11 Proposals For A Better DCS

September 16, 2009

Incoming graduate chair Peter Marbach has been leading a series of “town hall” meetings to build support for a Department of Computer Science conference, organized by graduate students. The stated goals are a) to improve the quality of the graduate experience by promoting interaction between areas and b) to build the Department’s reputation in industry and academic circles by effectively communicating our successes.

I’m happy that Peter cares enough to try something big, and doubly happy that he is trying to listen to us and involve us in his work. But, while the goals are lofty, I don’t agree with the conference approach. I think that there are lightweight projects that would contribute more to interaction between areas, and I feel that hosting a conference is not an effective way to build a reputation for our work. I would also criticize the proposal for lacking specificity (especially budget, organizing team, schedule) and accountability (no simple success or failure metrics).

I hate to criticize an ambitious and well-meaning proposal without suggesting a replacement. Accordingly, I’ll list some specific proposals below, without taking credit for any of the ideas. Before I do, I’ll list a few of the community projects that already exist.

CSGSBS “big” events

The CSGSBS hosts large, general-interest events a few times a year, including an upcoming BBQ.

CSGSBS “little” events

The CSGSBS coordinates weekly events including cookie breaks, movie nights and pub nights. There are attended by a small percentage (~10%) of the grad student population, and most weeks attract the same crowd again and again.

Research-In-Action showcase

Annual conference hosted on campus with posters and demos from graduate students. A few industry and political guests attend. Typically heavily-weighted on demo-friendly DGP work.

Area talks

Most areas have weekly talks. Outsiders are welcome in theory, though an interested outsider would have to find the right mailing list and often the material is too advanced.

11 Proposals for a better DCS

1. Lightweight talks

1.a Area intro talk series

Each week, a different area gives the talk. At a casual, accessible level, the talk covers why the area is interesting, the most important fundamental closed and open problems in the area, key results, and how each student’s work in the area fits into the big picture. The speaker recommends a survey paper or introductory book for students interested in more. No posters, textual or mathematical slides or chalkboards allowed. Each proposed speaker is vetted by the organizer in advance to ensure a minimum public speaking ability.

1.b Monthly Mini-TED

Monthly casual 1.5 hour meeting in a lounge with coffee. Three speakers give TED-format talks on any beginner-accessible topic of intellectual interest not related to research. Given the variety of DCS member  interests, we might hear talks on sailing, design, typography, photography, quantum physics or hooping.

2. DCS Portfolio

2.a Technical reports database

For communication of early or small research, revive the technical reports database or provide a DCS-only externally-accessible site with ArXiv.org-like functionality.

2.b Publication portfolio

Provide an RSS feed with links to PDFs of all research published by members of the department.

2.c Entrepreneurial portfolio

Directory of links to companies founded by members of the DCS.

2.d Graduate success portfolio

Track the alumni network (undergrad and grad) and list the institutions they join after graduation.

2.e Blogroll

Directory and firehose of blogs written by DCS members. Michael Famelis has done good work in this area for software engineering. Extend it to the whole department. For bonus marks, provide a global WordPress installation on cs.toronto.edu for DCS member blogs.

2.f Software Portfolio

Directory of software written by members of the DCS. This can include side projects, not just research-related work. Subsection for links to software to which DCS members have contributed. For bonus marks, provide distributed version control hosting and virtual private servers for hosting web applications.

3. Activities

3.a Pick-up sports

Weekly pick-up soccer or ultimate games in the summer and fall.

4. Structural Changes

This is the controversial stuff.

4.a Common-space whiteboards

Move whiteboards out of private, locked areas into public spaces where they can foster discussion.

4.b Cooperative Education

This is the biggest proposal, with highest cost and greatest potential payoff.

For the undergraduate program, provide institutional support for four-month co-op terms after first and second years, and break the PEY into four consecutive four-month terms. For MSc students provide institutional support for corporate research internships between completion of class and start of full-time research. A common infrastructure would support both missions.

Why co-op?

Work terms create a feedback loop with in-class education. Real work motivates study and supports good study habits. Programming skills honed in the workplace support theoretical learning in the classroom or lab. Problems discovered in the workplace can motivate research directions.

Work terms create a feedback loop with companies. With more work terms, more companies see more students. Companies are eager to hire students who have worked for them before. Companies are willing to contribute resources to universities that are an important part of their HR pipeline.

Weakness of the PEY

I claim the following weaknesses in the PEY program:

  1. Students make long-term commitments on their first-ever round of interviews. They often have no professional experience, and thus are not yet qualified for advanced work. This restricts their placement to beginner work, but this commitment to beginner work lasts sixteen months instead of four.
  2. Some students just don’t fit with some companies. In other cases, companies abuse long-term students for menial tasks such as manual testing. Four-month terms salvage these cases, without restricting a student from returning to a particularly good employer.
  3. Students see only one corporate culture, team culture and set of engineering practices before graduation.
  4. Students only get one round of interview practice.

Each of these is addressed in the proposed model.

Addressing Common Objections to Co-op for the DCS

I have heard two common objections.

  1. Prohibitive cost: This program would be expensive to administer. Answer: Everyone wants something for nothing, but a program like this is an investment. Other universities successfully mitigate the costs of similar programs. We can do it too.
  2. Competition with Waterloo (and other schools): The implementation of such a program would position us as an alternative to Waterloo’s well-oiled machine, and would contradict our years of PEY marketing. Answer: We should neither hesitate to adopt a good idea just because someone else uses it, nor should we allow a desire for self-consistency to trump innovation.

Summary

A frank an open discussion of ideas might lead us to improvements in our graduate experience, provided we restrict ourselves to specific proposals with measurable outcomes.

Here’s where you come in

  1. Leave a comment in your real name, and rank the ideas you care about. “++” for “would actively support, defend and work for”, “+” is “support passively”, “-” is “oppose passively” and “–” is “would actively oppose, attack and work against.”  So if you hate co-op, dislike pick-up sports, but would fight to see whiteboards moved to hallways, write this: [3.a-, 4.a++, 4.b–].
  2. Go find someone else in the department, force them to read this through and make them think through their opinions on these issues.

Let’s send a message about how to make our department better.

Advertisements

32 Responses to “11 Proposals For A Better DCS”

  1. Rory Says:

    Again, I have to applaud Peter for his enthusiasm toward improving the quality of life here in the DCS. I do have to agree with Aran, however, that a large conference style event probably isn’t quite the right approach. I think that, if the aim of said even is to give members of the industrial sector an icon they can associate with U of T, this icon should be the quality of our work and strength of community, not the size of the buffet at our gala weekend conference. Now, if said conference were used as a vector with which to publicize our work and community, I would support it whole-heartedly, but I think that some of the smaller-scale changes that Aran mentions need to be implemented in order to achieve these levels of cooperation and pride before we have justification for such an undertaking as this conference.

    In short, a large-scale, formal research conference seems premature. Start small, build quality, then present. While U of Waterloo is symbolically tied to its co-op program, most of the other top name CS schools (berkley, stanford, mit) are simply associated with high quality and collaboration. Let that be the metric with which we are judged.

  2. Steve Says:

    All of your items 2.x are absolutely essential. I’ve no idea why we don’t have them.

    I like the idea of an annual conference, but perhaps because I’ve seen it work well elsewhere. When I was at Sussex, the grad students had an annual 2-day retreat. It was residential, at a field centre far enough out of the city that meant you really were fully engaged. Lots of games, full bar facilities etc., encouraged evening socialization. Faculty were prohibited from attending, except that every year they would invite one of the newest faculty members to come and give a keynote. Every grad student was expected to present a paper at the conference regularly (but not necessarily every year). The entire point was for building a sense of community, so it was entirely inward focussed (i.e. it only satisfied the first of Peter’s goals).

    Thinking on this now, I’m tempted to say that Peter’s two goals are probably incompatible – the more you try to satisfy one, the less you’ll satisfy the other.

