Workshop Report: Southern Appalachian Symposium on Programming Languages and Systems (SASPLAS)

Phil Pfeiffer

25 October 2005

Thank you for putting this together and giving us (students) the opportunity to attend.  Thank you to the sponsors who allowed this event to be fee-waived for students—otherwise I might not have been able to attend.

Executive summary. The first SASPLAS workshop was held on 22 October 2005 at East Tennessee State University.  The workshop was attended by 74 people from 14 regional schools, 10 of whom spoke and 4 of whom presented posters.  Feedback about the workshop was largely positive, apart from concerns about the physical arrangement of the room in which the workshop was held.  The larger-than-expected turnout was due in part to the lack of an admission fee, thanks to $3,500 in donations from IBM Research, Microsoft Academic Research, and the UT/ORNL Science Alliance.  The $500 that remains from these contributions will be used to support a second SASPLAS, if one is held, or donated to a workshop with comparable goals, if not.  If a second SASPLAS is held, the host should think carefully about retaining certain aspects of the first SASPLAS—like the emphases on informality and student presentations-and revisiting others, including strategies for setting a date, distributing the workload of conference management, improving various aspects of the workshop's logistics (e.g., venue, format, and accommodations), and involving more nearby schools in the event.

Introduction.    SASPLAS, a student-oriented symposium on programming languages and systems, was held on 22 October 2005, at East Tennessee State University, in Johnson City, TN.  The workshop was attended by 74 people, including 56 students, 13 computing faculty, 3 keynote speakers, and 2 alumni of ETSU's CIS department.  Participants came from 14 regional schools, including Appalachian State Univ. (2), East Tenn. State Univ. (33), Furman College (1), King College (4), Middle Tenn. State Univ. (1), North Carolina State Univ. (2), Northeast State Technical CC (7), Pikeville College (1), Tenn. Technical Univ. (10), Univ. of Kentucky (1), UNC-Chapel Hill (1), Univ. of Tennessee-Knoxville (3), Virginia Tech (2), and Wofford (1).

The program started at 9 am.  It ran until 5:40 pm instead of 6 pm, thanks to the brevity of the talks in the third student session.  Presentations included

• two hour-long keynote talks by
  • Dr. Beverly Sanders of U. Florida, on design patterns for parallel programming, and
  • Messrs. John Mugler and Thomas Naughton of ORNL, on the OSCAR cluster toolkit;
• nine thirty-minute student talks by
  • three Ph.D.-level students (Brian Crawley of Kentucky, Huandong Liu of Tennessee, and Roy Tan of Virginia Tech);
  • four MS-level students (Matt Estes of Tennessee Tech, Jonathan Chapman of Appalachian State, Zach Lowry of Middle Tennessee, and Trey Buck of East Tennessee State); and
  • three BS-level students (Ryan Miller and Hunter Davis of North Carolina State (joint), and Liz O'Quinn of Wofford);
and
• a forty-five-minute poster session, by students from Virginia Tech (Matt Thornton), Tennessee (Nick Buchanan), East Tennessee State (Robert Henry), and Wofford (Liz O'Quinn).

The day also included two fifteen-minute breaks and a forty-five minute lunch break.

People in attendance were asked to evaluate every presentation, based on two questions:  If the speaker were to give the talk another time, what should stay the same, and what should change (and why)?   Forms were collected following every session, and turned over to the presenters.  A final evaluation form asked for "keep this/change that" feedback on the workshop as a whole.  The results of 29 evaluations that were returned are summarized in Appendix B, and discussed throughout this report.

Publicity.  Preparations for SASPLAS began in July, with an initial canvassing of web sites for regional computing departments. Calls for presentation (CFPs) were e-mailed in late August to computing faculty at 72 universities, colleges, and community colleges in Kentucky (4), North Carolina (25), South Carolina (11), Tennessee (18), and Virginia (13).  Supplemental publicity included a September mailing of SASPLAS posters to 77 schools; e-mails to ETSU computing and honors students; e-mails to graduate students with interests in programming languages and systems at Duke, UNC-Chapel Hill, South Carolina, Tennessee, and Virginia; and phone calls to computing faculty at (at least) 34 schools.

