Comparing university costs for subscripion & OA

[Peter Suber <peters@earlham.edu>]

[Ann:  Since I'm responding to issues raised by your message, and 
not issues raised by earlier messages in this thread, you may 
want to change the subject line to something like "Comparing 
university costs for subscription and OA journals".  --Peter.]

Ann:  Thanks for your reminder of Yale's per-article costs for 
subscription-journals.

The trick is to find equally accurate numbers for Yale's costs in 
a hypothetical future in which all subscription-based journals 
have converted to open access.

For example, your comparison below assumes (1) that all OA 
journals charge author-side publication fees and (2) that 
universities would pay all these fees on behalf of their faculty.

Currently, the majority of OA journals charge no author-side 
publication fees at all and, where fees exist, many more funding 
agencies than universities are willing to pay them.  Of course, 
both patterns may change.  But where we have good information 
about present patterns, it should anchor our speculations about 
the future.  Or at least we should model possible futures under a 
range of plausible assumptions rather than under one set of 
assumptions we know to be untrue to current patterns.

I'd like to see someone fine-tune the comparison in order to take 
into account that only some fraction of OA journals charge fees 
and only some fraction of these fees will be paid by 
universities.  If it's arbitrary and contentious to pick just one 
fraction for these calculations (whether today's fractions or the 
assumed 100%), then we should use a range.  The least contentious 
would be a full range like 0, 10%, 20% ... 100%.  After we do 
that, we can talk about which future is more likely.  But at 
least we'd have more than one picture on the table, including 
several that rest on more realistic assumptions.

For more detail on these comparisons, the assumptions underlying 
them, and ways to refine them to improve their accuracy, see my 
article, "Good facts, bad predictions" (June 2006), 
http://www.earlham.edu/~peters/fos/newsletter/06-02-06.htm#facts

Best,
Peter Suber

-----

At 08:38 PM 3/18/2007, you wrote:

>David:  Why do I feel we are covering old ground and old 
>inaccuracies below??
>
>Sometime in 2004 there was discussion both on this list and in 
>other venues, noting that the publishing output of large 
>research universities is considerably higher than your numbers 
>below, and thus -- if a per article fee of any size (over $900) 
>is to be charged for STM, these universities will need to find 
>considerably more funds than they are currently spending on 
>subscriptions, where the cost is shared by many readers.
>
>I don't intend here to disagree about what is the best 
>cost/price model for publishing research, but rather to repeat 
>some data from my December 2003 seminar presentation on this 
>topic, based on approximate (conservative) publications numbers 
>from Yale -- which is not by any means the largest STM article 
>producer among ARLs:
>
>Number of STM articles published (most indexed by ISI
>with an estimate for the rest):                   3,600
>(this excludes humanities journals)
>*I estimate the above number is about 10% on the low side and
>that the real number was closer to 4,000)
>
>STM journals budget that fiscal year              $3.6M
>
>On this basis, our per-article STM purchase
>cost was:                          $900-1,000
>
>Assuming those same STM 3,600 - 4,000 articles
>@ your $2,500                           $9M-10M
>
>@ $1,250 (which is LESS than PLoS now charges
>and also less than the top BMC journals)          $5M
>
>It's almost impossible to calculate the humanities numbers as the citation
>sources for them are much more scattered and meager, and the citations
>patterns are very different to STM. Social Sciences fall somewhere between
>the two and are not estimated above.
>
>Ann Okerson/Yale Library
[...]