Re: Open access to research worth

[Stevan Harnad <harnad@ecs.soton.ac.uk>]

On Mon, 26 Sep 2005, John Houghton wrote:

> Diamond did a decomposition of US faculty salaries and found a positive
> correlation between salaries and citations... However, to my reading, it
> is an explanatory variable not a causal variable... one cannot 'decompose'
> forwards. 

I agree. Causality has not been demonstrated by these data. However, as
citations are being counted and used, causality is likely, in both
researcher and research assessment. And their validity and predictive
value will only increase, as open access reaches 100%. Research *is* about
usage, not merely publication.

   http://www.ariadne.ac.uk/issue35/harnad/

> In the introduction to the Diamond article that Prof. Harnad points us
> to, it says: "Readers should be cautious in drawing certain conclusions
> from Diamond's article. Diamond is not saying that every additional
> citation is worth X amount of dollars. Economists are interested in the
> structure of wages and in its components, and they present their data to
> show that structure. Diamond does not claim that there is any simple,
> automatic connection between citations and salaries.  There is no real
> evidence of such a causal connection..." Salaries are not set on the
> basis of citations... a 50% increase in citation for the 85% of
> researchers that do not self-archive will not lead to them earning an
> extra 2.5 million pounds. There would probably be a lot more
> self-archiving if it would.

I agree that my figures are merely a crude estimate (though much better
than a raw publication count), and I chose the conservative end of the
observed 50%=250% citation increase partly for the reasons Prof. Houghton
adduces here.

But let us hope that there will indeed be more self-archiving, once its
effects on citation impact become more widely known.

> Prof. Harnad's second calculation (not using Diamond's numbers) relates
> the value of citations to research funding. However, as I see it, there
> is no necessary or direct relationship between the research funds input
> and the impacts (outputs/outcomes).

I agree. We are talking here merely about probability, not necessity.

> Knowledge has some unusual economic characteristics. When you increase
> the components going into an automobile factory by 10% you tend to get
> something close to a 10% increase in output (cars). However, putting 5
> people on the job would not necessarily have given you the theory of
> relativity in 20% of the time it took Einstein.  There is no simple
> relationship between the dollars in and the value of the outputs, so
> there can be no simple relationship between RCUK expenditure and the
> loss from potential citations.

Correct. But usage is a far better measure of how much the RCUK is getting
for the money it spends on research; and not self-archiving loses 50%-250%
of the potential usage. And that's what I was translating into pounds
sterling. It is just the same as saying that if you pay one pound for a
battery, and you could get 50% more usage out of it if you kept the unused
battery in your fridge, you are losing 50p's worth of potential value for
your money if you don't keep the unused battery in your fridge.

> I think there are ways to approach the potential economic and social
> benefits of open access... through such mechanisms as estimating the
> value of increasing the rate of accumulation of knowledge... but it is a
> far from simple question.

Citations are a good second approximation, better than publication counts;
but we can do much better with measuring and predicting both the epistemic
and the monetary value of research. The 100% open-access corpus -- toward
which my approximations were meant to be another incentive -- will be the
database from which many rich future measures will be designed and mined.

Stevan Harnad