Meeting with Marshall Brinn

Marshall indicated that he liked the design document, and thought it was a good start.  He hadn't had a chance to review it in great detail, but he had gone over it to some extent.  He noted that we would have to be careful in identifying TIEs, since the directions ALP as an individual program might want to go in might not necessarily be those where there was the most synergy with CoABS.

At Brian's request, I briefly reviewed the background of the CoABS program, its general goals, status, as well as the goals, status, and timeline of the ALP-CoABS integration effort I'm working on, and the general approach I've been taking.  I reiterated that the intent was that the actual work on implementing any TIEs was to be done by the actual participants in the current programs, not by third parties (although Todd and Jim will make the actual decisions on how to proceed).  I also noted that the current design document was just trying to identify potential TIE opportunities and project relationships, but that the end product was intended to be a true "design" of several TIEs that (a) are feasible and (b) would, when performed, demonstrate that something useful had been learned/accomplished.

Brian briefly reviewed where the ALP program is.  He said they'd demoed 80 clusters, this year were going to go to 260, but this is only a drop in the bucket relative to the 10000 or so total number of logistics-related organizations to be modeled.  He said their clusters model some organizations with good fidelity, but others are modeled only approximately (for demo purposes).  Much of the work on "clusterizing" particular organizations (and the systems they used) had been done by the individual logistics organizatons.

Brian suggested that at some point a meeting with the key CoABS participants and himself would be appropriate to nail down details of TIES.  I suggested that the next CoABS PI meeting (in July?) might be one place to do this.

He commented that the CoABS projects cover a lot of technical ground.  I said I was using two general approaches to narrow the scope:

• separate the architectural issues from those focused on specific single-agent concerns or applications and concentrate on the architectural ones
• look at the issues already being focused on by the CoABS TIES (as being among the more important ones)

In response to a question of mine, he said that the way ALP envisioned interoperating with agent-based command systems was that those would be "clusterized" too.  In interoperating with non-clusterized agent systems, there was a question of "who wraps who first", but he noted that each system can wrap the other for interoperability purposes (providing gateways).

Brian noted that there were some details of ALP that didn't seem quite correct in my writeups, and that the Plugin Developers Guide (PDG) I was using had been superceded.  For example, clusters (in particular, the expander, allocator, and assessor components) have no smarts at all (no rules);  instead, rules are entirely within plugins, and plug-ins grab tasks as they arrive, rather than clusters handing them out to the plug-ins (this can create some difficulties, and control of that process is currently an issue).

Brian asked what I wanted from him, and I said at this point, mainly:

• answers to specific technical questions, and
• comments on TIE-related ideas I bounce off him

He is extremely busy (our meeting was interrupted twice, and had to be cut short).  We agreed that we could handle the rest (of what I wanted at this meeting) by email or phone.   My immediate next steps will be to send him some additional administrative questions about CoABS access to ALP and ALP information (and also to get a more recent PDG).