stodol@diku.dk (David Stodolsky) (02/08/90)
This is a Call for Discussion of a possible change in the Guidelines for
creating newsgroups (How to Create a New Newsgroup). The single transferrable
vote (STV) procedure has already been used on several occasions where the
current Guidelines proved inadequate. This change would conform the Guidelines
to current practice and is supported by a recent voting methods poll. Procedures
for scoring such votes have been developed and posted. Also, many preliminary
votes have been run recently to avoid putting forward a name that will meet
resistance in a Guideline's vote. This is not satisfactory, since only a small
number of people participate in these votes, and the choices may therefore be
unrepresentative. Preliminary votes were not supported in the voting methods
poll.
The name, "Interest Group Surveys," has been chosen to avoid the adversarial
tone that is associated with votes, and to reflect more accurately the function
of these activities.
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Interest Group Surveys, called votes under the Guidelines, allow people to more
effectively use Net resources. They have two functions. The first is to
determine if enough people are interested in a topic. The second is to find a
good name for a newsgroup. Large mailing lists load host machines and the Net,
they also create bounced mail which absorb the time of skilled personnel.
The assumptions:
A newsgroup with 100 persons is less costly to maintain than a mailing list.
This determines support needed to create a newsgroup.
Persons not planning to participate in a group are permitted to express their
preference among names without influencing group creation.
================= STV in less than 25 lines ======================
Following the discussion period, all names that have been posted (seconded names
appearing first) will be included in an Interest Group Survey
announcement. Each name will be represented by a single character, with "a"
representing "abstain".
Respondents follow these rules:
1. While unmoved characters remain:
Select your preferred name and move its character to the subject line
(place each one after [to the right of] the ones already moved).
2. Place an "a" after the character representing the last name you find
acceptable. If you do not support creation of a newsgroup with any of the names,
put the "a" before the options.
Votes are scored by repeatedly removing the options that have the least support
from the front of responses, until the majority option is removed. Then subtract
the "a"s removed from the total responses to find the support for creation of
the newsgroup (for a worked example, see "Single transferrable vote counting").
If this is 100 or more, the group shall be created.
Number of responses received, number supporting creation, number supporting the
most preferred name, and the how each person voted, shall be posted.
============ end STV in less than 25 lines ==========================
The stopping rule used above is intended to work even if "abstain" is the
majority option.
The above procedure can be elaborated to include a "no group" choice as well as
an "abstain" choice. This would permit votes that either support, express no
support for, or oppose creation of the new group. Thus, proposals could be
defeated no matter how many people voted for them. This violates the first
assumption above, and it might then be useful to violate the second. That is,
votes against the creation of a new group can not express a preference among
names.
I doubt if discussion will resolve these possibilities. So a vote on a change to
the Guidelines would most likely include:
1. "STV in less than 25 lines", as above.
2. "STV in less than 25 lines", including a "no group" choice (support must be
greater than opposition to create the group [i.e., majority rule]).
3. "STV in less than 25 lines", including a "no group" choice (support must be
greater than opposition to create the group), and expressing preference among
names not permitted by those opposed to creation of the group.
4. Current rules.
-----------
Suggestions for changes to "'STV in less than 25 lines', as above", and other
options for the vote, would best be mailed to me as well as posted. I will
include them in an update to the proposal, which will appear no sooner than a
week after this posting.
These changes assume some modification to the Guidelines sections about
discussion and so on.
