June 2012 - mpi-forum - mailman3.cels.anl.gov

Re: [Mpi-forum] Voting in July (and beyond)
by Dave Goodell 14 Jun '12

14 Jun '12

On Jun 14, 2012, at 10:53 AM CDT, David Solt wrote: >> At the Japan meeting we took consensus from the room because there was >> no clear ruling on this known to *anyone* in the room. > > I disagree. I felt there was a clear ruling on this and stated that at > the Japan meeting. Perhaps my "anyone" statement was slightly too strong. Nonetheless, I don't recall many (any?) others with a similar level of conviction. > I don't know how people can be so out of touch with > what goes on at the meetings to not know how we have been voting for > years. The way that this happens is that both interpretations of the rules have always yielded the same result in the past. Only at the Japan meeting did we encounter a set of votes that would result in different outcomes under each interpretation. As Jeff has repeatedly pointed out, the voting rules web page is ambiguous, and reading it tends to just confirm the reader's view, regardless of which view that is. Years of voting in which the results match up with someone's interpretation (true for both interpretations) further reinforces a particular viewpoint. -Dave

1 0

Re: [Mpi-forum] Voting in July (and beyond)
by David Solt 14 Jun '12

14 Jun '12

> Even from this text, there is multiple possible interpretations. :-\ Yes, but only one correct interpretation. Dave From: Jeff Squyres <jsquyres(a)cisco.com> To: Main MPI Forum mailing list <mpi-forum(a)lists.mpi-forum.org> Date: 06/14/2012 11:10 AM Subject: Re: [Mpi-forum] Voting in July (and beyond) Sent by: mpi-forum-bounces(a)lists.mpi-forum.org On Jun 14, 2012, at 11:50 AM, Rolf Rabenseifner wrote: > Was nobody looking at our rules: > http://meetings.mpi-forum.org/secretary/voting.php > - Measures pass on simple majority. > - Formal votes will be taken by roll. > If you are not present, your organization > does not vote (effectively the same as abstain). Please see my prior emails (from a week or two ago) about this: - there are multiple accepted definitions of "simple majority" - "quorum" is not defined - there is also a question of the definition of "not present" -- does that not present in the room? (which is how I have interpreted it, but there has been some bending of that rule on occasion) Or not present at the meeting at all? Or not present that day? Or ...? > Here you can see "abstain" == "not voting", > and wiki clearly tells "simple majority" > means more than 50% of the passed votes. Even from this text, there is multiple possible interpretations. :-\ > Why was there a discussion about this? > > Jeff, how old is this file? SVN says: ------------------------------------------------------------------------ r312 | jsquyres | 2010-06-16 16:30:51 -0400 (Wed, 16 Jun 2010) | 2 lines Changed paths: M /trunk/secretary/index.php A /trunk/secretary/voting.php Add voting rules ------------------------------------------------------------------------ The file has not changed since that date. In mails a week or two ago, I discussed the history of the MPI-3 procedures document, too. -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/ _______________________________________________ mpi-forum mailing list mpi-forum(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

1 0

Re: [Mpi-forum] Voting in July (and beyond)
by Jeff Squyres 14 Jun '12

14 Jun '12

On Jun 14, 2012, at 11:50 AM, Rolf Rabenseifner wrote: > Was nobody looking at our rules: > http://meetings.mpi-forum.org/secretary/voting.php > - Measures pass on simple majority. > - Formal votes will be taken by roll. > If you are not present, your organization > does not vote (effectively the same as abstain). Please see my prior emails (from a week or two ago) about this: - there are multiple accepted definitions of "simple majority" - "quorum" is not defined - there is also a question of the definition of "not present" -- does that not present in the room? (which is how I have interpreted it, but there has been some bending of that rule on occasion) Or not present at the meeting at all? Or not present that day? Or ...? > Here you can see "abstain" == "not voting", > and wiki clearly tells "simple majority" > means more than 50% of the passed votes. Even from this text, there is multiple possible interpretations. :-\ > Why was there a discussion about this? > > Jeff, how old is this file? SVN says: ------------------------------------------------------------------------ r312 | jsquyres | 2010-06-16 16:30:51 -0400 (Wed, 16 Jun 2010) | 2 lines Changed paths: M /trunk/secretary/index.php A /trunk/secretary/voting.php Add voting rules ------------------------------------------------------------------------ The file has not changed since that date. In mails a week or two ago, I discussed the history of the MPI-3 procedures document, too. -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Re: [Mpi-forum] Voting in July (and beyond)
by David Solt 14 Jun '12

