The two proposed models seem to have very different scope and goals, they don't seem to be two alternative approaches to the same goal. I think this makes them challenging to compare.

It was mentioned during one of the earlier conversations that the multiple item model helps solve some long-standing challenges. What are those? Does the multiple holdings model not address them?

Which of these models do folks want tomorrow? Which do folks want in a few years time? Is the answer to both of these questions, the same or different?

How important is modelling bound-with / analytics relative to ease of use for independent items (which I imagine are the majority)?

I think it is important to note that these models effectively take inventory (and closely related areas like circulation) down two different paths. It is possible to combine those paths later, though that will have the complexity of both paths.

What important aspects have I missed from my summary of the models below?

Models

Multiple Holdings for an item

Emphasises that a bound-with item is a single unit of circulation and that the copies of the bibliographic entity within it are not separate.

Characteristics

• Items may represent either a single copy of a bibliographic entity or a set of them bound-with each other
• There is no way to describe information specific to each copy of a bibliographic entity
• Identifying whether an item is bound-with or not relies on inferring it from whether an item is related to one holding or many
• Is a structural change to the relationship between items and holdings records
• Any user or process which interacts with items in any way will need to change

Questions

• Does this model fulfil the needs of analytics?
• Which holdings record should be used when determining effective location, call-number etc? If the system picked at random, would that be acceptable?
• Which instance should be used when determining bibliographic information? If the system picked at random, would that be acceptable?
• Can an item be associated with no holdings?

Implementation Characteristics

• Effort is spread across any parts of the system that interacts with items
• Requires breaking compatibility changes in inventory
• Teams are (hopefully) familiar with making breaking compatibility changes and how to manage the impact of them
• Maintenance effort is likely related to
  • how many parts of the system need to infer whether an item is a bound-with or not
  • how important it is to have specific information about copies

Multiple items for a bound-with

Emphasises that the copies of bibliographic entities within a bound-with are separate, whilst trying to retain that they cannot be circulated handled independently.

Characteristics

• All items represent an individual copy of a bibliographic entity
• Some (either independent or primary) items also represent a unit of circulation
• Introduces an interdependency between items
• Fulfils the needs of analytics
• Any user or process that need to update / change items may need to be able to distinguish and understand the three different types of item e.g. they need to know a barcode of a dependent item cannot be changed directly
• Requires copying information between items within a bound-with to keep them consistent e.g. barcode, location etc
• Requires removing some the constraints on an item, e.g. barcodes can no longer be unique

Questions

• Which items can be requested, only the primary / independent, or any?
• Which items are checked out, is it only the primary or all?

Implementation Characteristics

• Effort is concentrated on
  • inventory (weakening constraints, synchronising information between items)
  • parts of the system that need to identify whether items are bound-with or not
  • parts of the system that need to edit / change / update an item
• Other parts of the system may have relied upon constraints in inventory that are being weakened e.g. only a single item with a barcode
• Complexity of identifying and understanding items which are either independent, primary or dependent items is spread through out the system
• The complexity introduced for categorising items and updating them does not reduce over time

Multiple items and single unit of circulation for a bound-with (Marc's more radical model)

Separates the ideas of copies of bibliographic entities and units of circulation. In effect, this model makes the idea that there isn't a one to one relationship between copies and units of circulation (that the multiple items model expresses) explicit.

Characteristics

• Items represent an individual copy of a bibliographic entity
• Items no longer represent a unit of circulation, these are modelled separately
• Units of circulation are made up of a set of items (independent items are effectively a unit of circulation with a single item)
• Areas of the system that rely on both copy information and unit of circulation information might be more complex
• Circulation processes would rely on the unit of circulation (they would still needs to understand items, holdings and instances to provide bibliographic information)
• Requires changing many of the circulation models (which might involve complex upgrade processes)
• Requires moving some of the attributes of an item to a unit of circulation (which will require upgrade processes)

Questions

• Is it an item or a unit of circulation that has a status (this model currently assumes the latter)? This might be especially interesting for processes like cataloguing, where all of the copies in a unit of circulation would go to that process, yet not all might actually be catalogued.
• Which parts of the system needs to understand both the bibliographic copy and unit of circulation aspects e.g. circulation, orders / receiving?

Implementation Characteristics

• Requires breaking compatibility changes in inventory (and probably circulation)
• Effort is concentrated on
  • inventory (introducing the circulation unit record type)
  • parts of the system which interact with items and care about unit of circulation attributes
• Significantly more effort required than other options for upgrading existing records to work with the new model
• The complexity is explicit in the model and may reduce the maintenance once the new model has settled in e.g. either interact with an item or a unit of circulation, don't need to be infer the two from items