Wednesday, February 16, 2011

If I tweeted here is what I would tweet

If I tweeted there is what I would tweet:
  1. There was an interesting blog post that responded to Aaron Sterling's Chemoinformatics Post. See here for this interesting discussion.
  2. See here for info on the Watson's performance on Jeopardy.
  3. Why did Watson bet $937 on the Final Jeopardy question yesterday? (Tuesday) Zero would make sense. The max amount to guarantee victory even if he got it wrong would make sense. A nice round number to look more human would make sense.
  4. The winner gets $1,000,000, second place gets $300,000, third gets $200,000. Ken Jennings and Brad Rutler have said they will give half of their winnings to charity. Watson said he would give all of his to charity. QUESTION: If you were on the show would you give 1/2 to charity? For Jennings and Rutler it isn't really hard to do since they already have lots of money from their prior Jeopardy appearances (though it is still admirable). But how about YOU?
  5. Officially the authors of the CCC papers will be notified on Friday Feb 18. I have been told that this is an approx both ways. Could be earlier, could be later.
  6. Some CCC notifications (all?) have been emailed. The list is not posted! At one time the quaint notion was that people should not find out if there paper got in or not by seeing someones telegraph message, website, or blog. Is this quaint-but-idiotic notion still in effect?
  7. I got a copy of the complete Knuth Vol 4 in the mail today. How long have we been waiting for this? Knuth held a contest to name what we now call NP-completeness for use in this book.
  8. Publishers have lists of people they send books to. As an editor for SIGACT NEWS I am on some of those lists. Today I got two copies of the exact same book. Both mailed to William Gasarch, SIGACT NEWS book review editor, Dept of CS, College Park MD, 20742. What was the book: Bijective Combinatorics.
  9. emailed me a subject heading of gift of infinite choices. Alas, the actual letter said you can choose from 250 million styles and designs..
  10. 12 days of Christmath (That is NOT misspelled.)
  11. Also this one (I give two pointers in case one goes away.)


  1. I would imagine Watson's betting strategy also takes into account it's confidence in the answer, which wasn't a round number.

  2. If you were on Twitter, you wouldn't have been able to use 3, 4, 5, 6, 7, 8, 9 because they exceed the rather twee and gimmicky 140 character limit. (Yes, I am on Twitter.) The others would depend on how you handled the URLs.

  3. How can 5 and 6 both be true?

  4. Watson wouldn't wager zero because the winnings from the game are to be added to those of the second game to determine the winner of the match.

    The betting strategy is described (sort of) here:

  5. I expect using odd dollar amounts when prompted was just a gimmick. It emphasizes how Watson thinks differently than us, with an implication that because he chooses more precise numbers, he's doing a better job than we would. The daily double where he chose another precise number elicited a laugh from the audience.

  6. The goal of this Jeopardy match is to have the most money after two games. They just finished the first one, so Watson is positioning himself for the next game. I assume he bet such a small amount both because he likes his chances so far, and because apparently isn't too strong on US cities. But betting $0 means he has zero confidence and/or sees zero need to increase his money in that round. Probably neither was true.

  7. Anon 3- I would have posted 5 and later 6.

    Anon `who is Watson'- click on the link I provided to find out.

    Andrew- That is one of the reason I don't tweet.

    My prediction about Watson: it will be the basis for question-answer systems; however, we will then have to ask computers questions in the form of an answer.

  8. It's only volume 4A, of what will, at the very least, be a four-part volume (4A to 4D). At the present rate of one part of a volume per thirty years, the entire "volume 4" will be published before the end of the century, but just barely

  9. Stars of TV show Stargate Universe supposedly found out it was canceled when fans starting tweeting their "sorry"s online after reading about it. The info didn't go out to the actual people involved in the show in a timely way with enough lead time if this is true. That's just wrong. Finding out your paper didn't get in from others telling you that your name isn't on some list posted would be wrong too. The gawkers can read the list after the authors have time to find out individually.

  10. Last Anon- fair point. Now here is the question of interest: you email
    the authors at time X about the status of their papers. When should you then post the list? I was suggesting
    X + 5 minutes which I suspect you think is too soon and you may be right. But now its your turn:
    X+ 24 hours? Or since we are computer scientists perhaps X+16 hours? I ask nonrhetorically and with respect- how much time is reasonble? Far less than it used to be, but how much less?

  11. I ask nonrhetorically and with respect- how much time is reasonable?

    For some conferences there are forced paper merges that must settle down before the list is sent out so there can be a much larger delay than 24 hours. I have never heard of that for CCC, though. If that isn't the case, then the next day seems fine.

  12. There was a built in minimum of $937 for final jeopardy. It was to prevent the humans from seeing a $0 bet on Tuesday, and taking advantage of the fact that they knew Watson was scared to guess on Wednesday.

  13. Andrew: I'm sure you meant: "If you were on Twitter, you wouldn't have been able to use 3, 4, 5, 6, 7, 8, 9 because they exceed the rather twee and gimmicky 140 characte"

  14. The 140 character limit was originally introduced for compatibility with the SMS protocol. I don't know why 140 characters and not 160, which is the SMS limit, but presume the 20 character difference is used by Twitter for its own data. It wasn't a gimmick.

  15. I would suggest x + 48 hours to 72 hours. That is enough time to check e-mail for most people in the field and also time for them to perhaps mention it to a few others personally (for example, co-authors not contacted directly).