14 Jun '12

> At the Japan meeting we took consensus from the room because there was no clear ruling on this known to *anyone* in the room. I disagree. I felt there was a clear ruling on this and stated that at the Japan meeting. I don't know how people can be so out of touch with what goes on at the meetings to not know how we have been voting for years. Thanks, Dave From: Dave Goodell <goodell(a)mcs.anl.gov> To: Main MPI Forum mailing list <mpi-forum(a)lists.mpi-forum.org> Date: 06/14/2012 10:30 AM Subject: Re: [Mpi-forum] Voting in July (and beyond) Sent by: mpi-forum-bounces(a)lists.mpi-forum.org On Jun 14, 2012, at 9:15 AM CDT, Jeff Squyres wrote: > On Jun 14, 2012, at 9:25 AM, Richard Graham wrote: > >> First, I would not note that as a mishap - a specific question came up, and the forum did it's best to address the question, and interpret what majority vote means. > > I'm not debating what happened. But the mishap = mistake, and this was clearly a mistake, regardless of good intentions. > > So however you want to define it, I don't think that anyone disagrees that the voting rules were different in Japan than they had been in all prior meetings. They may not, however, have been *observably* different. If we never had a close vote in the past then nobody but you would have known that the voting rules were changed. >> We need to keep with what was the consensus at the meeting, > > I'm not disagreeing with the fact that we need to abide by the voting results as they were decided in the room in Japan (because everyone knew/understood the "Japan" rules when they were voting). Absolutely, almost anything else would be crazy and/or a waste of everyone's time. >> and change that ONLY after further discussion. So, it at this stage pass mean yes > 0.5 * total vote. > > No. > > The Japan rules were a mistake. > > It is not proper to change the voting rules: > > a) with discussion from only one meeting, and > b) when the secretary -- i.e., the guy who knows the rules -- was not even present > > In July, voting should be the same as it has been for all prior 2.x and 3.x Forum meetings: pass = yes > no (abstains don't count). > > *** We =CANNOT= abolish "abstain" at the whim of one mis-guided decision from one meeting *** You are effectively asserting that your voting rules are more correct because you said that they came first, even though they were not clearly documented to anyone else. >> We can have this as the first item for discussion on Monday, but have to have a discussion before we change. > > I do not think it is proper to change the voting rules based on any one meeting. To me, changing the voting rules should require 2 formal votes (just like text) using the currently established rules. We have no procedure for this. AFAIK we never voted the original rules into place. At the Japan meeting we took consensus from the room because there was no clear ruling on this known to *anyone* in the room. Both members of the steering committee that were present (Bill & Rich) felt the "abstains can't change the denominator" view was correct. I think that an in-person discussion about this on the first day of the upcoming meeting is the most appropriate way to deal with this debate. Perhaps a two-votes-over-two-meetings-scheme is appropriate, perhaps not. When we have made other procedural changes, such as the "ticket 0" concept, I don't think that we used a two vote process. I don't see this as substantially different. The most important thing for us to do is not bicker over which rules were correct at various points in the past, but instead we should attempt to agree on rules that work for the Forum going forward. That could be either interpretation of simple majority, or possibly some more complicated scheme (can we somehow shoehorn instant-runoff voting into the process? ;) ). -Dave _______________________________________________ mpi-forum mailing list mpi-forum(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

