Another (unadvertised) web page under your direct
control is handy here. The conference committee want this information: so
also tell the participants.
The students wanted ribbons on their badges similar to those accepted for
student posters only. Certainly helpful when presenting their posters.
Other post-acceptance administration.
- There was a lot of messing around about hotel bookings, since the web site
pushed students to book immediately, but the organisers wanted to encourage
room sharing and to optimise placement in the light of their commitments.
- Check that all the consortium participants also get all the announcements
from the student poster organisers.
- I should have sorted out much earlier the time they were required to be
there for the first social event the night before, as they need to book their
travel early.
- Similarly, it wasn't clear to them when the conference ended and what that
meant for their travel plans (given travel time to the airport etc.).
Consequently several of them in effect missed the last day of the conference
(see the feedback). This is a mixture of not very clear general conference
website, plus their inexperience: I should have been on top of this.
The expenses of the doctoral consortium are covered more or less
completely by a grant from the NSF. Because the money must be paid to a US
academic, it was applied for by Gary Olson. He said it only took 5 minutes,
re-using an old application. Currently the NSF is very happy with this, and
fully expects to fund next year's in Vienna. Carol Klyver suggests that the
amount should be raised: it hasn't increased despite inflation for a while,
and the increasingly international group means not all expenses are covered.
The policy document is out of date (not revised since 1994) in the dollar
amounts mentioned, and in only offering 1 day's expenses to panel members,
while a full 48 hours labour is now required. In my opinion you should lobby
to have this changed now. The money should be recalculated, the $1500 limit
reviewed, and both policy and the amount of the NSF grant revised together.
I would contribute the following points to the wider experience from the
organisers of the student poster stream:
Better guidance and above all examples of the "sketch" required for submission.
Set a file size limit and stop them sending complete posters that only print
out on giant colour printers: this is inconvenient for the review process.
Provide advice on other things presenters might bring to a poster: business
cards, printouts of the abstract in the (CD only) proceedings, whether they are
interested in job offers. A photograph of themselves so others can find them
at the conference.
Given student inexperience, perhaps develop a little advice on presenting a
poster: why it's worthwhile, what to do, how not to feel embarrassed, lonely,
and rejected. (Send this out with the acceptance email? Put it in their
conference packs?)
Perhaps organise a little student-only bonding e.g. special poster preview for
all doctoral consortium and student posters to look at each others' posters, to foster community
among the students at the conference.
Having the posters in the "commons" where all tea and coffee breaks are served
(and other stands and exhibits are) is excellent.
Mixing up categories of poster by research topic was a great idea. Scheduling
other posters and student posters for different times (but same place) was a
great idea: reduce crowding, but raise the chance of people noticing other
posters on the same research theme.
The second student poster presentation scheduled slot was a dud in 2003, as
everyone was directed elsewhere for tea, coffee, and a plenary talk at that
time. They wuz robbed.
Poster board size: warn about leaving a margin if the rim is metal, and give
the actual size the poster can be, not the furniture-centered view of how much
space the board takes up in the hall.
While the main part of this document was about rationale and what was done
because it seemed natural, this is a summary of some of the things that I
didn't think to do but would wish to do in future.
Get policy document changed. Review the amount of money applied for to NSF.
The call for participation:
- Maintain integration of student opportunities
- Require 30-worders
- ?Make sure mentoring is available and signalled
- Get the conference web site to point to a page of your own for signalling
the stage you are at, and what emails they should or should not have received.
- Organise online examples, particularly for a) poster sketches b) personal
research statements.
Get a delegate list out of the organisers so as to be able to target people to
visit the doctoral consortium students' posters. AND so as to invite
"notables" to meals. A list of paper presenters would do almost as well.
These are best obtained a week or two before the conference, but prepare the
organisers for this request.
Better support for the poster presentations.
Precirculate the students with the list of panel/faculty, plus their URLs. In
fact start the doctoral consortium as a digital community by collecting
everyone's web page in one place, plus the accepted students' documents
(abstracts etc.) (perhaps by a URL to a page with all the documents and/or a
big zip file for mass downloading and printing). Make it clear everyone
should have read all the abstracts in advance.
30 word descriptions required in the call, or anyway later. You will need
them, but also probably good on a web page for everyone.
Ask for a flipchart so you can have things like the talk running order
permanently displayed.
A student volunteer is not really essential, but nice in case running around
has to be done for this and that.
We might distinguish between what we want and how to get it; between:
a) Aims, desired outcomes, requirements,
b) Implementation, techniques, methods.
The aims, major and minor, are summarised above in
"Summary of the key requirements".
The major methods are in effect embodied in
"the CHI recipe for a doctoral consortium", also
summarised above. All that remains is to summarise a few points about
smaller scale methods that probably are important in achieving the aims.
- Start the community in advance, digitally. It may be that actual
interaction isn't appropriate or wouldn't get everyone involved. But simply
sharing pointers to web pages as well as the Abstracts allows everyone to
privately envision the others, and lay the ground for community on the day by
making them all seem to some degree familiar already.
- The lowest level methods for promoting good discussion.
- Making sure everyone says something early on, however trivial, just to set
the subconscious precedent in their minds. Perhaps a bit more systematic
turn-taking round the whole group would also be good periodically.
- Chairing to promote quality first.
- But perhaps, having someone (the chair?) connect contributions together;
showing you've listened and heard what previous people said. This is a) the
difference between a discussion, and set of unconnected fragments of monologue
(the opposite of the feminist gibe that women converse, men merely take turns
giving monologues). b) clapping is polite applause; but a deeper politeness
is to show that you were listening and valued what was said by having it
change what you say.
- A possibility for the chair is to keep track of who has made comments, and
begin to call explicitly on those who haven't. Potentially this is trading
off equal participation against maximising the quality of comments, although
not if individuals refuse the opportunity until they have something important
to say.
- Another possibility for the chair is to reserve the last few minutes of
each slot for
- Comments on presentation rather than content
- Feedback from the speaker (and others) on the quality of the comments made
e.g. too broad, too narrow, what else they would like comments on.
In principle this would keep the meeting focussed on quality by getting
regular feedback on it in time to affect what happens next.
- Consider putting in work on intra-group relationships e.g. shifting seating
places regularly.
- Consider doing more work on introducing other researchers to students:
- At meals
- Sit-ins for a talk or two
- At poster times
Web site logical path:
[www.psy.gla.ac.uk]
[~steve]
[chi03]
[this page]
[Top of this page]