Alinghi, Bacon, & Continued Re: [NEC] 2.3: Power Laws, Weblogs, and Inequality

S. Alexander Jacobson alex@i2x.com
Tue, 4 Mar 2003 18:35:01 -0500 (Eastern Standard Time)


On Mon, 3 Mar 2003, Dave Long wrote:
> Many products and many media may be entirely
> random, without being uniformly distributed.

In fact, if rankings were entirely random, they
would have a Normal distribution thanks to the
Central Limit Theorem
http://stat.tamu.edu/stat30x/notes/node88.html

> We see correlations (power law correlations
> even) in random processes.

But now you are hyothesizing a random process.
The question here is how does this randomness
arise?  Are people flipping a coin?  Are they
typing random sets of characters into their
browsers?  If you want to claim any process here
at all you have to explain how it works (whether
random or not).

> If you'd like to
> claim talent alone drives the distribution,
> that's well and fine, and perhaps even true,
> but since a set of talented bloggers is more
> complex than one of identical bloggers, it'd
> be good to first demonstrate that it's worth
> the additional complexity:  can one make a
> better prediction of blog ranks by means of
> the additional detail of talent?

If we assume that blog preferences are correlated,
then we can predict blog rankings by analyzing
changes in blog preferences.

Given that people make blog recommendations all
the time, I think this assumption is not
unwarranted.

You are sayiig that this concept of preference
adds complexity.  I am saying that we have NO
other simpler story for predicting blog rank.

That current blog rank resembles prior blog rank
is not a prediction.

-Alex-



>
> Alinghi va a gonfie vele.
>
> :: :: ::
>
> Bacon, _Advancement of Learning_ (1605)
> > Surely, like as many substances in nature which are solid do putrify
> > and corrupt into worms; so it is the property of good and sound
> > knowledge to putrify and dissolve into a number of subtle, idle,
> > unwholesome, and, as I may term them, vermiculate questions ...
>
> There were six or seven millenia of
> urbane wits before Bacon, but he is
> one of the first of modern times to
> suggest that looking for keys might
> best be accomplished, not under the
> streetlight of holy books and their
> literature of logic and/or metaphor,
>
> > Well, obviously ["blessed are the cheesemakers" isn't] meant to be
> > taken literally; it refers to any manufacturers of dairy products.
>
> but in the dark of experimentation,
> even if it meant stubbing toes and
> temporarily groping blindly.*
>
> "Fire burns, here and in Persia" is
> small stuff next to grander stories
> that thinkers tell, but if it isn't
> ubique, semper, ab omnibus, it can
> at least be shown where- and when-
> ever one can find a small boy with
> a book of matches.
>
>   Foogilly Bargilly / Nicolas Bacon's son
>   Disputed disputing / Holding it moot
>   Knowledge (hence power) can / Be found inductively
>   Investigatively / Checking each root
>
> :: :: ::
>
> > I am not sure what you mean by showing A.  To me
> > it is obvious that consumers have correlated
> > preferences.
>
> But do blog preferences correlate in a
> path independent manner (readers agree
> on talent), or in a path dependent one
> (readers reflect historic preferences)?
>
> >               If they didn't, all products and all
> > media would be entirely random.
>
> This is where you lose me: we need to show,
> not just assert, that blogs aren't random.
> The obvious way to do so is to show they're
> predictable; when I tried that, I failed.
>
> Density is easy to believe in, because it
> easily predicts whether pool toys wind up
> at the top or the bottom, independently of
> the path they take from the release point.
> Is talent like density?
>
>
> -Dave
>
> :: :: ::
>
> * Bacon may not have realized quite how much
> this world has to teach, once it is asked: (_Novum Organum_, 1620)
> > 112. In the mean time, let no one be alarmed at the multitude of
> > particulars, but rather inclined to hope on that very account. For the
> > particular phenomena of the arts and nature are in reality but as a
> > handful, when compared with the fictions of the imagination, removed
> > and separated from the evidence of facts. The termination of our
> > method is clear, and I had almost said, near at hand; the other admits
> > of no termination, but only of infinite confusion. For men have
> > hitherto dwelt but little, or rather only slightly touched upon
> > experience, whilst they have wasted much time on theories and the
> > fictions of the imagination. If we had but any one who could actually
> > answer our interrogations of nature, the invention of all causes and
> > sciences would be the labour of but a few years.
>

___________________________________________________________________
S. Alexander Jacobson                   i2x Media
1-212-787-1914 voice                    1-603-288-1280 fax