To be honest, I think there are some issues with considering analysis as a service, which many people may think is part of the core function. I hope to have some blog posts that kind of relate to what you are taking about, but it may not quite be the right time to post those yet. I also am also hoping to have some more high-level discussions with a larger number of people, but I am currently focusing on maximizing agreement with my supervisor (and, in terms of the discussion slides, I think it would really be best if portions of those ideas that are agreed upon are put forth by someone more authoritative at City of Hope, rather than myself).
However, these are some things that I think are important to consider:
1) I can tell that I need to work on fewer projects more in-depth. So, I think having limits on the number of projects / PIs could be important, but that is the sort of thing that I am trying to figure out. I've heard someone suggest a limit of no less than 5% effort, but I think some sort of limit of no more than ~5 PIs may be a good idea (but I don't know exactly what that limit should be, or whether that should be considered more of a guideline). For example, if my optimum number of PIs to support is 5 but I am trying to support 10, then that is going to either affect the quality of results for all of the projects or it will be difficult to provide fair support.
2) It is extremely important that people realize the "initial" results need additional critical assessment with several rounds of discussion / analysis before preparing a paper for publication. I always say "Round1" for those initial results, and I try to make this clear. However, I think there is some room for improvement in terms of setting appropriate expectations.
3) It is important that skill sets be gradually expanded, and project leads (in the labs) need to work on projects at a pace that doesn't compromise understanding. While the eventual blog post may be a little clear, I have some notes on first-author (or equivalent) comments / corrections here.
4) While there are a lot of details that I need to sort out (and I think having a per-sample charge for existing cores may help when needing to sort out a lot of details and work on changing expectations), I would learn towards suggesting a percent effort system in the future (but whether that is most suitable for a "core" versus a "department" versus a "training facility" is something that I am not entirely sure about).
5) Offering indirect support via training (instead of direct support where you keep track of the details for everybody's project, and are involved in publication) may also be an option worth considering
6) It's importance to realize that there is a certain amount of permanence to your projects - you can have questions / post-publication review 10+ years after your paper is published.