Results 1 to 5 of 5

Thread: Any explaination for k=19249 vs 67607

  1. #1
    Moderator vjs's Avatar
    Join Date
    Apr 2004
    Location
    ARS DC forum
    Posts
    1,331

    Any explaination for k=19249 vs 67607

    I was wondering if anyone had a good explaination why k=19249 generally has less pending tests than k=67607.

    k=19249 always has ~1% less pending tests although the density for k=19249 is 20% greater than k=67607.

    As an example the density for 19249 is very close to k=22699 only 2% smaller yet k=22699 always has roughly 16% more tests than both 19249 and 67607.

    I just find it odd.

  2. #2
    I love 67607
    Join Date
    Dec 2002
    Location
    Istanbul
    Posts
    752
    do you know for how long this has been the case?

  3. #3
    Moderator vjs's Avatar
    Join Date
    Apr 2004
    Location
    ARS DC forum
    Posts
    1,331
    Quite a few weeks now I'd even hazard to say months.

    If you look in a little more detail at the assignments for n vs k

    http://www.seventeenorbust.com/stats....mhtml?k=67607
    http://www.seventeenorbust.com/stats....mhtml?k=19249

    I thought initally that there may simply be more secondpass assigned to 67607 than 19249. Incompleted tests etc...

    I believe today is one of the only days I have checked where k=67607 has less tests than 19249, 261 tests vs 262 tests.

    An oddity is all, there are a few other k's that have more than they should and other less of course. It's just strange that these two generally show the same trend.


    Something to discuss... don't nessarily think there is a problem.

  4. #4
    I love 67607
    Join Date
    Dec 2002
    Location
    Istanbul
    Posts
    752
    does anybody know the reason why there is so many pending secondpass tests? was it also created by a runaway client?

    there seems to be not much activity at secondpass, but way too many pending tests. and I guess this has been the case for a month or more.

    BTW, the discrepancy at 19249 is -16%, whereas all others are within +/- 5% range. This analysis is based on number k/n pairs left for 0-20m vs. number of pending tests.

  5. #5
    It is quite possible that it is a statistical effect.

    If you consider coin tossing, the probability for long leads is incredibly high.

    Some citations from Feller, An Introduction to Probability Theory and Its Applications:
    "In 20000 tossings it is about 88 times more probable that Peter leads in all 20000 trialas than that each player leads in 10000 trials. In general, the lead changes at such infrequent intervals that intuition is defied."
    Even if this is no fair coin tossing thing, it could be an explanation. the numbers of tests are not very high, and tests remain pending a long time, sometimes months.
    Unless serious statistical analysis is done, I wouldn't bother.
    Don't ask me to do this analysis, though.
    Yours, H.
    ___________________________________________________________________
    Sievers of all projects unite! You have nothing to lose but some PRP-residues.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •