+ grid + BOINC could be 2 things:
 - condor stuff -- execution, taking advantage of oxford desktop condor pool
 - data grid side -- how to distribute results

+ backfilling the EGEE grid w/ BOINC jobs

+ using BOINC as a way to harvest spare cycles at CERN
 - more control than volunteers
 - more I/O capacity
 - therefore, more options for the BOINC jobs you can run

+ most CERN applications are linux, most desktops are windoze
 - virtualization
 - emulation
 - not strictly BOINC or Condor per se, but related

+ Baker lab uses Condor as a front-end
 - hook into the BOINC stuff
 - treat BOINC as another grid resource to send work to

+ getting BOINC into fortune 500 companies
 - "nothing in it for us, everything to lose"
 - how to craft a good story for them?
 - enterprise grid option?
 - IBM partners with all sorts of groups for enterprise grids (UD, Globus)

+ using already installed BOINC base as a grid expansion
 - BOINC has limited apps, but grids can be any app.
 - changes to BOINC to really accept those apps.

+ BOINC & security has big implications for grid/BOINC pipelines

+ multi-purpose grids vs. BOINC-specific stuff

+ BOINC is light-weight server
 - can install it all locally, nothing on the internet at large
 - took about a day to setup
 - on-going support? -- not much

+ Condor and BOINC are more of a happy family than BOINC + Globus?
 - seems like general impression (carl)
 - LHC@Home was started as masters project -- official topic was BOINC + grid
 -- looked at mapping grid into/out-of BOINC
 -- this is much more complicated
 -- published master's thesis on this
 -- the approaches here (Condor backfill, and EGEE backfill) are much easier
 - CERN's BOINC people are quite supportive
 -- EGEE/BOINC backfill supported by high up in CERN (erwin laure - CTO EGEE)

+ british e-science grid:
 - originally seen as a renegade project
 - but now, one of the only groups that gets useful work done

+ academic grid world:
 - lots of tensions between different grid systems
 - inter-operability the new buzz-word
 - accenting common issues that effect everyone (like virtualization)

+ concrete steps:
 - write up these efforts, present to Miron + Laure

+ have the startd as a BOINC application?
 - need to have a self-contained BOINC bundle for the work-units
 - what about security and trust?

+ adaptor that takes work-units as a Condor job?
 - that's what the CERN master's students were trying to do, though not w/ Condor

+ the whole BOINC + MW thing as previously discussed
 - have MW get WUs from BOINC, then farm them out to a grid (or glide-in)
 - world community grid would be very interested in this stuff

+ BOINC WU's on the grid:
 - EGEE-style backfill
 - Condor-G-enabled EGGE-style backfill
 - BOINC + MW sharing work, w/ MW talking to glide-in/condor-g pool
 - stripped down, scripts to glue WUs into a form that grid resources will eat
 -- would this have to be application specific? [no, could be generic]
 -- really stripped down stand-alone app manager
 -- provides a subset of the services the BOINC client does, but grid-enabled
 -- BOINC API already detects if there's no BOINC client and will run stand-alone