1 0

Re: [Mpi-forum] Voting in July (and beyond)
by Jeff Squyres 14 Jun '12

14 Jun '12

On Jun 14, 2012, at 11:25 AM, Dave Goodell wrote: >> *** We =CANNOT= abolish "abstain" at the whim of one mis-guided decision from one meeting *** > > You are effectively asserting that your voting rules are more correct because you said that they came first, even though they were not clearly documented to anyone else. Yes and no. I'm saying that we've all been collectively crazy if, for the past 4 years of MPI 2.x and 3.0 meetings (and the MPI-2.0 meetings in the mid-90's), we've had an "abstain" position that meant nothing. Since I think we're *not* all collectively crazy, the only sensible conclusion is that "abstain" actually means something other than the other voting positions. I have been asked about the voting rules many times, and I have discussed how I record the votes publicly in meetings. I have *always* stated (and recorded) that pass = yes > no. The PHP on the web site reflects this. The spreadsheets in subversion reflect this. >> I do not think it is proper to change the voting rules based on any one meeting. To me, changing the voting rules should require 2 formal votes (just like text) using the currently established rules. > > We have no procedure for this. AFAIK we never voted the original rules into place. Agreed. My statement above was intended as "To me (meaning: this is my opinion), changing the voting rules..." Indeed, in my first mail on this thread, I stated "...discuss potentially changing the voting rules for future meetings ... and when/how to transition to new voting rules." > At the Japan meeting we took consensus from the room because there was no clear ruling on this known to *anyone* in the room. Both members of the steering committee that were present (Bill & Rich) felt the "abstains can't change the denominator" view was correct. Regardless of end voting results, this is not how the votes have been computed and recorded for the past 3-4 years. We all know how the Forum can be completely swayed from one meeting to the next -- that's one reason we have the "text must be voted in twice" rule. Indeed, look at what happened to Mohammad's ticket -- broad consensus one meeting, voted down the next. Forum-swaying very much depends on attendance. > I think that an in-person discussion about this on the first day of the upcoming meeting is the most appropriate way to deal with this debate. I am wary of this simply because of "the Forum can be swayed at a single meeting" effect. That's why I propose: 1. Until voting rules are formally changed, we should continue as I have recorded the votes for the past 3-4 years (i.e., pass = yes > no) 2. We clarify/create new voting rules over a 2 meeting period *** This proposal means that we will close out MPI-3 with a (mostly) consistent set of votes. And we can start MPI-<next> with a potentially new set of voting rules. > Perhaps a two-votes-over-two-meetings-scheme is appropriate, perhaps not. When we have made other procedural changes, such as the "ticket 0" concept, I don't think that we used a two vote process. I don't see this as substantially different. My $0.02: it is substantially different. Ticket 0 changes are broadly considered to be irrelevant -- we could not do them, and the standard wouldn't have a different meaning. Non-ticket-0 changes have substantive changes. Changing the voting rules is a substantive change, and must be done deliberately. Otherwise, we could just change the rules every meeting. > The most important thing for us to do is not bicker over which rules were correct at various points in the past, but instead we should attempt to agree on rules that work for the Forum going forward. Fair enough. I don't think anyone wants to re-interpret old votes. My point is that changing the computation in a single meeting should not be a basis for going forward. -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Re: [Mpi-forum] Voting in July (and beyond)
by Rolf Rabenseifner 14 Jun '12

