These notes respond to Henry Pisciotta's recent questions 
about usage data RLG will supply for the AMICO testbed, and 
his illuminating statement of Carnegie Mellon needs and 
recommendations. They incorporate some results of our 
discussions with David Bearman and Jennifer Trant in December:
-  RLG won't impose any limitations for accessing the
   different reports; AMICO will link to reports from testbed
   web site
-  reports will be broken out by institution
-  works will be identified by AID rather than title

Considering the specific topics Henry raises:

Timespan - Reports will cover the period starting with the 
August date when the Testbed became publicly available.

Frequency - There will be monthly reports for each month of 
the period, even retrospectively.       

Distribution/Share more information with universities - All 
the reports for each institution will be available to any 
institution, through the AMICO web site.

Cost per use - The monthly "Report on Searching" (3.2a) for 
each institution will meet the requirement of showing the 
number of searches from as a basis for calculating cost per 
use.

Zero results - This is another topic that came up in our 
December discussions. We will provide such a report, monthly, 
broken down by institution.

Confidentiality - It's reassuring for us that Henry doesn't 
see any objections at Carnegie Mellon.  The fact that the 
downloads aren't identified with particular users (or even 
identifiable sessions) may also be helpful, from this point
of view.
     
Use by type of institution - We'll break reports out by 
institution, which should solve the problem of distinguishing 
contributor behavior from end user behavior.  (The distinction 
between "Endeavor" and "Zephyr" logs isn't relevant here.)

It's always helpful to have actual user requirements 
articulated, and we know that can be difficult until there is
something to react to.

Arnold Arcolio
RLG