221
edits
Changes
Jump to navigation
Jump to search
Line 1,767:
Line 1,767: + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
Talk:Essay:Best New Conservative Words (view source)
Revision as of 14:46, August 1, 2010
, 14:46, August 1, 2010→PERFECTION: 20-40-80-160 BY CENTURY
: Typos and counting errors are, of course, inevitable; your own comment above has an error in its last seven words. Errors can be found in the greatest of works, such as [[Bernard Riemann]]'s famous mathematical lecture. None of this undermines the value of Riemann's work ... or ours. The best new conservative words do double by century, and it would be nearly impossible to identify such a large number closely fitting that pattern unless the underlying pattern existed.--[[User:Aschlafly|Andy Schlafly]] 12:06, 28 July 2010 (EDT)
: Typos and counting errors are, of course, inevitable; your own comment above has an error in its last seven words. Errors can be found in the greatest of works, such as [[Bernard Riemann]]'s famous mathematical lecture. None of this undermines the value of Riemann's work ... or ours. The best new conservative words do double by century, and it would be nearly impossible to identify such a large number closely fitting that pattern unless the underlying pattern existed.--[[User:Aschlafly|Andy Schlafly]] 12:06, 28 July 2010 (EDT)
You are right: It´s not about typos - it's about the phenomenon I'll explain in the following section
[[User:RonLar|RonLar]] 10:46, 1 August 2010 (EDT)
== Too Good to be True ==
Imagine the hypothesis: <i>There are as many male writers as there are female
ones</i>.
To proof this hypothesis, round for round, a player '''A''' names a male author,
then a player '''B''' a female one. The round one player runs out of names - and
the other doesn't- the hypothesis is falsified. If both stop in the same round,
the hypothesis is true.
But what can be said about the validity of the hypothesis before this last
round? There are thousands and thousands of authors, what do we know when we
reach round 300?
Nothing. At least nothing about the ratio of female and male authors! In fact,
imagine the game with player '''A''' naming ''two'' male authors for each of
player '''B''''s female writer. After 300 rounds, this game is far from being
finished, but we can't conclude anything!
Unless this game is played to its end, it's just an exercise in futility.
Statistics try to find methods which allow to draw conclusions without exhausting the whole population1
So, what's another way to create data for validating or falsifying the
hypothesis? Let's think of player '''S''' who says that he will gather authors
at random - and that this list will beautifully exemplify the hypothesis.
And so it does: he gives a couple of hundred names and - lo and behold - exactly
half of them are males! He even goes a step further and says that each two
consecutive name would mirror the hypothesis.
When would you start to suspect that he is not an honest player? The probability
that a pair of authors consists from a man and a woman is 1/2. And in a random
list of names, you'll find many such pairs. But if '''S''' announces: the next
pair is gender-mixed, the one thereafter, too, and the next one, again - that's
like predicting three ''heads'' when tossing a coin three times. If he does it
ten times, you would be a fool not to think that something fishy is happening,
i.e., that the names are not taken from a random list of authors, but in fact
are chosen deliberately. His data is just ''too good to be true''!
Such a list has no more significance than the first couple of hundred names
generated by our players '''A''' and '''B'''.
Andy, I hope you see how this scenario applies to your method of generating
''best conservative new words''.
At least four times you announced the start of a new ''layer'', and each time,
you were able to complete it perfectly. Though the probability to find the
combination 1-2-4-8 for 15 words may be bigger than for each other combination,
it's at best 2.6508% . Doing this four times in a row yields odds of
1:2,000,000.
Again, your findings are just ''too good to be true'', therefore, they are not
believable, and don't support your claims made in [[Essay:Conservapedia's Law]].
[[User:RonLar|RonLar]] 10:46, 1 August 2010 (EDT)