Logistics.  About 5/6 of the 88 people who pre-registered for SASPLAS attended; the greatest number of no-shows (13) were from ETSU.  55 of the participants stayed through the closing remarks.  Attendance was about double the initial estimate of 38—an estimate that was derived by adjusting the number of people who attended the first MASPLAS (cf. http://www.research.ibm.com/masplas ) downward, to compensate for the greater dispersion of academic institutions in southern Appalachia.  In retrospect, this estimate proved low for three reasons:

• SASPLAS, unlike the first MASPLAS, was free to participants.
• The number of ETSU computing students far outnumbered the number of computing students at the site where the first MASPLAS was held.
• More people were willing to drive for more than two hours to ETSU and to incur the cost of an overnight stay in a motel than was originally anticipated.

This unexpected interest in SASPLAS, coupled with a surge of last-minute registrations and a late registration deadline (8 October), forced a last-minute switch to a less than ideal venue: a pair of adjoining dining rooms in the ETSU student union.  These rooms had their good points: they were large enough to host the workshop; small enough to keep people from feeling dwarfed; convenient, in that they allowed people to eat while listening to talks and or discussing posters; and—most importantly—they were available, along with support staff.  Unfortunately, the chairs in these rooms were uncomfortable; the screen was hard to see from the back of the room; and a north-south table arrangement and a support post in the middle of the room made it harder for some participants to see the speakers.  Speakers also had to make do with a fixed-wire microphone: a wireless mike would have cost an additional $140 to provide, due to the Center's requirement that a dedicated technician be on site when a wireless mike is in use.  Several of the student speakers stood too far from the microphone, creating additional problems for some of the talks.

The lone problem with catering arrangements involved a failure to request more cheese and less meat for lunch, in anticipation of the number of vegetarians at the workshop.  Otherwise, people seemed satisfied with the food.

Costs.  The workshop's sponsors—IBM Research, Microsoft Academic Research, and the UT/ORNL Science Alliance—pledged a total of $3,500 for SASPLAS.  About $3,000 of this was used to fund SASPLAS (cf. Appendix A).  Additional support was provided by ORNL's CCS group (Dr. Stephen Scott, manager), which sponsored John Mugler's and Thomas Naughton's travel to SASPLAS.  Other unexpensed costs, which were borne by ETSU and the conference's host, included postage, day-of-workshop posters, travel between Tri-Cities Airport and ETSU for Beverly Sanders, and dinner for the speakers.

Wrapup.   Slides from talks are currently being posted to the SASPLAS website.  The website's main page will be reworked before the end of the semester, with two new subsections: this report, and a collection of workshop photos, courtesy of ETSU student Richard Metcalf.  The unspent $500 will be used to support a second SASPLAS, or donated to MASPLAS if the second SASPLAS proves infeasible.

Retrospective.  SASPLAS was organized as a low-key, low-cost workshop on computing: one that allowed some students a chance to polish their presentation skills, and others to participate in a conference-like event in an informal setting.  Judging from feedback from workshop participants (cf. Appendix B), the workshop met both goals.  Some participants also used the workshop as an opportunity to network with people from other schools, as evidenced by the requests for more break time.

The response to the CFPs was gratifying, in light of the haste with which SASPLAS was organized, and the time of year when the workshop was held.  SASPLAS drew 38 participants from 13 other schools, in spite of conflicts with homecoming celebrations, computing conferences, and other fall activities.

Other aspects of the workshop that seemed to work well included the following:

• scheduling the workshop for a Saturday: past experience with MASPLAS suggested that M-F does not work, due to conflicts with classes;
• the orange/blue/black SASPLAS poster: people remembered the poster, even if some didn't think carefully about what they were reading;
• soliciting abstracts from student presenters: this helped to ensure that participants had a well-defined topic to present;
• using a website and e-mail to compensate for the last-minute planning, by disseminating information about logistics as it became available;
• the use of four assistants to help run the workshop: two to staff the registration desk, one to take photos, and a fourth for extras, like making photocopies for poster presenters;
• providing participants with a simple, individualized workshop packet, consisting of a name tag, a pen, a two-sided program, a set of evaluation forms, and ten sheets of loose leaf paper;
• the use of a backup laptop for the workshop: even though it wasn't needed, it was good to have;
• using faculty as session chairs for the five speaker sessions;
• the inclusion of one talk and two poster presentations on concerns other than programming languages and systems research: this helped to highlight the changing nature of computing research (a point that was made during the opening remarks);
• giving student speakers 25 minutes for talks, and telling them in advance that the 25 minute time limit on talks would be strictly enforced;
• the use of simplified evaluation forms (what worked well?/ what could have been done better?) to collect feedback from participants: detailed questionnaires and tight schedules don't mix, and at least one participant requested more time for completing forms as it was;
• the use of the website to disseminate speakers' materials: abstracts and slides for each of the eleven talks and two of the four poster presentations should be on-line by the time the website is complete.