--
David S. Stodolsky, PhD Internet: david@harald.ruc.dk
Department of Psychology : stodol@diku.dk
Copenhagen Univ., Njalsg. 88 Voice: + 45 31 58 48 86
DK-2300 Copenhagen S, Denmark Fax: + 45 31 54 32 11hirchert@ux1.cso.uiuc.edu (Kurt Hirchert) (02/10/90)
1. A person can support, oppose, or have no opinion on the issue of creating a newsgroup with a particular charter. 2. A person can have an opinion on what a newsgroup with a given charter should be called, regardless of his/her opinion on whether the group should be created. 3. In some cases, the issue of name may affect a person's position on creation. In order to account for these variations, I suggest a vote string something like adcYeAbN would be interpreted as saying I like name "a" best, "d" next best,..., and "b" least. If names "a", "d", or "c" are chosen, I vote YES. If name "e" is chosen, I ABSTAIN. If name "b" is chosen, I vote NO. YES, ABSTAIN, and NO options would be required to come in that order. Vote evaluation would come in two stages: 1. First apply the single transferrable vote procedure to the lower case letters to determine what is the most acceptable name. 2. Then count the YES and NO votes corresponding to that choice to determine whether you have at least 100 more YES votes than NO and at least twice as many YES votes as NO (as per our existing procedures). (One way to implement the above is to remove an uppercase letter from a vote string whenever the immediately preceding lowercase letter is removed. Thus, once a lowercase letter achieves a majority, all you need to do is count the first remaining uppercase letter in each string.) What I'm really interested here is the method, not the syntax. If we're concerned about people not being about to produce both upper and lower case letters, we could use a syntax like adc/e/b in place of adcYeAbN . I return you now to your regularly scheduled brawl. -- Kurt W. Hirchert hirchert@ncsa.uiuc.edu National Center for Supercomputing Applications
chuq@Apple.COM (Chuq Von Rospach) (02/10/90)
hirchert@ux1.cso.uiuc.edu (Kurt Hirchert) writes: >In order to account for these variations, I suggest a vote string something >like > adcYeAbN I like this. If we can make it even more obscure and complicated, we can keep enough people from voting that it'd bring back the backbone cabal, where only folks really motivated will learn the jargon needed to join in. -- Chuq Von Rospach <+> chuq@apple.com <+> [This is myself speaking] Rumour has it that Larry Wall, author of RN, is a finalist in the race for the Nobel Peace Prize for his invention of the kill file.
woods@snowmass.scd.ucar.edu (Greg Woods) (02/10/90)
I do not wish to get into an endless discussion, but in my opinion, any
voting procedure that requires programs that are not a standard part of
an operating system in use on a significant number of machines on the net
in order to count votes cannot be considered to be easily verifiable.
Remember that in order to be verifiable, Joe Random User needs to be able
to verify that the posted votes really add up to the result claimed by
the vote taker (mayube not EVERY Joe Random, but some large percentage of
them anyway). The STV procedure does not meet this requirement. I have
no objection to its use in initial name surveys; I think anything that
cuts down on the flame wars over names is a good thing. But I don't think
we want to replace that with flame wars on the voting results instead, nor
do we want to put even more power in the hands of the vote takers.
Votes which are difficult to verify will do just that. I am opposed to
using the STV procedure for official newsgroup creation votes.
I have stated my case here, so my lack of response to any rebuttals
of this article does not indicate any change in my position. Rather it
indicates my desire to stay out of a never-ending discussion by allowing my
opponents to have the last word.
--Greg
P.S. I'm all for changing the word "vote" to "survey" to more accurately
reflect reality.peter@ficc.uu.net (Peter da Silva) (02/10/90)
Firstly, I'm not sure what the intent of a separate "abstain" and "no group"
entries is. You are requiring that there be 100 more "majority" votes than
"abstain" votes. This makes the "abstain" votes de-facto "no group" votes.
After running a few STV votes, I have come to the conclusion that a fairer
way to do things is for each voter to indicate whether they think the group
should exist or not independently of where they think the group should be in
the hierarchy. I would like to propose the following change:
The vote is in two parts. First, indicate on the subject line what your vote
on the existence of the group is: YES or NO. Then, following the YES or NO
indicate your preferences for a name. If you have no preference, leave this
part blank. The preferences should be handled as listed in the original call
for discussion, except that no "abstain" entry would be used. Only list the
groups that you have a preference for. Your preferences will be counted even
if you vote against the group.
If there are more than 100 more YES then NO votes, *and* the name that wins
a majority of votes has more than 2/3 more votes than any other candidate,
the group is created with that name. If eiither of these conditions does not
hold true, the group is not created.
Subject: YES bdcaf on the JUNGIAN vote.
--
_--_|\ Peter da Silva. +1 713 274 5180. <peter@ficc.uu.net>.
