On 2015-02-06 12:30, Mark Brader wrote:
> Björn Lundin:
>>> The cyclic still applies. Saying what year it was does not
>>> change anything, at least not in my understanding.
>>>
>>> If correct date is March 12 2012 then
>>> Dec 30 2011 is closer than Nov 5 2012
>
> Erland Sommarskog:
>> Yeah, but since you have given the year, an answer of Dec 30, would
>> imply an answer of 2012-12-30, which is further away.
>
> No, he's saying that if the correct date was 2012-03-12 then an answer
> of Dec 30 will be taken as 2011-12-30 *even though* we know that the
> year is 2012.
Yes, because it is closer in time of the event.
I do not care about the year, just the num days off.
(with the exception of counting feb 29 or not)
So, no guess can be off by more than 183 days (366/2)
Dec30 Mar12 Dec30
---- T1 ---- Tc -----T2----
-2011---|-----2012------|----2013---
It is supposed to be timelines above,
proportional font is needed to see correctly.
Tc=Correct date=Mar 12, 2012
I give year 2012
someone answers Dec 30.
Then T1 is taken as the first occurrence of
a yearless date with the answer (as date), that occurs _earlier_
than the correct date, going from the correct date to earlier dates.
That is, Dec 30, 2011
T2 is the first occurrence of a yearless date
with the answer (as date) that occurs _later_
than the correct date, going from the correct date to later dates.
That is, Dec 30, 2012
T1 and T2 is on each side in time of Tc
Tc-T1 = Mar 12, 2012 - Dec 30,2011 = 73 days
T2-Tc = Dec 30,2012 - Mar 12, 2012 = 293 days
Eventhough 2012 is given, a yearless answer of Dec 30
yields 73 off - NOT 293 days.
Given year has no impact on answers, nor on days off.
However, as Erland said, it might _imply_ one year or another,
but it really does not imply anything.
Or should at least not be interpreted as important in this knock out
--
Björn