    Having said that, I’d love to see DCS run a grad students conference, and I don’t necessarily think it even needs a clear vision and success criteria to be defined up front – try it, see what emerges, and then refactor. That’s how the research showcase evolved.

  3. aran Says:

    @Steve – a conference like the one you’re describing is something I would support without reservation. That sounds amazing.

  4. plagal Says:

    Just a note: the Planet is not for SE only, there are feeds from people’s blogs from all around (it just so happens that more SE people do blog 🙂 ).

  5. Rory Says:

    @steve I don’t necessarily agree that Peter’s two goals of attracting industrial contacts/reputation and promoting a collaborative, holistic grad student community are incompatible, in fact I believe that having a strong community will very much aid in attracting outside eyes. However, I do feel that one single initiative will not satisfy both objectives. Efforts to improve the community should be solely concerned with said community, just as efforts to improve industrial relations should not appear to be anything other than that. By mixing the two, I think industrial contacts will get a less-than-polished look at our research, and grad students may feel more like marketing tools than researchers.

  6. Rory Says:

    Also, I really like the idea of the private grad retreat. Eliminating outside influence/observation is one sure way to get people to open up a bit. One of my earlier visions when this discussion began was a conference in which the end result was a small number of massively co-authored works – sort of a code sprint for academics.

  7. Jorge Says:

    Like 1.a, love 2.x.

    Incidentally, 1.a is already a bit hard to pull off (due to lack of interest from lots of people), and noting this gets us halfway to understand why the larger conference is probably unfeasible.


  8. 1.a+ (don’t have the time to deliver, but attending is a definite)

    1.b++ (seen this work all summer at a different school — I think it would work well at U of T)

    2.a+

    2.b++ (exists from a recent U of T grad. http://www.bibbase.org/ http://www.cs.toronto.edu/kr/publications/ )

    2.c+

    2.d+

    2.e++ (done. as mentioned, it’s not just an SE blog)

    2.f* (meh)

    3.a+

    4.a++

    4.b+

    As for the conference, I would check out NESCAI ( http://www.cs.cornell.edu/conferences/nescai/ ). They didn’t hold one in ’09 as far as I can tell, but it was a very well (student) run conference for students. Few profs were keynotes, but it was all students presenting in the talks. My guess is that the students running it had to focus on their research — organizing anything of that magnitude can’t be good for productivity.

  9. Lee Zamparo Says:

    [1a +, 1b +, 2.x ++, 4b ++]

    Not wildly swung by any of the other proposals.

  10. Mike Conley Says:

    1a +

    1b +

    2.x sounds like a no-brainer, and I’m wondering why this hasn’t been done yet. ++

    3.a +

    4.a ++

    4.b ++ (I’d also be jealous of all the undergrads who get to do this)


  11. I like most of your ideas Aran.

    I say let’s start with the easy/cheap stuff that we can do without too much help from the department: 1.a and 1.b, the area talks and the mini-TED.

    I’m willing to help organize one/both. Who wants to help? Send me an email.


  12. […] September 17, 2009 · Leave a Comment Today I’m joining a meeting with our Associate Chair for graduate studies.  It’s part of a series of meetings aimed at gathering student input towards shaping our department.  One issue that everyone seems to agree upon is the barriers to inter-group collaboration: grad students are divided between four buildings (BA, SF, PT, CCBR), so research groups located in different buildings rarely (or never) see one another.  This is a big downer, as it inhibits new and surprising collaborations.  How can we make collaboration easier?  Aran Donohue has a few good proposals. […]

  13. Golnaz Elahi Says:

    Aran, I like your ideas. You asked for rating them:
    1.a ++ I love this idea, we can initiate it fast and easy. Each research group can host a cookie break and people in that group give short and cool presentations.

    1.b ++ This would be awesome, if the talks are awesome, otherwise, a very technical talk specially from theory people is enough to cut me from attending.

    2.a +
    2.b +
    2.c +
    2.d ++ and I have a comment on this. Down there.
    2.e +
    2.f ++
    3.a I’d like these activities, but I need to develop a very large i* model to finally conclude how sports can improve the reputation of our research groups and increases the chances we will get good positions later in the job market.
    4.a +
    4.b I have very minimal experiences here, so have no opinion.