/ \
\_.--._/ Xenix Support -- it's not just a job, it's an adventure!
v "Have you hugged your wolf today?" `-_-'djgrabin@phoenix.Princeton.EDU (David Joseph Grabiner) (02/12/90)
In article <L=N11X4xds13@ficc.uu.net> peter@ficc.uu.net (Peter da Silva) writes: >The vote is in two parts. First, indicate on the subject line what your vote >on the existence of the group is: YES or NO. Then, following the YES or NO >indicate your preferences for a name. If you have no preference, leave this >part blank. The preferences should be handled as listed in the original call >for discussion, except that no "abstain" entry would be used. Only list the >groups that you have a preference for. Your preferences will be counted even >if you vote against the group. >Subject: YES bdcaf on the JUNGIAN vote. This system doesn't allow expression of a common opinion: that you dislike a particular name so much that you will vote against a group with that name. When the *.aquaria vote came up, I would have abstained on rec.aquaria, but I voted NO on sci.aquaria. Under this system, I wouldn't have been able to vote at all, because I certainly didn't want to vote NO on rec.aquaria. If the surveys are done separately, first selecting a name by a binding STV vote and then holding a regular vote on the name selected by the net, this problem is removed, at the cost of adding some time to the group creation process. In many cases, the extended time isn't needed. The STV with YES, NO, and ABSTAIN votes, which has been proposed here before, solves this problem. Another solution, which I have proposed before, is to say that any group which passes despite 100 NO votes, which usually means unresolved dispute over the name, must then have an STV survey to determine the name. My proposal is somwehat simpler, but I prefer the STV with YES, NO, and ABSTAIN included. This would allow me to vote: YES: ABSTAIN: rec.aquaria rec.aquarium rec.pets.fish NO: sci.aquaria sci.aquarium -- My name is David Joseph Grabiner. I am called djgrabin@phoenix.princeton.edu. My name is called David. I am the person solely responsible for these opinions.
peter@ficc.uu.net (Peter da Silva) (02/12/90)
In article <13779@phoenix.Princeton.EDU> djgrabin@phoenix.Princeton.EDU (David Joseph Grabiner) writes: > This system doesn't allow expression of a common opinion: that you > dislike a particular name so much that you will vote against a group > with that name. Abstain from the YES/NO part and vote on the other groups in the choice section. Because of the 2/3 rule, that's as good as voting against the group. (let's see, a=sci.aquaria, b=sci.aquariums, c=rec.aquaria, d=rec.aquariums, e=rec.pets.fish) Subject: N/A cde I suspect the idea can do with some tuning, Perhaps a simple majority in the name part should be enough. But it's a lot less complex than the YES/NO/ABSTAIN proposal... which I also believe to be incomplete in its explanation of vote counting. I haven't looked too closely at it. -- _--_|\ Peter da Silva. +1 713 274 5180. <peter@ficc.uu.net>. / \ \_.--._/ Xenix Support -- it's not just a job, it's an adventure! v "Have you hugged your wolf today?" `-_-'
lear@turbo.bio.net (Eliot) (02/15/90)
We are arguing on how to pick a name. While siding in general with Greg and Chuq on this issue, I see no reason why one couldn't use STV to pick the name before a call for votes to see if the group should be created. This would require for no serious change in the guidelines. -- Eliot Lear [lear@TURBO.BIO.NET]
woods@snowmass.scd.ucar.edu (Greg Woods) (02/15/90)
In article <Feb.14.14.09.55.1990.1047@turbo.bio.net> lear@turbo.bio.net (Eliot) writes: >I see no reason why one couldn't use STV >to pick the name before a call for votes to see if the group should >be created. This would require for no serious change in the >guidelines. I agree. The purpose of the discussion period prior to calling for a vote is to firstly decide on whether a group is really desireable, and second to decide on a name. The guidelines do not say how the latter decision should be made. I see nothing wrong with formalizing that with a preliminary vote on the name. Since that vote is not intended to be binding, it doesn't have to be easily verifiable. Those who are truly opposed to a name that wins an STV vote are still free to vote against it come official vote time, and those who don't care about the name can still vote for it. In most cases, I don't think such a preliminary vote is needed, but in cases where there is a lot of controversy over the name, taking a preliminary vote on the name might be one way of formally ending the discussion period and getting on to a creation vote. And when there *is* a big controversy over the name, I see nothing wrong with delaying the creation by running a preliminary name vote in order to reduce the flame wars over the name. --Greg
brad@looking.on.ca (Brad Templeton) (02/15/90)
