.ig
  T/6 -- Woods Hole version
Lederberg/REVISIONS  9/25/91
..

.ce 5
Communication as the root of scientific progress
Lecture presented by Joshua Lederberg
Sixth International Conference
International Federation of Science Editors
Woods Hole, Mass. October 16, 1991.

.ls 2
.fi
     Thank you John [Kendrew].  Your tacit question about what I am doing
here I share, but I've done lots of mischief together with Ken
Warren, some of it here --  it's usually come out to a good end.
Yes, I am very interested in scientific information.  I
don't do very much editorial work these days; I'm back working
in the laboratory after a lapse of twelve years and that has
kept me very busy trying to reacquaint myself with the literature
of my own field.  So I will offer you the perspective
of a scientific reader.  Now some people tell me that's a vanishing
species! For anyone to say that, even with some sense of irony, is
an atrocity.  One of my main functions with my own laboratory group
is that I try to be its principal reader.  If
something goes on in the world outside and none of us has heard about
it for two or three weeks, I'm the one who feels responsible.  I
want to be alert to events that might have a very important bearing
on the way we think about our own research, our planning, of the
data coming in, of the sources of error.

    Let me begin with a few
truisms, just to be sure that we are operating on a common ground
of reverence for the publication process.  Publication is, to start
with, just that!  \fIPublic\fR-ation.  It converts private to public
knowledge, in the service of registering a private claim of original
authorship -- in science, of discovery.  Above all, the act of
publication is an inscription under oath, a testimony.  It is accepted
as valid until firm evidence to the contrary; and there is an
extremely high standard of accountability for what is published under
a given person's name.  Just look at the daily headlines.  It
is the essential ingredient to make scientific work responsible in the sense that one cannot
readily retreat from assertions that have been signed, delivered to the
printer and made available to thousands.

    These publically asserted
claims also play an extremely important role in the allocation of
resources, the ability of different scientists to survive in the
competition with other
legitimate claims for expenditures, for support of laboratories,
for positions at the institutions, for space in the journals, for
the attraction of students and collaborators.  All these rest on
those claims, the evidence for which in the end is in the public
record.  Both author and audience benefit from the
successful assertion of those claims: especially credibility, that
one doesn't have to spend an inordinate amount of time reexamining
every detail of an individual's output if that person has 
established credibility through prior publication and exposure.

     Publication also results in a repository, constructing the
tradition of science.  Up to this point it can hardly be anonymous
in order to perform the functions that I have just indicated.  But
as time goes by, we have the reassimilation of the content of
scientific work and as it settles in and survives the criticism that
it should have had at its early stages of the process, it becomes the
common tradition, the unquestioned shared wisdom --
often becoming anonymous by obliteration.

     The literature is also a forum.  It's a gladiatorial arena for
competing claims, resolving discrepancies in data or interpretation.
There used to be oral duels, and we revel in stories like
Pasteur's confrontation with Pouchet that finally put spontaneous
generation to rest in 1864.
Today, our battles are more
often fought out in print, which is indeed appropriate because the
testimony then becomes available to the universe, not simply to the
immediate onlookers.  Despite the opportunity for very broad
dissemination, there is the paradox, nevertheless, that broadcast restricts
individuals'  access to feedback.  The publication system,
at least in principle, should allow a dialectic to appear in more 
symmetrical terms where anyone with something purposeful to say has a way
to get into the system.

     If the literature is a forum, it is also a rumen, a place for the digestion
and assimilation of the variety of inputs where scientific claims
go through a period of seasoning, modification, modulation.  Even the
truths look different five or ten years later regardless of explicit
criticisms.  We can expect a process of reinterpretation, a
post-historical reexamination of the meaning of their terms.
And now I only need to remind you of the term "imprimatur" (a
wonderful metaphor): the imprinted witness that, an article having
appeared in a refereed journal, it had survived a critical
process, a conspiracy if you like, of the editors and the publishers
and the referees -- that something has appeared which is worthy of the
shared interest and precious attention of the community.