    I wanted to be frank with Peter, but I am nice and shy, so I left the meeting early. I wanted to tell him, the problem is that I am not sure what kind of academic position I will get, if I’ll get any ever. I though U of T has this great reputation, but I have got this feeling and impression that Canadian universities tend to hire from American universities graduates, and American universities that we can get to are remotely-located schools in the middle of nowhere that just started growing. These are totally based on my personal observations and judgments over the last three years I have been here and only limited to SE lab: our graduates do not really get hired even in the middle-ranked schools. (Here Aran’s suggestion 2.d is great about a data base of success stories, so we know with what level of research and publications, where we are going to end up)
    But, the big but is, we do good research here. What I observes is our students win the best paper awards of top conferences and publish in the best venues.
    So, what is this mystery and how are we going to solve it?

    I asked Peter for a sort of cross research groups seminars, like SE seminars we have in our lab. So in the last years of grad school we improve the presentation skills for not only the SE audience but for other CS and general scientists audience, and we learn how to explain our research for non-SE and non-CS people; hopefully, to increase the chances that when we go for a job presentation, we break the cultural and biases they have, and we will get the jobs.

    If the idea of big, industry-involved, 2-days conference of Peter wants to survive, we need motivations for graduate students to participate. Free beer and a lot of fun (@ Steve) work, but seriously, for me at the last years of my PhD, I am only thinking of getting it done and getting that empty position. If Peter invites faculties from the universities that have some openings for the next year, I will volunteer to give a talk, sure!

    New students may have better suggestions of how they can improve their graduate program, my ideas are now biased because I am getting worried about the end of the tunnel.

  14. A. Turing Says:

    Someone will have to explain to me what “promoting interaction between areas” has to do with “improv[ing] the quality of the graduate experience”. Furthermore, I don’t see how a conference (student-led or otherwise) would improve either one. Most people will find most work incomprehensible or the work would have to be watered down to be a general overview of the field.

    Now, if we “build the Department’s reputation in industry and academic circles”, THAT can improve the quality of the graduate experience! But that’s everyone’s concern, ain’t it?

    Are Golnaz’s comments correct? Do “our graduates do not really get hired even in the middle-ranked schools”? Perhaps 2.c doesn’t exist for a good reason?

    UofT should decide whether it wants to be a top Canadian university or a top world university. If it decides it wants to compete on the world stage then it needs to adopt the following policies:

    1) Bribe the editors of U.S. News and World Report to rank North American universities rather than only ones from the States.

    2) End preferential admittance for Canadian students. End preferential hiring of Canadian faculty. The cream of the crop of 6 billion is better than the cream of the crop of 33 million.

    3) Advertise ourselves better at international universities. This is emphatically not the same thing as inviting local businesses to visit our campus for a conference. It may include sending our students to work with international research groups; it may be getting our faculty to ask their colleagues to recommend Toronto to their students.

    4) Follow-up with prospective faculty or graduate students who chose to join another institution. What were their reasons? What deficiencies did they see here or in their hiring process? Perhaps the things we’re discussing here aren’t the real issues.

  15. Abe Says:

    I think interaction between areas is a good idea but should be applied outside of DCS as well as in. I’ve mentioned this before but let me reiterate here: Theory grad students would benefit from interaction with the Math department; Systems should be closely tied with ECE; HCI could interact with psychology or cognitive science; Machine Learning should work with statistics; KR and Operations Research both deal with optimization and scheduling; and Computational Biology should be working with our medical and biological faculties.

    At the very least, grad students should be permitted to take courses outside of DCS for credit and to fulfill any breadth/depth requirements. Longer term, cross-appointed faculty could help bridge these divides.

    I imagine there are bureaucratic obstacles but the benefits include experience working with people outside of a single niche, increased research collaboration through cross-fertilization, the opportunity to tackle old problems in new ways, a broader perspective on one’s own field. Furthermore, these benefits go both ways: we can expect many students from other disciplines to want to take CS courses too.