Sometimes I think it would be a good idea to remove the "discussion phase" from the guidelines. Yes, sometimes the discussion causes people to change their minds, but it's pretty rare, and I doubt it has much effect, except the negative effect of making the 'vote' about personalities and names and other meaningless issues. Have you ever decided to "vote" yes for a group because the discussion phase convinced you there was interest in the group, and you wouldn't have "voted" so otherwise? The discussion phase only seems to cause NO votes based on pointless issues, and perhaps a few YES votes to counter anticipated NOs. Why do we have it? We can discuss what we need during the survey. Your survey response is supposed to indicate YOUR interest in the group, not your opinion of the debate! Who gets changed from YES to NO by discussion, except over the name etc.? Nobody. People get changed from not-vote to NO. That's about it. -- Brad Templeton, ClariNet Communications Corp. -- Waterloo, Ontario 519/884-7473
peter@ficc.uu.net (Peter da Silva) (02/15/90)
In article <Feb.14.14.09.55.1990.1047@turbo.bio.net> lear@turbo.bio.net (Eliot) writes: > I see no reason why one couldn't use STV to pick the name before a call > for votes to see if the group should be created. Because the people intent on such clumsy fraud as has happened in the past will equally ignore any pre-vote poll. My STV proposal, currently on the table, would be to hold these two votes in parallel. To give the person proposing a group with a controversial name some incentive to go ahead with the STV vote, I'm suggesting a partial relaxation of the 2/3 rule. As has been shown many times in the past, you can't expect people to behave rationally unless you give them some incentive to do so. -- _--_|\ Peter da Silva. +1 713 274 5180. <peter@ficc.uu.net>. / \ \_.--._/ Xenix Support -- it's not just a job, it's an adventure! v "Have you hugged your wolf today?" `-_-'
jeffd@ficc.uu.net (Jeff Daiell) (02/15/90)
In article <E2S1.H1xds13@ficc.uu.net>, peter@ficc.uu.net (Peter da Silva) writes: > In article <Feb.14.14.09.55.1990.1047@turbo.bio.net> lear@turbo.bio.net (Eliot) writes: > > I see no reason why one couldn't use STV to pick the name before a call > > for votes to see if the group should be created. > > My STV proposal, currently on the > table, would be to hold these two votes in parallel. Holding them together would also reduce group.creation.time, and several folks have noted that the process takes too long as things stand. Jeff Daiell -- Thank you for not coercing.
peter@ficc.uu.net (peter da silva) (02/15/90)
In article <96250@looking.on.ca>, brad@looking.on.ca (Brad Templeton) writes: > Sometimes I think it would be a good idea to remove the "discussion phase" > from the guidelines. Yes, sometimes the discussion causes people to > change their minds, ... But that's not what the discussion phase is for. It's to firm up the group name and charter. There have been instances where the name of a group was changed during the discussion phase (and, admitedly, there have been cases where the discussion phase clearly indicated a name change was in order and no such change was made). It also provides a way to *quickly* weed out things like the Depeche Mode newsgroup. -- _--_|\ Peter da Silva. +1 713 274 5180. <peter@ficc.uu.net>. / \ \_.--._/ Xenix Support -- it's not just a job, it's an adventure! v "Have you hugged your wolf today?" `-_-'
woods@snowmass.scd.ucar.edu (Greg Woods) (02/16/90)
In article <96250@looking.on.ca> brad@looking.on.ca (Brad Templeton) writes: >Have you ever decided to "vote" yes for a group because the discussion >phase convinced you there was interest in the group, and you wouldn't have >"voted" so otherwise? The point of the discussion period, according to the article you followed up and also according to the guidelines, is to 1) Decide if a new group is really desired; and 2) Decide on a name for the group. Neither my article nor the guidelines say anything about using the discussion period to convince people how they should vote. Yes, it gets used for that, but that isn't the main purpose of it. And yes, as a matter of fact I *have* seen proposals dropped or names changed as a result of the discussion period. --Greg
woods@snowmass.scd.ucar.edu (Greg Woods) (02/16/90)
In article <O2S16.5xds12@ficc.uu.net> jeffd@ficc.uu.net (Jeff Daiell) writes: >Holding [name and creation votes] together would also reduce group.creation.time, >and several folks have noted that the process takes too long >as things stand. I don't buy this. The creation takes 14+21+5 = 40 days if done according to official rules. I don't see that the need for any new group is so critical that 40 days is anything to get riled up over. And even if a preliminary name vote is held, since that can be done concurrently with the first 14 days of discussion, it doesn't necessarily even lengthen the process that much. --Greg
brad@looking.on.ca (Brad Templeton) (02/16/90)
