Economic analysis of software

(Changed subject line as we were well past the initial topic ;-)

On Wed, 29 Oct 2003, Carlo Daffara wrote:

> There no (sensible, at least) way to measure use value for internally
> developed software; what various groups (OECD, various ICT
> observatories) have done is to measure various proxy indicators, like
> workforce employed in SW development outside of the boxed software
> industry, and multiply for adjusted average salary (plus overhead).

  I am wondering what these measurements based on proxy indicators really
tell us? I've been looking for a more researched number to indicate what
I've been using the 95%/5% figure for which is just based on "empirical
evidence" of the number of lines of code (quantity, not quality).


  Once you get into proxy indicators don't you end up with the same
problem that the BSA/CAAST does with their so called "piracy studies"?  
Their use of proxy indicators has lead their studies to measure more their
level of bias than anything useful about what they claim to be measuring.  
For example, their methodology can't objectively differentiate between
machines with so-called "pirated" software and machines with FLOSS or
other competing software to their members.
  http://weblog.flora.ca/article.php3?story_id=379


> Some articles and discussion are available from
> http://www.unece.org/stats/documents/2002.04.sna.htm (I suggest to start
> with http://www.unece.org/stats/documents/ces/ac.68/2002/12.e.pdf to
> have an overview).

  Thanks for the links.  I am wanting to pass this on to a colleague at 
Canadian Public Works that is working on full cost accounting 
methodologies for software.  I'm still not sure how much of this fits in 
or otherwise includes FLOSS.

  As an example, trying to find the value for Bellanet of their Online
Proposal Appraisal <http://www.bellanet.org/opa/> would not fit adequately
into any of their 4 categories of software, or any combination I can think
of.  This was an in-house developed application that because it was
released as FLOSS has had considerable outside resource amplification.  
How does one account for that resource amplification in dealing with the
fact that the value of the software increased for Bellanet at a level that
considerably outstripped resource input?

Note: I have a few other Government of Canada FLOSS projects linked at 
http://www.goslingcommunity.org/links.shtml which would have a similar 
problem being measured.

  For anyone reading this, I am wanting to list any other Government of
Canada projects you may know about!  Please let us know.

> Various economists I have talked with (remember: I am a poor engineer :-))

*grins*  I too am just a computer consultant, not an economist or lawyer 
which are areas I seem to get into conversations about.

> tend to avoid the use-value/sale-value comparison, because there is no way
> to relate them together; sale value (even of FLOSS, when substituting sale
> with the personnel pay or the consultancy) can be directly observed on
> balance sheets.

  This still sounds like it is measuring software cost, not software value
which is an entirely different thing.  You want value to go up, but
software costs to go down.

  Is there really an objective relationship between cost and value in
software when there are many competing methodologies involved?  We seem to
be moving into some of the same problems as the TCO studies which treat
software as a one-way procurement of a manufactured product rather than
knowledge development projects which one can participate in peer
production methodologies (open collaborative models for the production of
public goods, whatever you want to call them).

  This problem was written about in "Why Free Software's Long Run TCO must
be lower"
http://www.members.optushome.com.au/brendanscott/papers/freesoftwaretco150702.html
which spoke of "Total Cost of non-Ownership".

> I have extracted data from UNECE, OECD, EITO, some smaller observatory
> around Europe, and some articles; I tried to un-bias my European view
> integrating data from US department of commerce.

  I don't believe you gave us a reference to your paper yet, just the
number you used to start this sub-thread.


On Tue, 28 Oct 2003, Carlo Daffara wrote:

> My estimates (based on total economic value) place 2001 percentages of
> boxed versus custom/inhouse software at 20% up to 24% maximum.

> cheers
> Carlo Daffara

P.S.  I went looking for your messages in the archives at
http://www.ssc.com/mailman/private/gov-list/2003-October/thread.html#276
and can't seem to find them.  Are you a member of the list with the email
address you are replying with?  I suspect other participants in this list 
are just seeing my replies and not your original messages.

---
 Russell McOrmond, Internet Consultant: <http://www.flora.ca/> 
 Governance software that controls ICT, automates government policy, or
 electronically counts votes, shouldn't be bought any more than 
 politicians should be bought.  -- http://www.flora.ca/russell/