[Moims-dai] Evolution of the OAIS RM

[David Giaretta]

Hi Don

 

See my comments below.

 

..David

 

From: MOIMS-DAI <moims-dai-bounces at mailman.ccsds.org
<mailto:moims-dai-bounces at mailman.ccsds.org> > On Behalf Of D or C Sawyer
Sent: 10 February 2019 16:14
To: MOIMS DAI List <moims-dai at mailman.ccsds.org
<mailto:moims-dai at mailman.ccsds.org> >
Subject: Re: [Moims-dai] Evolution of the OAIS RM

Hi David, 

I have read your comments to my brief paper and I am, frankly,
disappointed.  I was hoping that you may have abandoned your efforts to make
recursion a fundamental underpinning to the OAIS conceptual model because I
don’t think this helps understanding of the model nor do I think it fits
with how most Archives see what they do,. 

Recursion has been baked into OAIS from the very start. For example,
Provenance Information, Representation Information, Content Information etc
are all special types of Information Objects. It has been that way from
version 1 of OAIS. Therefore, all the things we say about Information
Objects apply to Provenance Information, Representation Information etc.  As
I said, it has been this way from the start. I am not making this up – check
the standard. We are explicit about recursion for Representation Information
because it allows us to explain Designated Community. We did not explicitly
discuss recursion for Provenance Information because we thought we did not
have to in version 1, but during the latest update it became clear that it
is beneficial to make it explicit. 

Therefore I also don’t think it is beneficial to auditors and will not be a
basic framework that can underlie their thinking. I believe frameworks that
Archives find productive will also be the frameworks that auditors will find
useful.  Good archival practice must be practical and not based primarily on
an abstraction that has not been proven to be useful.  We’ve defined RepInfo
to be a recursive concept because this fits the reality of RepInfo.  I
submit that most archives do not see what they do as  being based on a
recursive concept. It is not enough to say that one can implement a
recursive framework. It must be practical to populate and use, in my
opinion. I suspect you are, at this point, in wide disagreement but please
read to the end where I think I highlight the basis for our disconnect.

 Our extensive public and private discussions, together with your response
to my brief paper, is consistent with your continuing effort to incorporate
more recursion into OAIS as a basic conceptual framework. 

As I mention above, I am not shoehorning recursion into OAIS – it has been
there from the start.

The strident tone of a number of your comments, such as your opening
comment that my paper “suffers from a number of fundamental
misunderstandings and fatal flaws” seems intended to dismiss discussion of
the issues I’ve raised.  Maybe it is just that you think your views, and
intent, are so clearly correct that anyone should see and agree. In good
conscience, I can not.

I certainly believe what I am writing is correct; it was extensively
discussed in the weekly telecons and on the review site. Moreover I have
pointed to what has been in OAIS from the start which supports this.

The one factual error in my paper, which I noted in my response to Mark, is
the statement that PDI is no longer applicable to non-digital Content
Information. This sentence should be considered removed, but this has no
impact on the points I’m making.  The simple truth, by definition, is that
PDI is now only applicable to the CDO in the conceptual model and this is a
very major change to the conceptual model. 

This is a problem only because, as I describe below, you require the CDO to
come from a Producer which is external to the OAIS – which is wrong, as the
definition from version 1 of OAIS has made clear.

This is not saying anything about 16363 or what any given archive actually
does.  You say that the note following the PDI definition in the revised
OAIS RM is meant to clarify the definition. It does no such thing.  As I
said in my paper, the note clearly says that the general concepts of
provenance, fixity, etc. are applicable to the RepInfo and other
information.  This is properly formatted in the note, using lower case,
because the note can’t contradict the PDI definition by using upper case.
Surely you see this.

You also say that the previous applicability of PDI was clearly being
misunderstood by archives, so this is a clarification. I have to wonder how
many archives actually were confused by PDI applicability, but I do know
that some key members of this working group misunderstood it.  

As I explained previously, the misunderstanding was that, for example,
archives calculated digests, which they kept as Fixity Information, based on
the Digital Object, not the Information Object. This makes practical sense,
and what the changes do is to recognise the logic of that understanding. 

This is what caused the PDI to be restricted to the CDO in the revision.
However once actual applicability was understood by the group, the proper
response, adhering to the guidelines for updates which you attached at the
end of your comments, would have been to make the clarification and not to
continue with the major concept revision. This clarification could have been
accomplished with a sentence of two. While there was the misunderstanding
within the group about PDI applicability, one could consider the major
revision to also be a clarification, but NOT once the misunderstanding was
erased.  As I noted in my paper, the result is a very weakened conceptual
model as PDI is now only associated with the CDO.  This is so obvious that
it makes one wonder why the major concept revision has been retained,
especially given the difficulties it has raised (e.g., the note, upper and
lower case).

My view is that this major revision is still in the OAIS model primarily
because you still want to reach a recursive basis for thinking about the
conceptual model. By making the PDI only applicable to the CDO, this is a
step in this direction as I explain below. My view is reinforced by your
interjecting, into my brief but standard view of the OAIS functional model,
the view that the OAIS can also be the Producer, by definition.  However the
only place this is actually addressed in the text is when the OAIS has
Content Information that needs updating and thus it submits the updated
Content Information to the Ingest Function so it can get back to Archival
Storage.  Perhaps you’ve forgotten the extensive email (private
communication) I sent you on these points.  This is the standard view that
anyone reading the whole document, and not just looking at a few
definitions, comes to.  