14 Jun '12

Was nobody looking at our rules: http://meetings.mpi-forum.org/secretary/voting.php - Measures pass on simple majority. - Formal votes will be taken by roll. If you are not present, your organization does not vote (effectively the same as abstain). Here you can see "abstain" == "not voting", and wiki clearly tells "simple majority" means more than 50% of the passed votes. Why was there a discussion about this? Jeff, how old is this file? Best regards Rolf ----- Original Message ----- > From: "Dave Goodell" <goodell(a)mcs.anl.gov> > To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> > Sent: Thursday, June 14, 2012 5:25:56 PM > Subject: Re: [Mpi-forum] Voting in July (and beyond) > On Jun 14, 2012, at 9:15 AM CDT, Jeff Squyres wrote: > > > On Jun 14, 2012, at 9:25 AM, Richard Graham wrote: > > > >> First, I would not note that as a mishap - a specific question came > >> up, and the forum did it's best to address the question, and > >> interpret what majority vote means. > > > > I'm not debating what happened. But the mishap = mistake, and this > > was clearly a mistake, regardless of good intentions. > > > > So however you want to define it, I don't think that anyone > > disagrees that the voting rules were different in Japan than they > > had been in all prior meetings. > > They may not, however, have been *observably* different. If we never > had a close vote in the past then nobody but you would have known that > the voting rules were changed. > > >> We need to keep with what was the consensus at the meeting, > > > > I'm not disagreeing with the fact that we need to abide by the > > voting results as they were decided in the room in Japan (because > > everyone knew/understood the "Japan" rules when they were voting). > > Absolutely, almost anything else would be crazy and/or a waste of > everyone's time. > > >> and change that ONLY after further discussion. So, it at this stage > >> pass mean yes > 0.5 * total vote. > > > > No. > > > > The Japan rules were a mistake. > > > > It is not proper to change the voting rules: > > > > a) with discussion from only one meeting, and > > b) when the secretary -- i.e., the guy who knows the rules -- was > > not even present > > > > In July, voting should be the same as it has been for all prior 2.x > > and 3.x Forum meetings: pass = yes > no (abstains don't count). > > > > *** We =CANNOT= abolish "abstain" at the whim of one mis-guided > > decision from one meeting *** > > You are effectively asserting that your voting rules are more correct > because you said that they came first, even though they were not > clearly documented to anyone else. > > >> We can have this as the first item for discussion on Monday, but > >> have to have a discussion before we change. > > > > I do not think it is proper to change the voting rules based on any > > one meeting. To me, changing the voting rules should require 2 > > formal votes (just like text) using the currently established rules. > > We have no procedure for this. AFAIK we never voted the original rules > into place. At the Japan meeting we took consensus from the room > because there was no clear ruling on this known to *anyone* in the > room. Both members of the steering committee that were present (Bill & > Rich) felt the "abstains can't change the denominator" view was > correct. > > I think that an in-person discussion about this on the first day of > the upcoming meeting is the most appropriate way to deal with this > debate. Perhaps a two-votes-over-two-meetings-scheme is appropriate, > perhaps not. When we have made other procedural changes, such as the > "ticket 0" concept, I don't think that we used a two vote process. I > don't see this as substantially different. > > The most important thing for us to do is not bicker over which rules > were correct at various points in the past, but instead we should > attempt to agree on rules that work for the Forum going forward. That > could be either interpretation of simple majority, or possibly some > more complicated scheme (can we somehow shoehorn instant-runoff voting > into the process? ;) ). > > -Dave > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum -- Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)hlrs.de High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530 University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832 Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner Nobelstr. 19, D-70550 Stuttgart, Germany . (Office: Allmandring 30)

1 0

Re: [Mpi-forum] Voting in July (and beyond)
by Dave Goodell 14 Jun '12

14 Jun '12

On Jun 14, 2012, at 9:15 AM CDT, Jeff Squyres wrote: > On Jun 14, 2012, at 9:25 AM, Richard Graham wrote: > >> First, I would not note that as a mishap - a specific question came up, and the forum did it's best to address the question, and interpret what majority vote means. > > I'm not debating what happened. But the mishap = mistake, and this was clearly a mistake, regardless of good intentions. > > So however you want to define it, I don't think that anyone disagrees that the voting rules were different in Japan than they had been in all prior meetings. They may not, however, have been *observably* different. If we never had a close vote in the past then nobody but you would have known that the voting rules were changed. >> We need to keep with what was the consensus at the meeting, > > I'm not disagreeing with the fact that we need to abide by the voting results as they were decided in the room in Japan (because everyone knew/understood the "Japan" rules when they were voting). Absolutely, almost anything else would be crazy and/or a waste of everyone's time. >> and change that ONLY after further discussion. So, it at this stage pass mean yes > 0.5 * total vote. > > No. > > The Japan rules were a mistake. > > It is not proper to change the voting rules: > > a) with discussion from only one meeting, and > b) when the secretary -- i.e., the guy who knows the rules -- was not even present > > In July, voting should be the same as it has been for all prior 2.x and 3.x Forum meetings: pass = yes > no (abstains don't count). > > *** We =CANNOT= abolish "abstain" at the whim of one mis-guided decision from one meeting *** You are effectively asserting that your voting rules are more correct because you said that they came first, even though they were not clearly documented to anyone else. >> We can have this as the first item for discussion on Monday, but have to have a discussion before we change. > > I do not think it is proper to change the voting rules based on any one meeting. To me, changing the voting rules should require 2 formal votes (just like text) using the currently established rules. We have no procedure for this. AFAIK we never voted the original rules into place. At the Japan meeting we took consensus from the room because there was no clear ruling on this known to *anyone* in the room. Both members of the steering committee that were present (Bill & Rich) felt the "abstains can't change the denominator" view was correct. I think that an in-person discussion about this on the first day of the upcoming meeting is the most appropriate way to deal with this debate. Perhaps a two-votes-over-two-meetings-scheme is appropriate, perhaps not. When we have made other procedural changes, such as the "ticket 0" concept, I don't think that we used a two vote process. I don't see this as substantially different. The most important thing for us to do is not bicker over which rules were correct at various points in the past, but instead we should attempt to agree on rules that work for the Forum going forward. That could be either interpretation of simple majority, or possibly some more complicated scheme (can we somehow shoehorn instant-runoff voting into the process? ;) ). -Dave