     May I tell you what I do as a reading scientist today?  Reading
the scientific literature has been my primary vocation for fifty years.
Books play a diminishing role.  Today they are mostly for targeted 
reference.  In the scientific domain, we rarely have the leisure today to read a book
from cover to cover.  A few
biographies command attention.  I just finished a
proof copy of Carl Djerassi's life story: "The Pill, Pygmy Chimps and
Degas' Horse";  another 
of that genre was Francois Jacob's revelation of the development
of his scientific work: "The Statue Within.
These are obviously not very contributory
to the details on how to do my next experiment, but they tell me a
lot about the scientific personality, providing object lessons and
models for emulation.  Rarely, I do see a work that compels total
ingestion -- for example
"Physiology of the Bacterial Cell" by Neidhardt, Ingraham and
Schaechter.  This is such
a magnificent synthesis at a fairly elementary level of exposition
that I really marveled at the deliberation and distillation that
went into the telling.  Wonderful books like that are
rare.  In printed form they surely will be the survivors of any electronic
revolution.  At an intermediary level of indispensibility as books in
print format are the Annual Reviews.  They are reference works
for whatever you have to look up; but they also give
a chance to browse through an enormous literature with some
coherence.  Compare an Annual Reviews of Genetics with current issues
of the journal Genetics.  Even if I had the time to read every
article, I wouldn't have the background to be able to place each one
of them in the appropriate context of what comes through.  And I
regard this as my home discipline!  People will spend varying amounts
of their time and energy as well in trying to understand what is going on in
science beyond the window of their own specific work in their research
and teaching.

     There are about a dozen journals that I subscribe to and maybe
seven or eight of them that I do scan from cover to cover: Nature,
Science, Proceedings of the National Academy  -- those are the very
general ones.  The Journal of Bacteriology, Microbiological Reviews,
Genetics, Biochemistry.  I pick up a hot paper now and then from The
Scientist, and look at The Sciences, New Scientist, American Scientist,
and Scientific American for general scientific culture.  That is a
textual sampling, not immersion.  You couldn't read every article
in a critical and detailed fashion in just the ones I have listed
within the number of hours that there are in the week.  What you can
do within a couple of hours a day is to scan that range of material
and try to pick out those things that might be of interest.  To follow
the structure of argument just
in one's own specialty, you must go to the detail of trying to check
the numbers on the graphs and see if they match the authors assertions
-- an arduous task.

    We are well served by those kinds of journals
in terms of maintaining a general currency about what is going
on in the field, and they match very well the energy and
interest and intellectual acuity that our scientific readers are able
to put into the process.  I see  no occasion for those to be
altered.  Most
scientists are very grateful for them:
what thousands of scientists will share as common
currency, to carry in their briefcase and read on the airplanes
and the commuter rides, with all the convenience of the present print
format.

     My main problem is how do you reacquire, retrace that
intellectual traffic.  What
do you do with all of your marginal notes and, how do you synthesize
a coherent system of what you've read?  Well to try to deal with this
on a current basis, I have Gene Garfield's wonderful products.  I get
the weekly Current Contents on diskette with all of its
embellishments.  I eagerly await the five or six diskettes that have
to be loaded, every week, and sometimes impatient about how long it
takes to load them and get going with that week's literature.  My
stored profiles work out reasonably well, but have to be embellished
from time to time.  You discover new keys, other notations that
authors insist on in changing fads and idiosyncrasies of language.
I can
warrant that my profiles recover on a current basis about 90% of what
I have read or would want to read.  God help me if I lose my notes on
the rest!

     Then, how keep up with what's closest to my immediate specialty?
Acquiring a couple or three papers a day is not hard.  And even with
a fairly detailed critical examination down to checking the points
on the graphs and so on, reading them as they come in is entirely doable.  My problem
is the arithmetic of accumulation.  After a decade, I've got about
10,000 papers that I have got to keep track of: the texts and my
marginal notes and so on.  And here my system is absolutely broken
down!  A technological fix is on the way: document scanners that can
store page images and digitize scripts on searchable media.  One or
a few CD-ROM's will take care of the storage.  But what a lot of
bother for information, yes full text, that I should be able to
acquire electronically in the first place.  The more so for specialty
journals and references to be searched on demand.