    Beyond that, I second Rory’s comments.

    For the ratings,

    1-: seems like more work that no one will want to do.

    2.a-: Technical reports don’t go on my CV so why bother?

    2.b+: Sounds good to get people’s info out there.

    2.c++, 2.d++: These are actually questions I asked when I joined the department. I assume past results predict future returns.

    2.e-: I don’t really see the point of blogging, generally.

    2.f-: Why is a central repository better than links off of homepages?

    3.a++: Starcraft!

    4.a-: We have shared whiteboards in my office and I’ve never seen one foster a discussion that wouldn’t have otherwise happened.

    4.b————————: Spending a year in four month increments slinging Java code sounds like a distraction from research. I think this would hurt rather than help most grad students. (For undergrads, internships are a good thing, but that’s what summers are for.)

    Rory said it better but, don’t be Waterloo, man! CMU, Berkeley, and MIT are not. In America, I’m familiar with Northeastern having a strong co-op program and, while it is a fine school, I’m *not* familiar with similar co-op programs at CMU, Berkeley, and MIT, even though I’ve had many friends at those places.

    Do good research; our reputation is based on this, not on internships. I’d rather focus on improving our research (e.g., through the cross-departmental collaborations mentioned earlier).

  16. aran Says:

    @A. Turing:

    To quickly get something out of the way, I’m upset by your anonymity. You should stand by your opinions, not least because we should know that you are a member of our community who’s opinions count. With that aside, your arguments deserve consideration.

    There are several ways to look at the link between interaction between areas and improving our experience. A simple one is social: A sense of working in a coherent community makes people happier. You’re welcome to disagree, but you’ll find many students and a large body of research against you.

    On conferences: First, they are not just about the presentation and understanding of work. They are also about the hallway connections and socializing. Second, a good conference presentation is rarely deeply technical. Technical details are best left to papers, and a good presenter shows the importance of their work at a level accessible to a broad audience.

    On our graduation outcomes: 2.d is the best way to find out one way or the other. But actually the facts are not relevant: If there is a perception of job-hopelessness, the perception itself is a destructive problem that should be addressed.

    Finally, on our stature as a top Canadian or top world university: It is not helpful to explode our scope. We do not need to globally optimize our entire department to improve it. We only need to make changes that make it better tomorrow than it is today. Put another way, like most universities, of course we want to be a great world university. (I disagree with you that we are not already a top world university). But all compete in a complex environment with political, social and financial considerations.

    Your first concrete suggestion isn’t serious. Your second fails to consider important political, social and financial issues, and also ignores the already-international nature of our school. Your third group of suggestions are not actionable.

    Your fourth suggestion is excellent, and is a simple expansion of 2.d.

    We are absolutely discussing real issues.

  17. Jocelyn Says:

    I agree with your proposals, except for 4b.

    The problem with doing a short (e.g., 4 month) internship is that by the time you’ve learnt enough to “be useful”, the internship is almost over. However, I agree that longer internships such as those offered through PEY also have their problems – mainly, students stuck in deadend internships.

    Question: do students have an opt-out clause when they sign up for a PEY?

  18. Jorge Says:

    Just wanted to add that Abe’s proposal (“At the very least, grad students should be permitted to take courses outside of DCS for credit and to fulfill any breadth/depth requirements.”) is one I fully endorse.

    We *are* currently allowed to take courses outside of CS for credit (though there’s some red tape to go through), but they don’t count for breadth. They should –perhaps not to the extreme of allowing us to get full breadth with purely external courses, but we should be able to cover at least one or two of our breadth requirements with interdisciplinary courses.

  19. George Says:

    I agree with Abe in most ways. I would love to have more interaction with statistics students.

    So here is what Abe said that I also agree with, changed slightly.

    1-: seems like more work that no one will want to do.

    2.a-: Technical reports might go on my CV but who cares about a repository specific to U of T? Not me.

    2.b+: Sounds good to get people’s info out there.

