[Mpi-forum] MPI Forum voting rules: new version

Aurélien Bouteiller bouteill at icl.utk.edu
Wed Jan 16 15:36:45 CST 2013


I'm citing point 5. IMOVE



Individual Meeting Organization Voting Eligibility (IMOVE): An organization is eligible to vote at a specific physical MPI Forum meeting if all of the following are true:

	• (a)  The organization is OOE.

	• (b)  An individual representing this organization registered for that spe- cific physical MPI Forum meeting before the first ballot occured.

	• (c)  The organization had at least one of its representatives physically present at any time during that specific physical MPI Forum meeting.

Once an organization becomes IMOVE for a specific physical MPI Forum meeting, that organization stays IMOVE for the remainder of that spe- cific physical MPI Forum meeting. For example, if an organization’s only representative leaves the meeting, that organization still remains IMOVE.

I remember the rationale about not having IMOVE change retroactively (and therefore having the result of votes known only at the end of a meeting), and I also see the flexibility with leaving a set of voting instructions for the last session when one has to catch a plane early on the last day afternoon.

But there is a lack of symmetry that is bothering that results from using IMOVE to compute Individual ballot quorum: 
votes that happen early in the meeting benefit from a small quorum, votes that happen late in the meeting suffer from a large quorum (which includes people that have left and are unlikely to have given voting instructions). 

 Maybe the individual ballot quorum should be defined with respect to the number of abstain only, and not as a function of IMOVE institutions that may leave without setting voting instructions. What about redefining rule 7. as "abstain votes must not represent more than 1/4 of total votes on that particular ballot" ? 


I also still think that the 7. rule of less than 1/4 abstain is too high. the 3/4 of yes/no is already ensuring strong consensus among people who care. As is, that 7 rule is just a convoluted rewrite of the rule that abstain/no are the same thing (to pass, a ballot has to get 1/2 of the IMOVE quorum). This is especially problematic if the definition of IMOVE includes a lot of people who cannot cast votes because they are not present anymore physically. But even in general terms, it may be interesting to have "abstain" be a vote that has less impact than "no" on the success of a ballot, a flexibility given by setting the bar of abstain to less than 1/3 or 1/2 (or any fraction in-between).


Le 16 janv. 2013 à 15:24, "Jeff Squyres (jsquyres)" <jsquyres at cisco.com> a écrit :

> We had a very good discussion about the proposed MPI Forum voting rules at the meeting in December.
> Attached is a new version of the document from me/Brian/Martin with all the feedback from the December meeting: changes are listed in red.
> The biggest change is that we added a chapter 3: Suggestions for Voters.  It's a 1-page bullet list of things that people voting should take into account when voting on proposals.  We also moved the concept of time limits on proposals to this chapter (so time limits are now no longer binding -- they're just a suggestion).
> Please let us know your feedback on this document.
> Martin: can we get some time to discuss / vote on this document next meeting?
> -- 
> Jeff Squyres
> jsquyres at cisco.com
> For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
> <paper.pdf>_______________________________________________
> mpi-forum mailing list
> mpi-forum at lists.mpi-forum.org
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

* Dr. Aurélien Bouteiller
* Researcher at Innovative Computing Laboratory
* University of Tennessee
* 1122 Volunteer Boulevard, suite 309b
* Knoxville, TN 37996
* 865 974 9375

More information about the mpi-forum mailing list