     Within a given specialty there are usually one or two journals
that specialists must see.  There may be only a couple of hundred
people who have such a level of interest that they will look at every
article.  There are the journals of broad appeal, and
then a very flat distribution of the other sources.  For my part,
an additional 30 articles a month -- perhaps half of them come
from about fifteen journals; you can probably extrapolate with
Bradford's Law to the rest.
90% will come from about thirty-five, and then there is a gradual
asymptote out to the vanishing returns from the total coverage
that the system is going to offer.
Every now and then an article does pop up from an
obscure place whence you had no systematic way to recover it; but
in retrospect it was really quite important.  

     So each of us faces the task of selective retrieval from a
cosmic domain of stuff that every other eager beaver in the world
has been busily putting into the repository.  Our present technology
enables an approximation with reasonable confidence.  Keeping track
of what you have accumulated on pieces of paper is the frustration.
That's not your bedside reading, well served by the print on paper
version.  The next step, to
integrate that into your own private library of useful knowledge,
is simply not achievable with last year's technology.

     The fact is that scientific literature inherently has grown
beyond the scope of any hundred people to have understood it and gone
into it in some depth; it is built in to the growth of knowledge that
past improvements in communication and storage aren't going to alter.
What are the consequences?  For one thing, the problematics
of assessing the literature reinforces all the other drives to specialization.  The
ambitions of scientists have changed, to focus on ever narrower
targets.  It's just too much hard work to master an interdisciplinary
area on top of all the other institutional obstacles.  Never mind
the intellectual conceptual problems.  Never mind the problem of
getting funding or moral and fiscal support, just to get hold of
the necessary expertise and information!  But that impediment is in
principle remediable.  At the same time, are we drowning in
information, inundated by the numbers of journals?  You know, when
you come to any specific issue, when there is some important special
fact that you would like to know all about, the shoe is very often
on the other foot.  My usual experience in asking a new question:
the odds are that the exquisite detail needed to take the next step
has just never been done.  So here, far from being drowned, I have
a great deficit of specific and detailed knowledge of exactly what
happens in such and such a system with such and such reagents, and so
forth.  Our systems for acquisition of that kind of material is not
perfect, but it is getting a lot better; and with devices like key
word searching, like related articles, like full abstract searching,
which is just about what the technology does offer today, I can feel
reasonably confident that explicit matters of factual detail, whether
somebody has done that particular experiment can be retrieved, but
often only with a lot of effort.  Much more difficult, has anybody
else had a good idea that would be pertinent to my search?  Those
keys are so much more difficult to catalogue.  Often it takes great
creative act to recognize that a concept developed in one context
really is pertinent to another.  So there will never be a guarantee
that those can all be acquired.  But there is at least the hope
of finding it in that literature, and it is very important hope to
try to preserve.

     There are different adaptations to that flood, and more and more
we do see, what I can only describe as a scandal, that scientific
literature is not always taken seriously any more.  In polls of
scientists many will say that the primary
source of their information about scientific work in their field is
not with the published literature.  It's by word of mouth, it's by
telephone networks, by attendance at meetings, and so on.  People
have got to do what they've got to do.  But I find those kinds of
sources so unreliable!  I feel very uncomfortable when the only
place that I have heard something is by word of mouth.  If I can't pin
it down, if I can't hold its source accountable by saying that was in
a published item, I can't look at it in detail, ruminate about it,
think through what second order reactions I would have, I don't know
whether my colleagues share that.  They may feel that they don't have
any alternative except to pick up what's on the rumor mill, but I
think great mistakes can be promulgated in that fashion.