In article <6316@ncar.ucar.edu> woods@snowmass.scd.ucar.edu (Greg Woods) writes: >The point of the discussion period, according to the article you followed >up and also according to the guidelines, is to 1) Decide if a new group >is really desired; and 2) Decide on a name for the group. Neither my >article nor the guidelines say anything about using the discussion period >to convince people how they should vote. Yes, it gets used for that, but >that isn't the main purpose of it. And yes, as a matter of fact I *have* seen >proposals dropped or names changed as a result of the discussion period. But would they not be dropped, even more quietly, if they started the survey right away, and saw there was no interest? I am not talking about banning discussion here (not that this wouldn't be a good idea! :-]) but simply starting the survey right away (if desired) and letting any discussion take place in parallel if people really want it. By and large, I am not sure we do want it. But you can have it if you like. The point was, the discussion doesn't *change* people's votes, it just adds new ones. So there is no problem with them taking place together. And if discussion calls off a proposal, that doesn't cause harm if the survey has started. The proposer simply abandons it. And yes, there's a problem with names, but there's always a problem with names and this is no better or worse. -- Brad Templeton, ClariNet Communications Corp. -- Waterloo, Ontario 519/884-7473
woods@snowmass.scd.ucar.edu (Greg Woods) (02/17/90)
In article <96971@looking.on.ca> brad@looking.on.ca (Brad Templeton) writes: >In article <6316@ncar.ucar.edu> woods@snowmass.scd.ucar.edu (Greg Woods) writes: >>And yes, as a matter of fact I *have* seen >>proposals dropped or names changed as a result of the discussion period. > >But would they not be dropped, even more quietly, if they started the >survey right away, and saw there was no interest? In some cases, perhaps. But I still think it is a good idea if people have some idea what they are voting for. That is part of what the discussion period is also about: to at least semi-formalize the group charter. >but simply starting the survey right away (if desired) >and letting any discussion take place in parallel if people really want it. That can cause other problems, such as what happens if the charter or name of the group changes as a result of the discussion in a way I don't like, but I've already voted? Then we get people wanting to change their votes and it gets lots messier than it already is. >The point was, the discussion doesn't *change* people's votes, it just >adds new ones I'm not sure I agree with this as I said above. It may not change someone's mind about a particular proposal, but it *might* change the proposal in a way that would cause someone's vote to change. For example, if the sci.aquaria name had been changed as a result of the discussion I might not have voted against it, I probably would have just abstained. > So there is no problem with them taking place together. I think there is. >And if discussion calls off a proposal, that doesn't cause harm if the >survey has started. The proposer simply abandons it. This gives far too much power to the vote-taker. It presumes the final result before the vote is finished. >And yes, there's a problem with names, but there's always a problem with >names and this is no better or worse. I think it would be a LOT worse if the name could change after some people have already voted. And what would we gain if we did this? A few days off the creation time? I don't think those few days are worth the potential extra mess. --Greg
brad@looking.on.ca (Brad Templeton) (02/17/90)
In article <6331@ncar.ucar.edu> woods@snowmass.scd.ucar.edu (Greg Woods) writes: > > I think it would be a LOT worse if the name could change after some people >have already voted. > > And what would we gain if we did this? A few days off the creation time? >I don't think those few days are worth the potential extra mess. What we gain is actually the removal of the "discussion phase" concept from the process, and thus, by and large, a lot of the "discussions." Most of these discussions, I would venture, are only read by a small minority of people. The rest of us use the 'k' key with great regularity on them, if comments I have heard are any indication. It would eliminate the "I think group x wuold be a good idea" postings we always get when a 'call for discussion' is made. We might well gain a lot this way. I see no problem in the name changing after people vote, but then as I have written many times, I think voting is the dumbest way you could devise for picking names, and I have yet to figure out why so many people think it's worth voting on. (Yes, I know, pseudo-democratic ideals are drummed into you in school. That's no excuse.) As if naming groups is a popularity contest of some sort. But whenever I or somebody else suggests something new, there are always those who leap to the defense of the status quo. Usenet, the network of arch-conservative anarchists. What a contradiction. -- Brad Templeton, ClariNet Communications Corp. -- Waterloo, Ontario 519/884-7473
" Maynard) (02/17/90)
In article <96250@looking.on.ca> brad@looking.on.ca (Brad Templeton) writes: >Have you ever decided to "vote" yes for a group because the discussion >phase convinced you there was interest in the group, and you wouldn't have >"voted" so otherwise? Yes: talk.politics.objectivism. -- Jay Maynard, EMT-P, K5ZC, PP-ASEL | Never ascribe to malice that which can jay@splut.conmicro.com (eieio)| adequately be explained by stupidity. {attctc,bellcore}!texbell!splut!jay +---------------------------------------- Free the DC-10!