    2.c+, 2.d+ (would have another + except for apathy)

    2.e-: I don’t really see the point of blogging unless the blog will be of exceptional quality. Most blogs aren’t.

    2.f-: Why is a central repository better than links off of homepages?

    3.a++: Starcraft! MOAR STARCRAFT!

    4.a–: We have a whiteboard in the hall near my office right outside the offices of some of my colleagues. It gets used, but this is really not a general question for the department. People in local office areas should be discussing the need for more whiteboards in new places if such a need exists. This is not a general concern for the department. So I am not against additional whiteboard, per se, just against the whole department discussing this or trying to do mess around with it on a large scale. If you feel the need for a new whiteboard where you work, lobby for that whiteboard, but don’t try and put a whiteboard in my hall (unless it is also your hall).

    4.b——

  20. aran Says:

    @Jocelyn: Your objection is common but I don’t think the evidence supports it. Suppose it were true that 4 months was too short a time for a student to accomplish useful work. Then no self-interested company or research lab would pay anyone for a four month term. Programs like Waterloo’s, Concordia’s, UBC’s, Carleton’s or Berkeley’s would rapidly collapse. Students would never find summer programming work.

    In practice companies have many achievable and worthy projects suitable for four months of work.

    For grad students, the four-month number is less important. Internships at other corporate research labs or other universities could be extended.

  21. Jocelyn Says:

    Oh, by the way, we do have a tech report database, although it’s just a public ftp and not something as nice like arXiv.org

    ftp://ftp.cs.utoronto.ca/pub/reports/


  22. @ Golnaz:
    “…but I have got this feeling and impression that Canadian universities tend to hire from American universities graduates, and American universities that we can get to are remotely-located schools in the middle of nowhere that just started growing. These are totally based on my personal observations and judgments over the last three years I have been here and only limited to SE lab: our graduates do not really get hired even in the middle-ranked schools”

    Yes, we do great research and we’re very comparable to top US universities. Look at papers in top conferences your field and you’ll probably will NOT find that they’re all from Berkley, Stanford and CMU, But, in my opinion, the problem is two-fold:
    1) Compared to US universities, we suck at advertising. I mean UofT doesn’t appear in Hollywood movies too often.

    2) The second factor, which I think is the dominant one, is a simple supply/demand question. How many good schools are there and how many mediocre/bottom-tier schools are there? The difference is huge. So not all great researchers will end up at MIT. Yes, you may have an advantage attending a top US school because of factor #1 above. But take a look at Edmund Clarke’s former PhD student page – they graduated from CMU under the supervision of a Turing award winner:
    http://www.cs.cmu.edu/%7Eemc/oldstudents.html
    And yeah, a good number of them found themselves in the middle of nowhere universities. (hope this makes the fact that there’s no light at the end of the tunnel more justifiable)

  23. Dustin Says:

    So, I really really hate talks. I think I’ve become desensitized in the past year. Fundamentally, talks represent OLD RESEARCH. The more professional you make a talk and the more you prepare it, the less relevant it is. The point of research is to look forward and if we really want to drive inter-department collaboration, we should be discussing ideas we feel really tentative and unsure about. DGP has weekly meetings where one person present what they’re working on, but half of the talk is the audience proposing or dissecting ideas. This is a very valuable experience, although I feel the speaker gets more out of it than the audience some times.

    I am intrigued, but need to read up more on “unconferences”,
    http://en.wikipedia.org/wiki/Unconference
    see also:
    http://en.wikipedia.org/wiki/BarCamp
    such as foocamp. If we as a group of very intelligent people can’t come up with something we find interesting to talk about, then we don’t deserve to have any sort of conference. The bonus for this is that, other than logistics, this doesn’t require much academic preparation, which is a major argument against other “talks”. Ideally, this could take place over a weekend, or even a single weekday. It is certainly riskier, because “deliverables” are harder to define, and whoever moderates it will really have to know their stuff.