     The telephone is a wonderful instrument,
but when I try to use that to get information, people who have what I
am looking for are all pretty busy.  I hate to impose on their time,
and if I do, there is usually a round of telephone tag of three or
four attempts to catch somebody before you actually do and try to get
hold of them for the information.  If it's a reference, I am
delighted.  If it's an attribution, it cannot be pinned down more
definitely than you know, this is what I think.  I don't feel like I
have made a great advance over what I have had before.  Not taking
literature seriously reinforces the trend that libraries in
desperation are cancelling subscriptions on journals that they don't
see being very much read locally.  And it doesn't make any sense to
have a local copy of a serial where perhaps one in a hundred titles
will ever be examined by anybody in that institution.  For some of
these journals, de facto, are approaching the point where they might
as well only print one copy, send it to the National Library of
Medicine or some other repository, and let it redistribute reprints
by interlibrary loan.  The economics obviously is insupportable.
The fundamental problem is trying to foist an inappropriate level of
number of vehicles on an outmoded mechanism for the purpose of
dissemination.  So that would fall of its own weight.  You can see
what I'm leading to: go from 1000 to 1 to zero print copies.

     Meantime, the libraries are in a great dilemma trying to figure
out exactly what to do.  They get a fight from the faculty -- what a
librarian hears when they want to drop a journal, you would think
that every professor was reading every issue of every journal in the
library!  For the operational procedures by which libraries can make
sensible decisions about acquisition priorities, they could get any
number of technical aids on that point*, but it does put them in a
very tough spot.  Besides the budgetary crunch, the libraries are
also running out of space.  The older stuff is deteriorating anyhow!
Maybe ink on paper was not a totally bad idea, for that reason alone,
provided one
clean copy remains available; unfortunately things don't always work out
that way.

     One direction things could take if we don't reform the system is
that invisible colleges will take over as the principal but 
unreliable routes of communication.  Archival
copies of material will eventually be sent in to
the repository, but there will be a limbo of material that doesn't
know if it is going to go to hell or heaven for four or five years,
while it is still cooking and unaccountably available, on a basis
far from equitable.  So, in due course there has to be a
wholehearted exploitation of the new technologies and I don't have to
plead for it.  It's happening because electronic networks are becoming
more and more available de facto to people working in a variety of
fields.  A couple of dozen of them now operate with a routine exchange
of preprints (sic).  The central problem facing the journal has been
a radical change in the economics and technology of printing, without
an adequate recognition of the essential value-added in the journal
process.
From Gutenberg's time until mechanized and computerized
composition, that was providing the capital and the
entrepreneurship and the organization to facilitate a process whereby
an expensive and precious printed article was the product.
It was characterized by rather high capital investment
in the initial composition of any material.  Once it was composed,
there was a rather low variable cost for further dissemination.  We
had a market system for determining what was worthy of that degree of
capital investment.  Well today the capital investment on the printing
technology is almost zero.  The important value-added is the editorial
process including issues of selection, then of editorial work and
improvement.  And that very precious imprimatur.  When something
comes out in a journal of high repute (to make a circular argument),
that's a journal worth my time and worth my attention.  If it is just
thrown up in the air without having undergone that kind of editorial
review, it will not have been refined in terms of both the
presentation, and perhaps even substance of the argument, and it won't
have the imprimatur of other people, whose judgment I trust, that it's
worth reading and can be relied upon for accountability.  Whether the
article then gets into \fIprint\fR is almost an irrelevancy at this point.
Any of a variety of media of communication could follow on that 
editorial process.  What we need to see more than has happened so far
is the marriage -- of that editorial role on the one hand, with a
production role that uses the electronic technologies rather than
the print, on the other.  And that's where the spontaneous bulletin
boards don't quite make it.  They quickly get filled up with
obscenities, literal and otherwise for lack of that sort of control.  I
don't mind the obscenities as long as I don't have to plow through
them, but I'd like a
truth-in-advertising framework that tells me, as I say, what's worth
reading.  I'd like to know that x,y or z editorial committee had been
established as a guide for what is worth capturing the
priority of my attention.

