Re: Google Group scistatmath
- From: David Winsemius <doe_snot@xxxxxxxxxxx>
- Date: Mon, 06 Nov 2006 09:35:14 -0600
"Reef Fish" <large_nassua_grouper@xxxxxxxxx> wrote in
news:1162793583.067876.157030@xxxxxxxxxxxxxxxxxxxxxxxxxxxx:
David Winsemius wrote:
Richard Ulrich <Rich.Ulrich@xxxxxxxxxxx> wrote in
news:1n4tk2t9jki6ufvs7vrkjv620u0b4otgn8@xxxxxxx:
On 4 Nov 2006 21:23:58 -0800, "Reef Fish"
<large_nassua_grouper@xxxxxxxxx> wrote:
[snip]
When I did a search of the STATISTICS about the two groups at
present, I found that there are not 18 subcribers to scistatmath,
while there are 1926 subscribers to sci.stat.math.
That 1292 suggests that there *are* still 100 lurkers who
never post, for every poster in a month, since it reflects
only the minority who read through Google-groups.
That estimate is flawed. A bit like using telephones as sampling
strategy in 1932. Many people read newgroups using "real" newsclients
drawing from "real" newservers. They would not be in the Google
counts.
I seriously doubt this theory of yours. Google must have the
subscriber count from the newsgroup itself.
Yet another expression of your newbie status. Because of your lack of
clue to how newsgroups are organized and distributed, you imagine Google
to be the center of the USENET universe. What do you imagine to be the
physical basis for this construct you are labeling, "the newsgroup
itself"? Where do you imagine this single entity, called "the newsgroup",
exists?
I can't imagine 1296 reads sci.stat.math via Google. From the ratings
in Google (star) on the threads, I can always count on two ratings of
1 (the lowest): 1 from Richard Ulrich, and one from Afonso. :-)
There are seldom more than 3 or 4 ratings on posts. If there are
anywhere near even 100 of the 1296 who subscribe via Google, there
would be many more who would rate the postings in the threads. I am
getting accustomed to the rating of "poor" by 1 user no matter what I
post. :-)
You are again conflating the reality of USENET with the epiphenomenon of
Google ratings. There is no such thing as "Google stars" in the bare
reality of USENET. Frivolous window dressing. OK, everyone, raise your
hand if you bother to rate Google postings.
This latest error on your part about USENET is akin to your repeated
error in citing the posting times displayed by Google as logically
related to when a posting appeared on any of the many other newservers
around the world. Posting time (at newsserver A) is not the same as
arrival time (at newsserver B) in USENET.
Another repeated error, driven by your lack of USENET clue, is expressed
when you complain about people using the "no-archive" marker in their
postings. All of those postings can be read by those of us using non-
Google newsreaders that have sufficient retention. Just because Google
removes it from its files does not make a posting disappear from USENET
servers around the world. At the moment Giganews' retention goes back to
mid-2003.
They would not necessarily even be countable by the newservers.
Subscription is an action performed at the client level.
But they can share with Google.
Exactly how would such count "sharing" occur? On what specification or
RFC are you basing your counter hypothesis. Can you specify an NNTP
parameter or protocol that would support such "sharing" among
newsservers? Suggested search term: "RFC 1036". Do not be misled when you
read that newsservers "subscribe" to feeds. That is logically distinct
from what happens when I subscribe to a group on any of my news-clients.
Or you could try using a real newsclient, unplug your cable (yeah, so
last century), then suscribe. Assuming the client already has a list of
available newsgroups, you should see no complaint from the client. There
I did it...cable out... now I am also subscribed to 0.test. No software
error message. Cable back in.
I think the multiple counting of the same subscriber is the big
inflated factor. Just look at the TOP posters. Richard Ulrich had
several different addresses listed in the all time list. The one
listing me as the all time top poster has not changed the count for
the last 1500 posts or so of mine, because that account had been
sabbotaged by a group of morons in rec.travel.travel who managed to
have Google bar TWO of my posting iDs by their repeated false abuse
reports.
Not everyone considers posting volume to be a figure-of-merit. I was
certainly not claiming that Google counts its own traffic correctly,
merely that users on other newservers could not be "seen", because only
news postings are shared among servers, not client "subscription"
statistics. And that is because "subscription" in the non-Google world is
a client-side event. There have been repeated guesses at the lurker to
poster ratio in various groups but no evidence based solution has
emerged.
I am sure I am counted at least 4 times by Google in its 1296 number.
Probably so have you.
I doubt that Google will share their methods with you. For all of its
virtues, Google is not "open source." You could be correct that I am
Google-counted multiply. Over the years I have posted to USENET through a
variety of ISP's as well as through the Google interface from different
machines.
--
David Winsemius
.
- Follow-Ups:
- David Winsemius lectures on USENET, Google, etc.
- From: Reef Fish
- David Winsemius lectures on USENET, Google, etc.
- References:
- Re: Google Group scistatmath
- From: Reef Fish
- Re: Google Group scistatmath
- From: Richard Ulrich
- Re: Google Group scistatmath
- From: David Winsemius
- Re: Google Group scistatmath
- From: Reef Fish
- Re: Google Group scistatmath
- Prev by Date: Re: percentiles estimation
- Next by Date: Re: likelihood estimation
- Previous by thread: Re: Google Group scistatmath
- Next by thread: David Winsemius lectures on USENET, Google, etc.
- Index(es):
Relevant Pages
|