    Regarding internships, I don’t see why we have to facilitate grad internships more than we do. I also see it as a very fuzzy line from when classes “end until research “starts”. I started research during classes, and went away to an internship Jan – April (went on a “leave of absence”). I had to do a tiny amount of footwork, but this was totally easy.

    Also, stop saying “University of Toronto is a world-class school”; it de-values us. I refer to a quote by Margaret Thatcher:
    “Being powerful is like being a lady. If you have to tell people you are, you aren’t.”

    Ratings of the ideas:
    1a- 1b– (I am for these as long as they de-emphasize talks)
    2a- (Don’t see the value of this. Research-wise, I am more interested in papers with similar topics at other institutions than papers of un-related topics of this one)
    2b++ (Currently I was just planning to add my paper to my personal website. I would totally like this)
    2c+ 2d+ (Gives me a directory of people to contact)
    2e++ (I’m on this and I’m not even in SE. Everyone email Michael now!)
    2f- (At least for myself, any code I write is at or below research grade. I do not consider formalizing the process of spreading code to be useful)

    3a++

    4a++ (We don’t need the department to do this AT ALL. How expensive are whiteboards? I have one hanging around my house I will be bringing in and affixing to a wall in the near future. Microsoft Research has most of it’s interior walls made of frosted glass, and there were sketches EVERYWHERE. It was awesome. Some were whimsical, but most were serious. I can see how some would argue that this would reduce the “professionalism” of the department, but, honestly, grow up. The hallways of Bahen and SE are dead, dead, dead. Just stand in a hallway for a few minutes and listen? Can you hear the research happening? No. Argh, throw up some whiteboards!)

    4b+ (Personally, I think the correlation between a good undergrad program and a good grad program is kind of weak. However, I suppose reputation bleeds from either side. I agree will all your reasons for shorter co-ops. You get more out of 3+ short, different projects than 1 long project, especially if you are less experienced.

    @aws 1) Actually UofT appears in movies all the time, except never as the “University of Toronto”. I wonder if you could legally require them to call it by the proper name…

  24. anonymous Says:

    Step 1: Have big idea.
    Step 2: Gather graduate student focus groups to discuss ideas.
    Step 3: In the groups, tell them that we are going to execute your idea. Ignore other ideas.
    Step 4: Send out email thanking graduate students for their participation, and telling them that the result of the focus groups was your big idea.

  25. Sherry Says:

    1.a ++
    like the idea of making it more accessible
    like the minimum public speaking ability requirement

    1.b ++
    most DCS members can do much more than CS. But we never know about it unless they play on the stairwell like i do 🙂

    All of 2: ++
    Most of the work by DCS members should probably be listed somewhere online already. However, we only know about those works we heard about or stumbled across while surfing. Having them all together would facilitate and encourage the reading of other works. It would also be simply much more impressive to have this huge archive list 😉 easy way to promote our successes.

    3 +
    sure, why not

    4.a +

    4.b +

  26. Henig Says:

    Just wanted to post some comments from an undergrad point of view. So I may misinterpret the lifestyle of a grad student but I will still give it a shot.

    Here we go:

    1.(a+b) I think that you should keep the momentum of the weekly talks. I mean that instead of having weekly talks and a monthly TED talk, combine both to have a three regular talks followed by one TED talk. Organizing a schedule that includes one interesting weekly talk is much more attractive.

    2.(* except of f) Sounds good, and sounds like a full time job 🙂 Good luck.

    2.(f) Centralizing all the source code can be hard to maintain and organize effectively. I think you can rely on people to put code online if they want to.

    3. What about booking squash, tennis, etc… and organizing students to play together on the same time?

    4.(a) Optimally yes, but you will definitely affect the length of the conversation and it’s quality. I think the benefit does not worth the trade-off. The solution can be to have whiteboards in both places and let students choose depends on the conversation’s subject.

    4.(b) Undergrad students institutional support after first year will be inefficient since the students will not acquire enough “tools” yet. What about after Second and Third years?

    The PEY program is mostly not enough self-contributing for the students. Students take a year off from school in order to work on basic assignments that will not completely reflect their real life in industry after graduation. Some students will actually get to work on interesting projects and learn a lot from it, but they are a small portion of the total number of students that attend the PEY program.

