Tuesday, November 18, 2008

STOC notes

It looks like we ended up with about 329 submissions. (Hopefully that won't change much at this point.) That's between 40-50 papers per PC member at 3 reviews per paper.

Thanks to everyone who withdrew the papers they weren't submitting. Otherwise, I had to go through and withdraw them manually myself.

Thanks to Shai Halevi for continuing to help me with the reviewing system.

Yes, I did get mail from about 10 people who hadn't known they'd need to file an abstract the week before, and I accommodated them. I think this should become a standard and everyone should get used to it. Again, keep in mind there's 40-50 papers per PC member; anything that makes their work easier is a good thing. (And Shai's interface for choosing preferences lets you see a paper's abstract pop up with a mouse rollover, so it's really nice to have abstracts early!)

Assuming things continue to go well, expect e-mail from PC members asking for you to handle a subreview before you head off for Thanksgiving....


Anonymous said...

Wow! Is this a record number of submissions?

What fraction of abstracts turned into papers?

Anonymous said...

Hi Michael,

Out of scientific curiosity, could you comment on how long it's taking to assign the papers since the paper submission this time round, vs how long it'd normally have taken if there were no early submission of abstracts?

Michael Mitzenmacher said...

Since I haven't been chair before, I can't say how long is normal.

I can say that it's much nicer to have abstracts when picking papers. I asked the PC to look at the abstracts in advance. The deadline was Monday midnight; I hope to have assignments done by tomorrow midnight, if the PC has everything in.

Anonymous said...

two small comments:
(1) I don't like submitting an abstract before the full paper because, even though the technical parts might be worked out, sometimes it's hard to know exactly how to "sell the paper" until it's fully written. In other words, I like writing the abstract last. I guess that means I should write the entire paper earlier and not submit at 11:59 pm!
(2) If you google "STOC 2009", the first hit is you blog bost titled "impending doom".

Anonymous said...

Unrelated comment: I hope you will consider dropping paper proceedings this time.

My reasons are not cost or weight but scientific - often authors don't write full version and the best version available of a paper is the conference camera-ready version.Paper proceedings force this version to be hard to read and ugly looking 12 page double-column.
If instead it would be 20 single column pages we'd have much easier to read papers.

You can use a service like http://www.lulu.com/ (or Kinkos) to accomodate people and libraries who still want printed and bound full paper proceedings.

Electronic versions are also useful if people that registered to the conference in advance could have access to a website with the actual papers a week or two in advance, we could think of the talks and print out for us whatever papers we want to have available during the conference. (That option can be implemented even without dropping paper proceedings.)

Boaz Barak

Anonymous said...

SODA'09 Proceedings can be viewed already on the SODA web page, in electronic form. Note that this raises the possibility of reducing the number of months between the submission and the conference.

Unknown said...

Dear Michael, thanks for keeping everyone posted. I'd like to volunteer for a very specific task.

When you post the accepted papers, it seems oh so likely that that their titles (and hopefully abstracts) will be posted on the web in plain text format. If you want someone to put them in easy to read html format, please let me know.

I did this for SODA for my own benefit and I would be super happy to produce (and run) a script that anyone could use. (i.e. go from this to this, but probably nicer looking since it'd be for everyone)

When the time comes, if you're interested, please feel free to email me eerac at cs.brown.edu. Also if you have any interest in setting up a blog for the conference in which each paper corresponds to a post (allowing people to post comments), I'd be happy to set that up too.

Anonymous said...

need to file an abstract the week before... I think this should become a standard and everyone should get used to it.

I disagree.

As a member of this STOC PC, I personally found no particular benefit in seeing the abstracts early. I briefly looked at the list of abstracts but the knowledge that many of the abstracts would not turn into final papers made it difficult to concentrate on them. (15-20% of abstracts either did not turn into submitted papers or were duplicates and this is typical based on the SODA experience last year.) Why should I waste effort deciding preferences for them? Moreover, a brief glance at the final paper was often a significant aid in choosing my preferences.

Large conferences with formal multilevel PCs may need the separate deadlines to coordinate with many outside members but with our smallish single-level PC, it doesn't seem to provide much benefit. If we need an extra day or two we should just build it in for the final version deadline but I don't see much of a need for that either - a couple of days don't make a difference in an 8+ week reviewing cycle.

The separate abstract deadline may have been good for submitting authors in that it forced them to start work early. On the other hand it means an extra week between the work being done and the time when it will be presented.

Michael Mitzenmacher said...


We'll just have to agree to disagree.

I can understand if YOU, personally, didn't find it useful to have the abstracts early. I, as PC chair, most certainly did. I can't speak for the other PC members, but I know at least some made use of having the abstracts early and I'll assume they appreciated it. There is some overhead in that not all abstracts become papers, agreed, but the gain is flexibility in timing. You can choose not to take advantage, but now you at least have an option.

For me, the tradeoff was in making the deadline MUCH earlier -- possibly Nov 10 itself, when the abstracts were due -- to make sure there was ample time available, or use this two-pronged approach. (Your estimate of an extra day or two in the schedule doesn't fly with me -- since I have kids, making papers due Nov 14 and doing this over the weekend just wouldn't have worked for me, and possibly other PC members.)

It does remind me of an issue that comes up every year with my undergrads, about the due date of the last assignment. I tell them I'll make it as late as possible to give them maximum flexibility for when they can do it. But there's always a subset of people who want the deadline earlier -- they know they'll just do it at the last minute anyway, regardless of when it's due, so giving them flexibility doesn't help them and hurts them in cases when they have exams coming up right after (that they're also not going to study for until the last minute) or just bothers them to have it over their head longer. Their psychology trumps their optimal strategy.

For me, timing flexibility is key. I think having papers due later and abstracts due earlier is a fine compromise between competing needs.

David Andersen said...

Hey, Michael - I noted that you said you had to manually withdraw incomplete submissions. I haven't used websubrev, so I can't comment on the relative strengths, but having just finished chairing HotNets using the HotCRP software, I have to give it a strong plug -- it's awesome, and there were very few painful manual steps like the one you mentioned. I'd give it a shot next time you chair something and have the (opportunity, painful requirement to) specify which conf. mgmt software to use.

And Eddie's fantastic about bug reports and feedback.


(p.s. you can put me in the "I like abstracts" camp as both a PC member and a chair.)

Michael Mitzenmacher said...

David --

I'm using HotCRP as a PC member for NSDI, and I'm enjoying it too. (I might have used it, but I couldn't convince anyone to host it for me.)

I think Shai's software also has a lot of nice features, and some combination of them (or the ideas) would be great.

And thanks for offering some agreement on the "abstract" issue.