Tuesday, May 10, 2011

Those Happy Samoans

Below is a post I wrote in 2009 but for some reason never made it to this blog. But I had better post it now because, as I found out via Tony Wirth, Samoa is moving across the international date line, 24 hours into the future. That should make Hawaii the last place on Earth to end the day, making conference deadlines an hour earlier.

So are the Samoans losing a hour to submit their papers or are they gaining twenty-three?

Update 12/28/11: Western Samoa is skipping this Friday to make the switch. American Samoa is not making the switch and will be the last place on Earth to end the day, an hour after Hawaii.

The deadline for ICALP 2009 was February 10th at 12:30 PM in Greece where the conference will be held. Did you sleep late and miss it? Why not have the deadline in late afternoon so people can keep working on their papers?

But in fact the deadline was Tuesday. At least while it was Tuesday anywhere in the world. So ICALP waits until the small country of Samoa approaches midnight before making its papers due. So the Greeks get all morning Wednesday. The New Zealanders have all day Wednesday until 11:30 PM. With Daylight Savings Time, the Kiwis are a full day ahead of the Samoans.

Nothing against the fine citizens of Samoa but it is not a hotbed of theoretical computer science. So it still seems strange to list the submission deadline as "February 10, 2009, (23:30 GMT-11)" in a time zone from which they will likely get no submissions instead of just giving the deadline as February 11, 2009 (10:30 GMT).

Not all conferences follow the Samoan rule. Complexity deadline is midnight Eastern on Friday. At EC we used Midnight GMT on Monday and we received a few complaints from people in the US. I just didn't want to have to be up at 4:30 AM Chicago time making sure the submission server wasn't crashing.

The NSF has a policy of submission of 5 PM in the submitter's time zone. So perhaps conferences could have a deadline of midnight in the place the submitter lives. This would give those Samoans the extra time they so richly deserve.


  1. Time-zone of the author sounds hard to enforce. Also, authors may purposely try to get co-authors from a place where the deadline is the latest. (Sonds like a good April Fools day post: Submissions from
    Samoa skyrocket!)

    Do those last minute corrections actually help that much?

  2. I actually think it makes sense to say "February 10, 2009, (23:30 GMT-11)" as deadline instead of "February 11, 2009 (10:30 GMT)": Then people think the deadline is on 10 February and finish their paper on that day instead of trying to finish it in the last minute of 11 February.

  3. IEEE 802.16 Working Group ballot deadlines are established as the end of day "AOE", for "Anywhere on Earth." This means that the deadline has not passed if, anywhere on earth, the deadline date has not yet passed.


    Some cs conferences explicitly point to this site for the deadline.

  4. > This means that the deadline has
    > not passed if, anywhere on earth, > the deadline date has not yet
    > passed.

    I maintain that it's Feb 1, 2003 in my office; does that count?

  5. The problem with the NSF deadline is that it's ambiguous as well as being hard to enforce. I would regularly end up submitting NSF material from other time zones (say while visiting my wife while long distance or friends) and I was never really clear about which time zone applied. Worse academics move frequently during grad/postdoc years so do you go by the address in the personal data section even if they've since moved and have mail forwarding? Is it sufficent to have your academic mail delivered to a friend in Samoa?

    And this doesn't even cover people who live in different time zones from where they work or who spend weekends in different places than weekdays (I used to regularly spend weekends in chicago while at Notre Dame).


    Frankly I prefer the way we do it in math. There is the "deadline" and then you wait a few days for stragglers since you need that time to read the on time submissions anyway. Then again in math it takes so long to type up good proofs we usually only require an abstract for a conference so if you don't make it in time for the proceedings it's not a huge deal.

    But even so why all the fuss about deadlines. Give people a bit of wiggle room and progressively increase your standards as submissions get later and later (a couple days and you evaluate it like any other submission at papers a couple months late get accepted only if they contain a formally verified proof deciding P = NP). You can always staple copies to the proceedings and you want the best conference and that means the best results.