    I believe that if students want to get a taste of industry before graduating they can find a part time job and work during summers. From an Economics point of view, the opportunity cost of students to attend PEY is their consecutive year after graduation. So students are actually letting go on more than $15,000 on average. Does it worth it? Are the new tools that they are acquiring worth the cost? And what about students GPA? A lot of students coming back from PEY are experiencing a decrease in grades because their Psychological state has now changed.

    The interview process for PEY can be done independent of the program. I think that it is the responsibility of the DCS to give the students the opportunity to get skills they need. I am a big +1 for interview workshops at the department and educate students about how to deal with stressful situations. (Maybe simulate interviews at the department by industry employees?)

    That’s it, hope it helps!

  27. abayomi Says:

    Aran, I think it is fantastic that you made the effort to put your ideas in writing and engage others in discussion about improving the department. I am new, so I don’t know if this is the norm. Nonetheless, it is a truly admirable effort.

    All the ideas put forward sound as though they will be helpful to someone, not me necessarily, as different people respond to and are helped by different things.

    I would support all of the ideas put forward by everyone, since someone thinks it would help them. I don’t feel passionate about any though, but I would help organise any of them with a minimum time commitment. When I do become passionate about an idea I will let you know.

  28. Marsha Says:

    Guys, lots of excellent ideas. I just saw this page and will study it in more detail, although if Aran were to do a bit of a summary of the feedback for his propose, it would really help us. And to second Aws’s observation about where students go: like Sagar Chaki, our own Arie Gurfinkel is in SEI at CMU. And like Alex Groce, our Michaela Bobaru is at JPL. These are excellent research labs that happen to hire formal methods grads. We have Shiva and Mehrdad go to Simula (either of them had academic offers in good places), Ou Wei return to China to become an academic, etc. Maybe our scale is a bit smaller than that of Ed Clarke’s but the quality of the placement is about the same!

    I do agree with you that we do not place our grads to top academic places in North America. This subject got raised several times and is being discussed on every level.

  29. Old Guy with Beard Says:

    Since 2004 I have repeatedly proposed that the department provide something very much like 2b. Everyone liked the idea, but no one was willing to expend any effort to implement it.

    I think it would be easy to implement using open source off the shelf software that I have identified. I was told that we couldn’t even ask grad students to volunteer to implement it due to union regulations about student work.

    My design looked like this
    – A prominent subpage of the DCS webpage called
    “Publications by our students”
    – A mechanism to allow grad students to add their
    publications to the page, either links or PDFs.
    – A mechanism to allow grad students to provide
    links to their home web page.
    – A mechanism for publication search by area
    (e.g. SE, KR …), by title or by author.


  30. I think I mentioned this before, but that’s already been mostly done for the kr department, and open sourced for any other group (inside U of T or otherwise) to use:

    http://www.cs.toronto.edu/kr/
    http://www.cs.toronto.edu/kr/publications
    http://www.bibbase.org/
    —> Author is a recent U of T grad student.

  31. Jorge Says:

    “Old Guy with Beard”: OK, you can’t ask grad students to volunteer to implement this, but could grad students volunteer on their own and get this in to the DCS website?

  32. Old Guy with Beard Says:

    I’d really like to say yes, but practically the answer is proabably no. (It’s not my decsion in
    any event.)
    – the underlying structure of the DCS web pages is quite complicated, it would take a lot of learning and work to craft something that fit in smoothly.
    – the proprietors of the DCS web would (reasonably)
    be nervious about 3rd party contributions.
    – the provision for allowing students to post
    things would probably require interaction with
    the cslab login/password mechanisim which would
    pose security risks if done by a 3rd party.
    – even if students volunteered without being asked
    the union might be annoyed, at least if the
    student’s work was incorporated into an
    official DCS web page. (this is my opinion,
    Chris Sparks can give you the official answer).
    Of course students can do anything they want
    in other webspaces (cf this blog).


Comments are closed.

%d bloggers like this: