From: Anders Sandberg (asa@nada.kth.se)
Date: Sun Jul 28 2002 - 09:59:12 MDT
On Sat, Jul 27, 2002 at 07:42:01PM +0200, Joao Magalhaes wrote:
> At 08:02 25-07-2002 +0200, you wrote:
> >I just got another idea about dealing with this, based on the Way of
> >Google. If you count the number of citations of a paper you get a
> >rough measure of its importance, and presumably the breakthroughs are
> >sets of linked widely cited papers. The problem with this is of course
> >that some breakthroughs take time to be recognized, so ideally they
> >should be evaluated frim time=infinity. So an approximation would be to
> >look for the number of widely cited papers after a set time, like two
> >or five years. If you count the number of papers with citation levels a
> >few quartiles above median, then you might get a kind of breakthrough
> >measure. Still hardly perfect, but calculable.
>
> But how can you use it to predict breakthroughs? I mean, you will need to
> wait years to have an idea of the impact of a paper.
Sure. But we can do with a shorter time if we accept the number as being
less accurate.
Note that this way of measuring "progress" is equivalent to looking at
the tail of the distribution of citations of individual papers. If it is
becoming a fatter tail, then "progress" is occuring.
> Perhaps a better
> measure would be to calculate the diversity of papers. For example, we now
> have fields -- e.g. bioinformatics -- that we didn't have a 10 or 20 years
> ago. Of course the question is how to calculate scientific diversity?
> Perhaps we can count the number of different course at Universities? Now
> I'm sure that has increased in recent years. A higher degree of
> specialization might lead to new discoveries in broader fields.
There are probably just as many economic and political influences on the
number of university courses as there are on papers.
One way of finding out the level of interdisciplinary work would be to
look at the scientific collaboration graph. In a compartmentalized field
researchers form small cliques, while in an interdisciplinary field they
would link all over the place. Similarly for literature citations. This
is actually something that could be measured fairly accurately using
small world graph theory methods.
Still, interdisciplinary work is not necessarily progress. It is just a
health sign.
> Anyway, I'm
> starting to think that having a faith in future technological progress is
> no different than having a faith in God. Now I feel depressed.
:-) Don't worry, the fundamental problem with this thread is that we are
discussing progress in therms of *quantitative* measures, not
qualitative. But real steps forward - Galileo's insights into the
scientific method, Maxwell's unification of electromagnetism, the
invention of the computer etc are all about changes in how things are
done rather than improvements on previous tech. Each technology curve is
a sigmoid, and the sum of many sigmoids might be a great exponential,
but real progress is about finding ways of moving in new directions
instead.
A greek thinker worried about firewood scarcity might have dreamed of a
better way of gathering firewood, a new fuel or even a more efficient
stove - all of which we have today - but our solution to the problem of
cooking is a power grid, something which is totally unlike just a better
cooking system but solves a vast number of other problems too.
> >GCS/M/S/O d++ -p+ c++++ !l u+ e++ m++ s+/+ n--- h+/* f+ g+ w++ t+ r+ !y
>
> What on earth is this?
A geek code, an ancient version (1.0.1).
http://www.mit.edu:8001/afs/sipb.mit.edu/user/rei/Docs/HTML/GeekCode.html
-- ----------------------------------------------------------------------- Anders Sandberg Towards Ascension! asa@nada.kth.se http://www.nada.kth.se/~asa/ GCS/M/S/O d++ -p+ c++++ !l u+ e++ m++ s+/+ n--- h+/* f+ g+ w++ t+ r+ !y
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:15:43 MST