    I think it will be the societies that provide the most
likely framework for the organization of those functions.  It won't make any money
to start with.  But the economics and the technology
will converge with the social necessities for this kind of
improvement.  Technically, we don't need much more than what we now have.
There are a few problems about transmitting graphics and formatting
manuscripts.  Some standards have to be established, and some minor fixes
especially on the graphics.  But we are basically right there.
Machines with gigabyte storage and ever smaller 25-meg processors
are very routine today.  You will find them by the hundreds in the 
laboratories and the libraries and so on, with a doubling of
capabilities per unit costs every couple of years.  So in ten years
today's "super-computer" will be available certainly in every
institution, and to a large degree in every laboratory.  Communication
links won't grow quite as fast as that, but if you consider the
bandwidth of a package of CD ROMs, you have a variety of technologies
for all the communication we need.  So those are not limiting factors
either.  They are not very expensive.  The machinery, the social
framework, the decisions involved, the wetware, the distribution
channels, the marketing and so on, really are all that stand in the
way.  There are not the same kinds of profit incentives that drive
paper publishing; so I think the not-for-profit institutions will
start taking over.  Perhaps the for-profit publishing houses will
provide the essential technical services because they can have the
economy of scale, the organization, the hardware and so on, and then
contract that out to the societies for providing the other elements
of the equation.

     That partnership could be a very productive one for the entire
scientific community.  One feature of that kind of a system to which
we have only a crude approximation today, is feedback, dialectic.
It shouldn't take a federal case for reactions to a paper to be
elicited from the scientific community and not just on the rumor
network, but some place where everybody else can see it.  This is the
bulletin board system of commentary and would complement what the
fixed board of editors would have to say.  If there is a good
dialectical system and the critical community has an opportunity to
express its views, even ex-post facto: that's how the scientific
process works at its best.  Here the economics and the
technology for dialectic give a great edge on the electronic systems
over the printed ones, if for no other reason than how to get
propinquity?  I mean, if an article has been printed and then a little
later on, I write a critical reaction to it, (even in the rare case
that the journal accepts that sort of commentary and further dialogue,)
they do not adjoin one another
on the shelves.  It's a nuisance trying to find them.  Let's
say I write something six months ago; Gene Garfield wrote a blistering
critique sometime after that.  How are the two of them going to be
brought together?  That kind of reshuffling of the units is very hard
with printed paper.  It's trivial, of course, to do it with electronic
media via the networks of linkage of material and commentary.  That
potential for reaggregation stands just after mechanized search and
tempo of availability as
the greatest advantage that these new kinds of media can offer.  

     Let me make one further comment about global access: something
very dear to Ken's heart, and to my own.  There was a
remark in my letter of invitation: "You may feel like you are in a
flood, but people in the Third World are in a real drought.  They
never get the journals that you complain of getting too many of".
And so forth.  The economics of sharing will shift dramatically with
these media.  For trivial marginal costs you can provide 100 CDs a
year which would far exceed the total volume of publication that
they could ever hope to get in any other way.  There is no other way in the
world that we can reduplicate all the paper libraries that are
we now have as a privileged treasure.

     Another feature about globality that electronic systems will
offer are built-in translation aids.  I am not talking about the
Nirvana of automated perfect smooth translation.  Most of
us here have a smattering of two or three foreign languages; a few
of you are great linguists.  But when I am reading an article in
German, which I am fairly fluent in, wouldn't I love to have a
built-in dictionary to help out when I run into a phrase that I didn't
understand?  I'll take the risks of that crude translation.  It may
come out with some of the ridiculous puns that you all know about.
Again this becomes trivially easy in terms of its marginal cost, and
will greatly extend the global accessibility of literature to a wide
variety of people whose command of the current international standard
English may not be perfect.  So these are some of the arguments for
the reforms that I hope you share with me and I would like to see
brought about.  Thank you very much.
.ce
 ----------------------

*  For The Rockefeller University library, I used the Science Citation
Index to prepare an index to the
frequency with which different journals were cited in their published
papers.  That would be a bad algorithm for acquisition of books and review journals.