Hi Cal, I understand the points you say but I would like to emphasize that Geant4 has made just one large production during just 2 weeks in december last year. At this time nor Alice, ATLAS or CMS were performing any large production. Then the only thing done is to install their last software version but this means to run a 10 minutes job per site. And this was done actually not to prepare a new production but to test sites, the software area was checked after the update to LCG2.4.0, the/opt/edg/var/ area of the CEs, the compilers (some sites have not installed g++ or basic X11 libs which can be needed by any other experiment). This installation was done by me and I am dteam. Again the reason to assist us to test the sites with their software here appears. In somehow is to take a certain software that we think is addapted to test the sites. We take Geant4 and we run it as dteam it does not mean that Geant4 has to become VO to run tests performed by me as dteam. And second, relative to the Geant4 production, at that time (december 2004) Geant4 was checking the performance of Grid for their needs, so if they wanted just to perform some tests to see the new environment I do not see any against to do it as dteam, not only because they are testing the Grid, but because me as dteam will use their software to test the status of the WNs and because Geant4 is a lower lever tool for all experiments. Many of them use them as basic tool, in a site with ATLAS, Babar, Biomed, CMS, LHCb areas I do not see so strange to have 1 Geant4 installation instead of 5 installations of the same software. Do not see Geant4 as a collaboration making full productions on a continuous way during the whole year, because it's not their situation. Full production has been run just once, in december 2004 during just 2 weeks. Now some small jobs are running to perform some tests again at low energy. The rest of Geant4 jobs are run by dteam persons eventually just one job by site and just to test the performance of the WNs with a good example of a software which can be run by any other experiment. Another question is if Geant4 begins a whole production as another VO, but this is not their case. Cheers, Patricia