From Newsgroup: alt.config

I don't know where to ask these questions. If this is the wrong place
please direct me to the proper resource.

How many active Usenet servers are currently in operation? Or how would
I find out? Is there a online list?

What online resources provide comprehensive and up-to-date Usenet
statistics and group lists?

--
Seamus
--- Synchronet 3.21d-Linux NewsLink 1.2

From Newsgroup: alt.config

On Fri, 8 Oct 2021 01:11:27 -0500, Seamus Hinkle wrote:

I don't know where to ask these questions. If this is the wrong place
please direct me to the proper resource.

How many active Usenet servers are currently in operation? Or how would
I find out? Is there a online list?

What online resources provide comprehensive and up-to-date Usenet
statistics and group lists?

First, you might want to try news.software.nntp. These are folks that
run Usenet servers. alt.config newsgroup is for getting feedback on
creating new alt.* newsgroups.

Also...

Here's the top 1000 Usenet servers:
http://top1000.anthologeek.net/

Here's a map of the Usenet from last year: http://usenet.ee/maps/graph3-n-20200402.png

--- Synchronet 3.21d-Linux NewsLink 1.2

From Newsgroup: alt.config

On 10/8/21 6:08 AM, Jason Evans wrote:

Here's the top 1000 Usenet servers:

Even this is predicated on a Usenet server submitting data to the Top
1000 list.

I'm aware of a number of private news servers that participate in Usenet
that don't submit to Top 1000.

I would not be surprised if it's impossible to accurately list all
Usenet servers.
--
Grant. . . .
unix || die
--- Synchronet 3.21d-Linux NewsLink 1.2

From Newsgroup: alt.config

Grant Taylor <gtaylor@tnetconsulting.net> wrote:

On 10/8/21 6:08 AM, Jason Evans wrote:

Here's the top 1000 Usenet servers:

Even this is predicated on a Usenet server submitting data to the Top
1000 list.

I'm aware of a number of private news servers that participate in Usenet that don't submit to Top 1000.

I would not be surprised if it's impossible to accurately list all
Usenet servers.

I agree with it being impossible to list all usenet servers but not participating with top1000 doesn't exclude them from being counted.

If they have no outbound feeds where posted articles on their server just
stay local, ok, that's possible.

But if they have any outbound servers they feed articles to, eventually
their path will be detected and then counted. I guess it depends on what you mean by "private server". But if it is a truly private server (no outbound feeds), does that really count as a usenet server?

I mean there are many weird entries on that list:

191 0.089544 iskon
200 0.077523 news
215 0.049966 adore2
234 0.032693 spln

and dozens of others. I really think none of those are really servers, some kind of article injection reference (end user?) into a server. Still ends up
in the article path so it is still counted I guess.

Plus out of the 1000 shown, 215 of them are *.fidonet.org.ru.

Same thing, I doubt all of those are servers but posts from those nodes are still counted.

-bruce
bje@ripco.com



--- Synchronet 3.21d-Linux NewsLink 1.2

From Newsgroup: alt.config

On 10/8/21 12:40 PM, bje@ripco.com wrote:

I agree with it being impossible to list all usenet servers but not participating with top1000 doesn't exclude them from being counted.

Agreed. But it does move thee task of enumerating ~> counting Usenet
servers from a simpler lookup type problem to a more complex header
analysis type problem.

If they have no outbound feeds where posted articles on their server just stay local, ok, that's possible.

;-)

But if they have any outbound servers they feed articles to, eventually
their path will be detected and then counted.

Maybe. Maybe not.

Consider, if you will, three parties / Usenet servers; A, B, and C.
Where B is the source of connecting A and C; A---B---C, and B is the
main connection to Usenet. B could have things configured to allow A
and C to share all articles with each other while not sharing their
articles with Usenet at large. Thus users of A and C could comment on
Usenet articles but their comments would not go out to Usenet at large.
There are probably a number of ways to do this, distribution headers and
/ or group filtering comes to mind.

I guess it depends on what you mean by "private server". But if it
is a truly private server (no outbound feeds), does that really count
as a usenet server?

Does a server need to /send/ articles /to/ Usenet to participate in
Usenet? Is it not sufficient to /receive/ articles /from/ Usenet and
have users read said articles? }:-)
--
Grant. . . .
unix || die
--- Synchronet 3.21d-Linux NewsLink 1.2

From Newsgroup: alt.config

Grant Taylor <gtaylor@tnetconsulting.net> wrote:

But if they have any outbound servers they feed articles to, eventually
their path will be detected and then counted.

Maybe. Maybe not.

It's more maybe than maybe not.