There are many things we do not spell out explicitly – otherwise the
document would never end. The definitions were worked on extensively and
provide the foundation of OAIS. You write as if the descriptions of the
Functional Model is required for conformance to OAIS – which it is not.  

A standard view that readers come to, and that thus facilitates
understandable unambiguous communication, is the basis for the development
and maintenance of the OAIS RM. Instead you have argued, many times in the
past, for treating the Content Information as ‘any’ information that the
archive wants to ‘preserve’ (not defined). If adopted, as you’ve previously
argued, you could then treat any information, such as RepInfo, PDI, etc. as
Content Information and its CDO would get PDI.  Thus there would be no need
to have PDI directly associated with RepInfo because the RepInfo can be
viewed as Content Information and thus get its PDI this way. This has been
your recursive view.  You should see that this would be a disaster for clear
unambiguous communication. I believe you have become so enamored with this
recursive view that you have lost sight of the clear communication objective
for the conceptual model. (Again, this says nothing about 16363 or how a
given archive implements its functionality.)

Based on our previous email exchanges your fundamental error arises from
refusing to read the definition of Producer – the OAIS itself can play the
role of Producer – so you insist that the Producer is external to the OAIS –
which is your initial fundamental error. This then leads you to write that
CI must come from outside that OAIS – another fundamental error. This then
leads you to write that PDI etc created by the OAIS cannot be regarded as CI
– hence this is also a fundamental error. 

 As further evidence, I note that you have interjected, in two places
within my paper, bullets describing general things the archive needs to do
to preserve (your term) some information and then you conclude that these
are, by analogy, associating PDI (actually pdi unless you take any/every
digital object as being a CDO). Of course I don’t object to anyone thinking
this way, and seeing some of the relationships, but it is not anywhere close
to the current conceptual model as anyone would understand it. It is also
not something that I think could be developed into a clear communication
model and it suffers from implying that every piece of information gets the
same level of attention within the archive.  I think one would have to try
to develop such a model from the beginning to see if it was useful.  I
believe trying to retrofit this type of thing into the OAIS RM would destroy
its utility.

The point I was making is that, for example, Provenance Information must be
preserved, otherwise how can the OAIS provide evidence about Authenticity.
You wrote that this would mean that “every piece of information gets the
same level of attention within the archive”. Well if things have to be
preserved then some effort would be needed but, as I pointed out in our
email exchange, you seem to have a very file based understanding of AIPs
whereas, as I pointed out, the important point is that the OAIS knows where
all the components of the AIP can be found, because the AIP is a logical
construct. As I point out in my comments in your paper, if an archive does
not know where the Provenance is, is not sure that it has not changed, who
can access it etc etc, then it is surely not able to fulfil the mandatory
responsibilities.

 I’m sure that we’ve been having such disconnect because you have this
broad recursive view that you feel can be used to help auditors focus on
what is needed – a kind of short hand concept way of looking at things.  I’m
focused on the function of the OAIS RM for clear, unambiguous, communication
and thus I stand by my paper as written, apart from that one erroneous
sentence.  It seems very clear to me that one can’t retrofit your recursive
view into OAIS without destroying its primary function – clear
communication.  I don’t think any of us want to destroy that, and thus I
hope you rethink your approach. 

The point I keep making is that recursion is not something I am forcing
into OAIS, it has been there from version 1, and I continue to be surprised
that you do not realise this. One reason that recursion is in OAIS is that
it does enhance clarity because there is a small set of concepts that can be
applied over and over again – one does not need to invent lots on
increasingly abstruse new details.

 

Cheers-

Don

 

On Feb 1, 2019, at 12:00 PM, David Giaretta <david at giaretta.org
<mailto:david at giaretta.org> > wrote:

 

Hi Don

As you will see in this marked-up document, you have made a number of
fundamental errors.

I agree with the general point that we need to keep the process under
continual review. But I believe we have applied the correct procedure in
this review and your concerns about the meaning of the changes is misplaced.
Regards

..David

-----Original Message-----
From: MOIMS-DAI <moims-dai-bounces at mailman.ccsds.org
<mailto:moims-dai-bounces at mailman.ccsds.org> > On Behalf Of D or C Sawyer
via MOIMS-DAI
Sent: 29 January 2019 04:24
To: MOIMS DAI List <moims-dai at mailman.ccsds.org
<mailto:moims-dai at mailman.ccsds.org> >
Cc: D or C Sawyer <Sawyer at acm.org <mailto:Sawyer at acm.org> >
Subject: [Moims-dai] Evolution of the OAIS RM

Dear All,

I’m attaching this short paper, addressing what I see as a deficiency in
the current 5-year review process, in an effort to improve the quality of
the next 5-year review.

<Evolution of the OAIS RM - with major comments from David
Giaretta.docx>_______________________________________________
MOIMS-DAI mailing list
MOIMS-DAI at mailman.ccsds.org <mailto:MOIMS-DAI at mailman.ccsds.org> 
https://mailman.ccsds.org/cgi-bin/mailman/listinfo/moims-dai

 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: winmail.dat
Type: application/ms-tnef
Size: 11998 bytes
Desc: not available
URL: <http://mailman.ccsds.org/pipermail/moims-dai/attachments/20190211/8424be76/attachment-0001.bin>