[Mpi-forum] MPI user survey - what purpose?
Underwood, Keith D
keith.d.underwood at intel.com
Wed Nov 18 11:07:43 CST 2009
We do need to remember what our options are here... We can do a survey that is possibly biased by "the answers we want", or we can get our feedback from the people coming to the forum who will each personally bias the user feedback based on the answer they want, or we can do this in the absence of feedback...
Does somebody see a 4th option that is actually good?
> -----Original Message-----
> From: mpi-forum-bounces at lists.mpi-forum.org [mailto:mpi-forum-
> bounces at lists.mpi-forum.org] On Behalf Of Bronis R. de Supinski
> Sent: Wednesday, November 18, 2009 7:37 AM
> To: Main MPI Forum mailing list
> Subject: Re: [Mpi-forum] MPI user survey - what purpose?
> I agree 100% with Dick's statement that survey support
> will not be a convincing arguement to move forward with
> large (or, IMO, small) issues. At most, we might get a
> few "Hey, we didn't think of that" ideas or issues out
> of this survey. Anything further would be a mistake.
> Feedback is useful but it must be filtered.
> On Wed, 18 Nov 2009, Jeff Squyres wrote:
> > Thanks all for the comments about the survey. Apologies that I have
> > not replied earlier; I have read all of the feedback and am making a
> > final whack at the on-line survey right now. I'm at SC09 this week
> > and my email access is limited to extremely late in the evening and
> > extremely early in the morning.
> > It's hard to disagree with what has been said; I'm in complete
> > agreement that surveys are very, very hard to do properly. At the
> > same time, perhaps it's just the optimist in me that feels like we
> > need to try to get some feedback on the "big" issues. We're making
> > MPI-3 for what people want to do, right?
> > I do tend to agree with Dick that some of the Big Complex Issue
> > questions may be of limited value far all the reasons previously
> > discussed. Perhaps they should turn into general text box answers
> > rather than a simple number evaluation -- that might give [slightly]
> > more useful results by forcing the user to *say* something
> > On Nov 18, 2009, at 6:16 AM, Richard Treumann wrote:
> > > I would like to hear exactly what the advocates of doing a survey
> > > hope to gain from it.
> > >
> > > A survey about what people use MPI for today might give useful
> > > information if the survey is well constructed. One that asks
> > > questions about what they expect to want to do over the next decade
> > > could conceivably give useful information though I think it would
> > > hard to construct well and hard to interpret.
> > >
> > > Asking questions about complex technical options by giving very
> > > brief overviews, each crafted by a person who knows what answer
> > > want seems to have almost zero chance of giving useful feedback. At
> > > most it gives the person with a particular agenda a way to argue
> > > that his agenda has strong support. Of course, in most (or all)
> > > cases, if the survey question had been crafted by a skeptic of that
> > > agenda, the support would not be seen and the argument would be
> > > the public is clearly not on board.
> > >
> > > In reality neither question tells us how the community will respond
> > > to the ultimate content of the MPI 3 standard. Responsible,
> > > professional polling organization have highly trained specialists
> > > crafting supposedly balanced questions and still freely confess
> > > a favorable response to question is not the same as genuine support
> > > for the underlying issue.
> > >
> > > I, for one, will not give it much credence when somebody brings
> > > forward a 20 page extension to the standard and says: "We must vote
> > > this in because my 20 word question about it got 85% favorable."
> > >
> > > Dick
> > >
> > > Dick Treumann - MPI Team
> > > IBM Systems & Technology Group
> > > Dept X2ZA / MS P963 -- 2455 South Road -- Poughkeepsie, NY 12601
> > > Tele (845) 433-7846 Fax (845) 433-8363
> > > _______________________________________________
> > > mpi-forum mailing list
> > > mpi-forum at lists.mpi-forum.org
> > > http://*lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
> > --
> > Jeff Squyres
> > jsquyres at cisco.com
> > _______________________________________________
> > mpi-forum mailing list
> > mpi-forum at lists.mpi-forum.org
> > http://*lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
> mpi-forum mailing list
> mpi-forum at lists.mpi-forum.org
More information about the mpi-forum