Like I pointed out, those entries for fidonet.org.ru in the top1000 list, I really doubt all of them (305?) are using or sending reports to anthologeek.net, those probably are single messages sent to fidonet and somewhere down the line is ending up being counted as a server.

Does a server need to /send/ articles /to/ Usenet to participate in
Usenet? Is it not sufficient to /receive/ articles /from/ Usenet and
have users read said articles? }:-)

I dunno.

If the pope shits in the woods and there is no one around to smell it, does
it still have an odor?

That top1000 site, when we first started to feed our stats into it, I was
under the impression it was for speed of articles posted. So if one system
got an article posted from somewhere in the world in 3 seconds and another system in 10 seconds, the first system got a higher rank.

That made sense back in the days UUCP was still common (or even bnews or
cnews) because some articles could take hours/days/weeks to show up.

Now with most systems on nntp, not really sure what the point is measuring
all that anymore.

I remember a while back (late 90's, early 2000's?) there was one system that constantly was in the top 5 if not in the #1 position and they were running
a pc using Windows something. When asked how they did it, the answer was
fairly simple, they had 100's of feeds and a shitload of ram.

So that top1000 site, besides being handy to see who is left, doesn't seem
to have any real world use anymore.

-bruce
bje@ripco.com
--- Synchronet 3.21d-Linux NewsLink 1.2

From Newsgroup: alt.config

On 10/10/21 7:02 AM, bje@ripco.com wrote:

If the pope shits in the woods and there is no one around to smell it,
does it still have an odor?

Almost certainly. The pope's digestive system isn't going to do
anything differently based on if there are people around or not.

That top1000 site, when we first started to feed our stats into it, I
was under the impression it was for speed of articles posted. So if one system got an article posted from somewhere in the world in 3 seconds
and another system in 10 seconds, the first system got a higher rank.

That's contrary to my understanding.

My understanding is that Top1000 is about the number / volume of
articles, and decidedly not about speed.

That made sense back in the days UUCP was still common (or even bnews
or cnews) because some articles could take hours/days/weeks to show up.

I haven't seen anything about stats to Top1000 being about /time/.

There's no time tracking as messages flow through Usenet servers like
there are in email Received: headers. Even if there were, you have the inconsistency of clocks to content with.

Thankfully, the number of messages that pass through a given server is independent of time.

Now with most systems on nntp, not really sure what the point is
measuring all that anymore.

I still see plenty of point in measuring the number / volume of
articles. If for nothing else, choosing better connected systems to
peer with.

I remember a while back (late 90's, early 2000's?) there was one system
that constantly was in the top 5 if not in the #1 position and they
were running a pc using Windows something. When asked how they did it,
the answer was fairly simple, they had 100's of feeds and a shitload
of ram.

And that makes perfect sense, a very large number of peers would
probably mean that it would get messages to the largest number of peers sooner. Thus the peer is more likely to have not seen a given message
and accept it from the extremely well peered server.

So that top1000 site, besides being handy to see who is left, doesn't
seem to have any real world use anymore.

I disagree. It's an indication of the number / volume of articles.
Time is incidental / after the fact.
--
Grant. . . .
unix || die
--- Synchronet 3.21d-Linux NewsLink 1.2

From Newsgroup: alt.config

Hi,

Am 10.10.2021 um 18:47 schrieb Grant Taylor:

On 10/10/21 7:02 AM, bje@ripco.com wrote:

If the pope shits in the woods and there is no one around to smell it,
does it still have an odor?

Almost certainly.-a The pope's digestive system isn't going to do
anything differently based on if there are people around or not.

That top1000 site, when we first started to feed our stats into it, I
was under the impression it was for speed of articles posted. So if
one system got an article posted from somewhere in the world in 3
seconds and another system in 10 seconds, the first system got a
higher rank.

That's contrary to my understanding.

My understanding is that Top1000 is about the number / volume of
articles, and decidedly not about speed.

That made sense back in the days UUCP was still common (or even bnews
or cnews) because some articles could take hours/days/weeks to show up.

I haven't seen anything about stats to Top1000 being about /time/.

There's no time tracking as messages flow through Usenet servers like
there are in email Received: headers.-a Even if there were, you have the inconsistency of clocks to content with.

Thankfully, the number of messages that pass through a given server is independent of time.

I guess time is an implicit factor. An article is accepted by a given
server from the first peer it offers it, from all other peers it is
refused. That system that won the race ends up in the path header, all
others do not. I see a lot more articles from a peer in the same
datacenter then from a peer on the other side of the planet.

Other factors are how well connected a system is (number of peers), and
how many articles are posted on that server, because that will make the
system show up in more path headers.

Cheers
Martin
--- Synchronet 3.21d-Linux NewsLink 1.2