home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky news.groups:24819 news.admin.misc:881
- Newsgroups: news.groups,news.admin.misc
- Path: sparky!uunet!gatech!emory!nastar!phardie
- From: phardie@nastar.uucp (Pete Hardie)
- Subject: Re: PROPOSAL: Yes votes >= 200+2*No Votes
- Message-ID: <1992Dec21.221311.17787@nastar.uucp>
- Organization: Digital Transmission Systems, Duluth, GA.
- References: <EMCGUIRE.92Dec12155241@mother.intellection.com> <BzBMAn.57y@rice.edu> <EMCGUIRE.92Dec17092630@fuller.intellection.com>
- Date: Mon, 21 Dec 1992 22:13:11 GMT
- Lines: 63
-
- In article <EMCGUIRE.92Dec17092630@fuller.intellection.com> emcguire@intellection.com (Ed McGuire) writes:
- >I think the way to avoid this is to agree in advance on some objective
- >metrics. I'm unsure what metrics to use, but there are some numbers
- >floating around which /may/ fit the bill.
- >
- >For example, the UUNET news statistics include the number of sites
- >where at least one person posted an article within the last two weeks
- >(25821 in the December 9 article). It's not so important that we have
- >an accurate count of netnews sites as that we have a factor in the
- >voting rule which reflects the growth of netnews. The question
- >remaining is what constant multiplier should be used with 25821 as a
- >coefficient. And I would expect that the average over a three or six
- >month period would be more reliable than just using the most recent
- >figure.
-
- I like this idea.
-
- However, playing Devil's Advocate for a moment, I see some objections that
- might come up:
-
- 1) This places power into the hands of UUNET, at least in that whoever it is
- in that organization reporting the statistics can skew them, in theory.
-
- 2) This places a single 'point of failure' at UUNET. Should UUNET get wiped
- off the face of the planet, how do we get the numbers for the next vote(s)?
-
- 3) UUNET is not 'central' enough for regional hierarchies, and non-USA
- hierarchies.
-
- Now I can counter some of these points immediately:
-
- 1) It should be possible to have the large sites UUNET feeds compile the same
- lists, and merge them for the same result, at least for the accuracy this
- idea needs, alleviating the problem of one person/small group skewing the
- net growth.
-
- 2) Similarly, the backbone sites can run the same compilation, and we can
- agree on an 'order of succession' should any one major site get removed.
-
- 3) Perhaps the regional hierarchies can set up similar sites for their own
- subnet traffic. I believe the European sites tend to feed through single
- sites in many countries, facilitating the collection of statistics. However,
- the debates I've seen of the control issues make those sites more likely to
- fall into item number (1) above
-
- >I agree that "traffic on this topic" is difficult to formalize and may
- >lead to flame wars. On the other hand, we don't know in advance until
- >it is tried. It is certainly possible to count articles in related
- >mailing lists and to estimate percent of traffic in existing groups,
- >and it would be up to the proponents of the new group to come up with
- >a consensus figure during the CFD period.
-
- For the rec,games.frp spilt, there were a good number of topic counts used to
- justify the split into various subgroups based on traffic, so this can be
- done.
-
-
-
- --
- Pete Hardie: phardie@nastar (voice) (404) 497-0101
- Digital Transmission Systems, Inc., Duluth GA
- Member, DTS Dart Team | cat * | egrep -v "signature virus|infection"
- Position: Goalie |
-