Aspects of workshop management that should be revisited, if someone agrees to host a second SASPLAS, include determining a date, distributing the workload, implementing selected improvements to the workshop's logistics and format, and working harder to find ways of including local schools.

• Determining a date .  SASPLAS was scheduled for October 22 because no other fall date was open, or seemed practical.  Dates before 8 October were rejected, to give people time to respond to the initial CFP, which was not mailed until 23 August.  October 8, 15, and 29 were rejected, due to conflicts with a regional storytelling festival, ETSU fall break, and ETSU homecoming, respectively.  Finally, later dates were ruled out, due to concerns about inclement weather discouraging participation from schools in Kentucky and Virginia. 

The choice of October 22, unfortunately, conflicted with homecoming celebrations at Carson-Newman and Furman (and maybe other schools); the annual CRA-W conference, which complicated the search for a female keynote speaker; and OOPSLA, which concluded two days before SASPLAS, and made attendance less attractive for at least one participant.  October, according to hoteliers, is also one of the busiest travel months of the year: scheduling any event in October risks conflicts with personal plans and events like conference committee meetings—which created a conflict for one would-be participant, a member of the Mars Hill faculty.  A final concern was the event's proximity to the ACM and CCSC southeast regional conferences, which were both scheduled for early November.

Given these observations, a better option might be to schedule a second SASPLAS for the end of September—or to move it to April, at a time of year when academic research projects are traditionally completed and there is less competition from "last-gasp" fall events.

• Distributing the workload .  Experience with MASPLAS, the workshop upon which SASPLAS is modeled, suggests the importance of community involvement in computing workshops.  The first MASPLAS, like the first SASPLAS, probably required at least 160 hours of work: i.e., about 10 hours of work per week, for the 16 weeks up to and including the event. 

MASPLAS persisted because faculty at different schools took turns hosting that workshop—and in sharing in the benefits of having a conference at one's doorstep.  This sharing of hosting duties is important, because the most of the work of hosting each workshop typically falls on the shoulders of one individual at the host school.  An active steering committee could also help a host by helping with publicity.  Publicity is particularly important for a newly established workshop like SASPLAS, where word-of-mouth is needed to supplement e-mail-based and postal calls for participation. 

The single most difficult aspect of organizing SASPLAS was finding the time to phone people at regional schools once the semester was underway—a problem that the host never quite completed.

As of 25 October 2005, no one has offered to host a second SASPLAS. However, four people who attended SASPLAS have offered their help with organizing a second: Jay Fenwick of Appalachian State; Dan Perry and David Frazier of Northeast State Technical Community College, and TTU alumnus Matt Estes.

• Improving logistics.   The person who hosts the second SASPLAS should arrange for speakers to present in an auditorium with a wireless microphone.  The host should also consider adding more slack time to the event, to allow presenters to discuss their work with attendees; providing support and actively encouraging software demonstrations; and moving the poster session to between the second and third talk sessions, if a three-session format is retained.  Another idea is to adopt the MASPLAS practice of hosting a panel discussion on careers in computing in place of the second keynote, with speakers from industry, a small school, a medium-sized school, and a full-fledged research university.

Access to the Internet, as per one participant's request, would be a mixed blessing at best.  Providing Internet access would distract people from the presentations and from speaking with others at the workshop.

• Encouraging more local involvement. A final concern is finding ways to encourage more schools that are within 1-2 hours of the host site to participate.  About 2/3 of the schools that are local to ETSU failed to participate in SASPLAS, namely Asheville-Buncombe CC, Carson-Newman College, Emory and Henry College, Hiwassee CC, Lees-McRae College, Mars Hill College, Milligan College, Mountain Empire CC, Tennessee Wesleyan College, Tusculum College, UNC-Asheville, U. Va.-Wise, Virginia Highlands CC, Virginia Intermont College, and Walters State CC.  The primary concern about this lack of participation is that the students at schools like these—students who would not ordinarily have a chance to attend a conference or a workshop—are the very students that SASPLAS is trying to reach. 