1 0

Re: [Mpi-forum] Voting in July (and beyond)
by Jeff Squyres 14 Jun '12

14 Jun '12

On Jun 14, 2012, at 9:25 AM, Richard Graham wrote: > First, I would not note that as a mishap - a specific question came up, and the forum did it's best to address the question, and interpret what majority vote means. I'm not debating what happened. But the mishap = mistake, and this was clearly a mistake, regardless of good intentions. So however you want to define it, I don't think that anyone disagrees that the voting rules were different in Japan than they had been in all prior meetings. > We need to keep with what was the consensus at the meeting, I'm not disagreeing with the fact that we need to abide by the voting results as they were decided in the room in Japan (because everyone knew/understood the "Japan" rules when they were voting). But that does not mean that the voting under those rules was not a mistake. That's a double negative, so let me rephrase to be clear: voting under those rules was a mistake. A mishap. It shouldn't have happened. It *wouldn't* have happened had I been there. Put simply: We have to accept the *results* of the Japan votes, but not the Japan voting rules. > and change that ONLY after further discussion. So, it at this stage pass mean yes > 0.5 * total vote. No. The Japan rules were a mistake. It is not proper to change the voting rules: a) with discussion from only one meeting, and b) when the secretary -- i.e., the guy who knows the rules -- was not even present In July, voting should be the same as it has been for all prior 2.x and 3.x Forum meetings: pass = yes > no (abstains don't count). *** We =CANNOT= abolish "abstain" at the whim of one mis-guided decision from one meeting *** > We can have this as the first item for discussion on Monday, but have to have a discussion before we change. I do not think it is proper to change the voting rules based on any one meeting. To me, changing the voting rules should require 2 formal votes (just like text) using the currently established rules. -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Re: [Mpi-forum] Voting in July (and beyond)
by Richard Graham 14 Jun '12

14 Jun '12

First, I would not note that as a mishap - a specific question came up, and the forum did it's best to address the question, and interpret what majority vote means. We need to keep with what was the consensus at the meeting, and change that ONLY after further discussion. So, it at this stage pass mean yes > 0.5 * total vote. We can have this as the first item for discussion on Monday, but have to have a discussion before we change. Rich -----Original Message----- From: mpi-forum-bounces(a)lists.mpi-forum.org [mailto:[email protected]] On Behalf Of Jeff Squyres Sent: Thursday, June 14, 2012 9:06 AM To: MPI Forum list Subject: [Mpi-forum] Voting in July (and beyond) Regardless on the Japan voting mishap and the conversation that has occurred since then (most of which was here on the list), I don't think there's enough time to change voting rules for July. Hence, I intend to use the same voting rules that I have used previously: yes, no, and abstain have different meanings. Votes pass if (num_yes_votes > num_no_votes). Rich: please add an agenda item for Chicago to discuss potentially changing the voting rules for future meetings (e.g., good points were made about majority definitions, consensus / near-consensus definitions, etc.), and when/how to transition to new voting rules. -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/ _______________________________________________ mpi-forum mailing list mpi-forum(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

1 0

Voting in July (and beyond)
by Jeff Squyres 14 Jun '12

14 Jun '12

Regardless on the Japan voting mishap and the conversation that has occurred since then (most of which was here on the list), I don't think there's enough time to change voting rules for July. Hence, I intend to use the same voting rules that I have used previously: yes, no, and abstain have different meanings. Votes pass if (num_yes_votes > num_no_votes). Rich: please add an agenda item for Chicago to discuss potentially changing the voting rules for future meetings (e.g., good points were made about majority definitions, consensus / near-consensus definitions, etc.), and when/how to transition to new voting rules. -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0