bionet.* policy change

Dave Mack dmack at net.bio.net
Mon Aug 22 21:47:17 EST 1994


In article <33alfo$gs6 at news.ycc.yale.edu>,
Una Smith <una at doliolum.biology.yale.edu> wrote:
>
>Una Smith <una at doliolum.biology.yale.edu> wrote:
>
>>>[Bionet.*] newsgroups are all gated to mailing lists, so should
>>>you decide not to carry them at your site, due to lack of
>>>interest or the extremely topical nature of these groups,
>>>you won't be depriving any down-stream users of the chance
>>>to use these resources.
>
>
>Dave Mack <dmack at net.bio.net> wrote:
>
>>This inference is completely incorrect.
>
>My statement above is neither an inference nor incorrect.

Indeed? I leave it the readers to determine whether or not
you inferred anything.

Your statement implies that e-mail and Usenet are equally
effective mechanisms for giving "any down-stream users [of] the chance
to use these resources." I consider this to be incorrect for
the following reasons:

1) it presumes that the mailing lists will always be available. No
such guarantee exists. Please let me know if you are interested in
taking on the chore of running them when the BIOSCI contract expires.
The bionet newsgroups will continue regardless of the status of the
contract or most anything else;

2) it presumes that the information content of a message arriving via
e-mail is identical to the content of one transmitted via Usenet. This
is incorrect - the Message-ID of a Usenet article is globally unique
and can be used as a reference identifier, whereas the Message-ID of
an e-mail message is not and cannot;

3) it presumes that e-mail delivery is as reliable as Usenet article
delivery - it is not. It is possible to get redundant Usenet feeds
to increase the probability of receiving all posted articles to
some set of newsgroups. No such mechanism exists for e-mail. A
mail message either gets there via a selected route (usually
direct delivery via SMTP, sometimes via a number of UUCP hops,
and X.400 is anybody's guess) or it bounces. (There is undoubtedly
some exception to this very general statement,i.e., some MTA out
there which tries multiple paths before giving up, but I believe that
this correctly describes the normal case.) ;

4) it presumes that all of the Usenet delivery capabilities are
somehow reflected in the mailing list distributions - not so. There
is no realistic way to ensure that a Usenet message which is cross-posted
to multiple bionet groups will be seen by all the subscribers to those
list but seen only once by any particular subscriber. In other words,
either you get multiple e-mail copies of the messages depending on
which mailing lists you're on, or you can't be assured of getting the
message at all, even though it's cross-posted to a newsgroup whose
mailing list you subscribe to.

>>Yes, all of the bionet
>>newsgroups are gated to mailing lists. We do this as a service
>>to those who, for one reason or another, cannot get Usenet but
>>are able to get e-mail. We DO NOT guarantee that this service
>>will be provided in perpetuity or that it will be provided to
>>anyone who wants it.
>
>Dave, do you guarantee that bionet.* will be available to 
>anyone who wants it, via Usenet?  How can you, when Usenet
>doesn't belong to you, and isn't under your control?

Of course, I don't. Do you guarantee that anyone who wants
them can get the bionet groups via the mailing lists? You certainly
seem to be offering such a guarantee to the Usenet admins when
you tell them that, by refusing to carry the newsgroups, they
are not depriving downstream users. But you cannot legitimately
offer such a guarantee since you control neither the e-mail
transport systems involved nor the mailing lists.

What I do guarantee is that I will make the best effort I can
to provide the bionet hierarchy to any site that wants it.
Fiscal constraints make it impossible for me to do that through
any mechanism other than the Internet (i.e., I cannot offer
long-distance UUCP feeds.)

>If loss of support from BIOSCI is a real, forseeable risk,
>then for all our sakes, please start to find alternative
>hosts for those mailing lists.  I think you owe this to 
>your subscribers.

Mailing lists are an obsolete and ineffective way of transmitting
information through the Internet.

When and if the time comes when we can no longer run the mailing
lists, I will try to find other sites to house them. It is more
probable that we will first reduce the burden by eliminating
e-mail to sites that we know are getting the bionet groups via
Usenet, and by requiring sites that receive multiple
subscriptions to run exploders, either into e-mail or through
mail-to-news gateways. (Note that we are already doing this to
some extent - e-mail subscriptions are split between our site
and Daresbury in the UK; we handle the subscriptions destined
for the Americas and the Pacific Rim, they handle Europe, Central
Asia, and Africa.)

Currently, our average e-mail output on net.bio.net is slightly
over 30,000 messages per day. We typically get several hundred
bounced messages per day. It requires human intervention to deal
with those and it's an absurd waste of human resources.

>I agree with you, there are no guarantees here.  But getting
>Usenet on every computer that biologists use isn't necessary,
>because the mailing lists can be moved to other hosts, with
>no effort required from the subscribers.

All that is necessary is that they get someone to get Usenet on
one computer on the same network. NNRP-capable newsreaders are
available for most architectures.

>
>>We most strongly urge anyone who wants 
>>access to the bionet newsgroups to attempt to acquire them
>>through Usenet rather than email. In particular, we will attempt
>>to provide NNTP feeds of the bionet hierarchy to anyone who
>>wants one and is connected to the Internet. If your service
>>provider does not have access to the bionet groups, you should
>>have them contact me.
>
>Dave, are you willing to allow individuals at remote sites to
>read the bionet.* newsgroups from net.bio.net?
>
>Note, I do *not* mean a feed from your server to another server.
>Many people have to use e-mail to access bionet.* because they
>have no local Usenet server.

We already do - through gopher. All of the bionet groups are
archived on gopher.bio.net (aka net.bio.net) and new articles
are made available as soon as they arrive in the "current"
selection (i.e., to read bionet.general, you would gopher into
gopher.bio.net, select "BIOFORUM/" from the root menu, page down
to the last item and select "current".) I will admit that gopher
is not the most adept mechanism for news reading. We plan to make
all of this available through WWW Real Soon Now. Neither of these
is a particularly desirable mechanism for distributing Usenet news,
since each article has to be transmitted across the net each time
it is read, whereas NNTP transmission only requires that
each article be sent once to each local server.

Your question omits the more problematic end of things: posting.
Read-only access to Usenet is lame at best. The security problems
involved in running an open NNTP posting server are well known, so
that option is out. However, INN-1.4 offers an NNRP authentication
option which might offer some flimsy shred of security.

So, to answer your question: maybe. I need to look into what
it would take to build a subscription-based front end to NNRP.
I'll discuss this with Dave K when he returns from vacation.

Incidentally, the opinions I've offered here are my own - Dave K and
Alan Bleasby may not agree with me on all or any of them.



-- 
Dave Mack
Manager, Computer Facilities
Senior Systems Administrator
IntelliGenetics, Inc./BIOSCI Project
-- 
Dave Mack
Manager, Computer Facilities
Senior Systems Administrator
IntelliGenetics, Inc./BIOSCI Project



More information about the Bioforum mailing list