Harvard-2017: Elements of the S4 Collaboration

Back to Harvard-2017 main page

Elements of the S4 Collaboration (Moderator: John Carlstrom)

[charge to moderator: find 2-3 examples to display as kickoff]

Post talks here.

• Collaboration Elements by Example: PPT, PDF -- Brenna Flaugher
• Governance structures -- Karen Byrum [PDF] , [PPT]
• Membership policy + mentoring -- Steve Kuhlmann PDF PPT
• Publication + talks policy -- Nathan Whitehorn PDF Discussion Intro
• Wrap Up -- Interim Planning -- Suzanne Staggs PDF, PPTX

• Prep for Parallels 2 -- science updates for Decadal: Colin, Cora, and Shaul PDF

Notes from session

• 3:30PM Collaboration Elements by Example - Brenna Flaugher
  • Collaborations grow and evolve, policies are to guide that evolution, not to set in stone what will happen infinitely in the future
  • DES example:
    • DES started with ~20 people at 5 institutions, now 400 scientists at 25 institutions
      • Q: what stage was DES at then compared to where we are now?
        • behind where we are now, because it was a brand new idea.
        • CD3 in 2008
    • DES had a single head (John Peoples) but management committee (with reps from each institution) was very important for the collaboration feeling like they were involved and represented.
    • DES has a brand new org chart now that it's in operations
      • Q: Were decisions made by the management committee binding?
      • Yes, all decisions went to them, and their decisions were binding. Director was chair of that committee but did not have veto power.
    • leadership roles are one of the big perks of a large collaboration and there needs to be a good process for filling them.
      • consider two people in each position and rotate one each year
      • volunteers / self-nominations
  • DESI example
    • forged from two competing groups (DES / SDSS)
    • institutional board created from *all* interested groups on both sides
      • started with ~75 people but quickly decayed
    • institutional membership and contributions reviewed every year
      • initially, this was mostly money, which is useful to convince DOE you're serious
  • LSST example
    • NSF / DOE division similar to what might happen for S4
  • CMS example
    • collaboration board has one member per region/country
  • takeaway message: big collaborations don't just happen, you have to work on them
    • flexibility is key
    • funding sources need to be reflected in org structure (agencies need to be represented)
    • need to be cognizant of postdoc & grad student timescales, incl. how to make transition between junior & senior roles
  • Questions
    • have you seen examples of good and bad leadership?
      • a good team works even when leadership is questionable
    • can you distill into a few core elements the structure that DOE and NSF will impose on us?
      • DOE won't impose a structure but has strong expectations for management
    • How did NSF accept separate projects for LSST?
      • No, it was multiple science collaborations, but still one project.
    • In these examples, how many leaders were elected directly?
      • DES does not elect directly, CDF did, DESC does
    • Did the division between the two camps in DESI persist throughout?
      • no, it pretty much died out over time

4:00PM Governance structures - Karen Byrum three examples of by-laws: IceCube (~300 members, 48 institutions, 12 countries); Mu2e (~230, 37, 6), DESI (530, 79, 9) by-laws define the role of the institutional board, which is the policy-forming body by-laws should define a series of things about the IB: who can be on the IB & how selected? voting process how to select spokesperson and/or chair? how frequently to meet? by-laws define the role of the executive board / committee (advises scientific leadership) by-laws define scientific leadership (spokesperson) more than one spokesperson? deputy? need one from each region / sub-field / etc.? how are they nominated, how do they accept? terms, elections, etc. Questions In IceCube, there is another informal committee (international funding agencies) that's project, we are talking science collaboration could be a resource board Will there be conflicts between project leadership and scientific leadership? needs to be strong communication between the two, including possibly science representation on project leadership Is there a clear demarcation between what project leadership and scientific leadership do? (E.g., who is in charge of producing maps?) project ends when instrument is build, then you're in operations so project management ends then? no ? lots of questions about project management...

4:25PM Membership policy + mentoring - Steve Kuhlmann What does "member" mean? senior-ish person, possibly with a required minimum %FTE on the project, often with associated postdocs & students when postdocs move on, usually given a new title and remain in the collaboration, but that new institution is not given a spot on the IB IB is final word on membership (sometimes 2/3 vote required) often have membership subcommittee as initial filter stay flexible! membership and publication policies can be coupled but don't have to be preference usually given to groups, not individual people at institutions one-person institutions often "associate members" (can band together and get representation) specific membership policy examples: IceCube (short) & DES (long) buy-in helps to buy stuff because money doesn't really turn on until CD2 example of in-kind contributions that can replace cash buy-in (DES document) external collaborator concept unique to DES? non-members can apply to use some part of data builder idea normal membership not always guarantee of authorship, but large infrastructure effort is nine questions about membership policy (see talk) Questions Do you have to have an IB? never seen a collaboration without one DESC does not have one, but it's a different beast Is cash buy-in really practical? Do you distinguish between people who pay a lot of cash and people who do stuff? most institutions on DES are only ~1/3 cash on DESI, any DOE institution could come in with no cash When does the buy-in happen? We don't have a project yet. institutions do buy in before approval, often at a discounted rate because of higher risk, but buy-in makes more sense for a defined project How do CMB collaborations work now, given that none of them have bylaws? perhaps there is a threshold of collaboration size beyond which you need them funding agencies know how to deal with collaborations, not amorphous groups of people

• 4:50PM Publication + talks policy - Nathan Whitehorn
  • What are the goals?
    • good science disseminated widely
    • appropriate credit assigned
    • protect younger people
    • somehow recognize critical dirty work (e.g., calibration)
  • Panel discussion with representatives from three large collaborations
    • Joe Formaggio (SNO, Project 8, KATRIN)
      • the only important thing the collaboration board really does is write down the publication policy
      • maintain member list (6 months to be added, 1 year off the project to be dropped)
      • different classes of papers (collaboration-wide, official but not all authors, not official but by collab members and usually circulated)
    • Charles Lawrence (Planck)
      • similar breakdown in papers
      • authorship was never automatic (had to say "I want to be an author")
      • "Planck Scientist" like builder on DES (can be on any paper)
      • always alphabetical, can be no other way
      • Planck talks policy didn't work, so not saying anything about it
    • Brenna Flaugher (DES)
      • DES has key projects and non-key projects with different policies
      • key papers are alphabetical, pretty much anything else can be first-author
  • Questions
    • How do you get to lead an analysis project?
      • on DES, need approval from working group
      • DOE wants a unified public front
    • What about key projects?
      • WG decides
        • so if you do a really good job, you are more likely to lose first authorship?
        • no, you can still write a first-author paper about part of the analysis
      • in Project 8, key analyses are identified at collaboration meetings
    • How well did these policies work in getting people credit?
      • what worked in Planck were written letters of recommendation; what didn't work was the talks policy
      • in SNO, credit was assigned by collaboration board, and it was important to cycle those people

Action items/Next steps

Summarize action items here