Appendix A:  Record of Expenses

Appendix B:  Participant Feedback

Appendix B.1. Evaluation form

If we were to organize a SASPLAS '06, what about the workshop would you wish to see the same?

6 comments about the variety of presentations.

4 comments about the overall quality of the presentations.

3 comments about schedule management.

3 comments about the informal/unintimidating atmosphere.

3 comments about the food.

2 comments about the length for student talks (25 minutes, + 5 minutes for questions).

2 comments about everything/practically everything going well.

2 comments about emphasis on student work.

2 comments about wanting to see repeat SASPLAS events.

2 comments about the inclusion of non-traditional CS applications (bioinformatics).

Local speakers.

Range of topics covered (right size of coverage).

1 day is good.

Really liked that it was geared to students who have never attended a conference also.

I very much enjoyed the opportunity to share my presentation and I would encourage fellow students to participate in such conferences.

SASPLAS in general, it's interesting to see what other universities are doing.

The info packet and sponsorship were excellent.

Keynote speakers were good, particularly the design patterns presentation.

Time for presentations seemed about right.

Evaluating the speakers is a good idea.

Really liked the presentation setup—type 1, 2, 3.  Worked well.

Great event and experience.  I enjoyed mingling with students from other schools.  Interesting topics.

Venue was good.

Lots of speakers, lots of feedback.

The general format is good. 

Interactivity.

Dedicated poster session.

Involvement from a variety of schools.

Overall it was pretty good.

"Yes".

Size of conference.

Great conference.

Keep it how the presenters are on the same level as the audience.  Very easy to get up in front of everyone.

I think that lengthwise it was about right, but I wouldn't go longer.

I think it worked well for us to have students from UT, in that they could talk to students from other schools, which had an opportunity to hear about what else is out there post-graduation.

If we were to organize a SASPLAS '06, what about the workshop would you like to change?  Why?

5 comments about room layout, including post in middle of room.

4 requests for more break time—mostly, to talk with other students, presenters.

3 requests for larger screens.

2 requests for lighting the speakers.

2 requests for live demos of software.

A little more time to fill out evaluations for each presenter.

Stickers made so we can get more people interested.

Poster session totally broke the flow of the meeting—move it to the end.

Increased diversity in student speakers and their topics.

More advice for transitioning students (undergrad to grad, grad school to "real world").

Consider: to help students with writing skills, a 2, 3, 4, page "extended abstract" of work would be useful.  On the other hand, that extra "submission requirement" may decrease submissions.  On the other-other hand, a persistent, regular SASPLAS can be planned for!

No talks ran long, some ran short; possible 20 minute talk with 5 minute question session.

Poster session should be between student groups 2 and 3 and should be 25-30 minutes.

More programming languages.

Somewhat better advertisement, because this was good.

Uncomfortable chairs.

Shorter keynotes.

The event as a whole was a tad too long.  Breaks are good though.

The name.  There are too many jokes for the name.

Topics should be more specific than "systems".

Maybe more social networking opportunities.

Don't hold during weekend after OOPSLA.

Home Internet access.

Maybe have a booklet with all the presentations?

Fewer talks and more time per talk, because it didn't feel like there was enough time to learn that much.

Maybe more relevant talks; biology talks and grammar talks didn't seem that applicable.

Starts too early.  (We're grad students!)

More people.  (Consequently: late start + more people = second day).

Giving first, second, and third prizes for the speakers may be a good idea.

Maybe parallel talks?—if enough speakers to warrant.

Multiple lines at lunch.

People use microphone.

Appendix B.2. (Unsolicited) feedback from post-workshop e-mails

SASPLAS was interesting for most of the presentations. Thank you for allowing students to go free. If there is another one next year, I would like to help out and attend.

I really enjoyed SASPLAS 2005 and the opportunity to speak at it, and I am looking forward to the possibility of a SASPLAS 2006, and I would be willing to help in whatever way I can.