7 The impact of the ADS on astronomy

It is difficult to judge the impact of scientific work. For scientific programs citation histories, personal honors and awards, and the success of students can give a measure of impact. For support type programs these measures do not suffice; the impact of the 200-inch Hale Telescope (Anderson 1948; Rule 1948) or the 4-meter Mayall Telescope (Crawford 1965) clearly extends beyond the papers and honors of their respective developers. The impact of large software projects is, if anything, even harder to quantify; the large data reduction environments, like AIPS (Fomalont 1981; Greisen 1998), MIDAS (Banse et al. 1983), or IRAF (Tody 1986) have transformed astronomy, but how much?

The ADS is perhaps unique among large support projects in that a reasonably accurate quantitative estimate of its impact can be made. This is because many of the services the ADS provides are just more efficient methods of doing things astronomers have long done, and found worth the time it took to do them.

We will assign to each of several ADS functions a time which is our estimate of the increase in research time which accrues to the researcher by virtue of using that function. Our fundamental measure will be the time saved in obtaining an article via the ADS, which we estimate from the time it takes to go to the library, find the volume, photocopy the article, and return to the office, as 15 minutes. We then estimate that reading an abstract, a reference list, or a citation history saves 1/3 of the full article time, or 5 minutes, and we arbitrarily assign a one minute time savings to each query.

We can now estimate the impact of ADS, in terms of FTE (Full Time Equivalent, 2000 hour) research years, by examining the ADS usage logs. We note that about half of the full text articles currently retrieved via the ADS come from the on-line journals, which certainly deserve credit for their work. Also we are ignoring several important (but hard to quantify) aspects of the ADS service, such as links from other web sites (e.g. the HTML journals), the synergy of joint ADS/SIMBAD and ADS/NED queries (e.g. that in Fig. 1), the bulk retrieval of abstracts and LATEX formatted references (about 200 000 per month), and the more than 10 000 000 references returned each month. We think that what follows is a reasonable estimate of the impact of the ADS on astronomy, and that the impact of the full Urania collaboration is substantially more.

Using the March 1999 worldwide combined ADS logs there were 113 471 full text articles retrieved, 195 026 abstracts (individually selected), 10 663 citation histories, and 3 702 reference pages retrieved, and 582 836 queries made. Using the estimated time savings above we find that the impact of the ADS on astronomy is 333 FTE research years per year, approximately the same as the entire Harvard-Smithsonian Center for Astrophysics.

If we crudely estimate that there are 10 000 FTE research years in astronomy each year the ADS can be viewed as accounting for 3.33% of astronomy. Currently the ADS contains 27 712 (11 834) articles (refereed articles) in the astronomy database dated 1998, so one way of expressing the impact of the ADS would be 923 (394) articles (refereed articles) per year.

While the efficiencies brought about by the technologies inherent in the ADS and Urania are permanent, and will contribute (compounded) to the accelerating pace of discovery in astronomy, one can ask what was gained by being first. Risks were taken in funding the early development and adoption of technologies via the ADS and Urania. Also, had nothing been done, the "winning'' technologies would eventually be adopted with very little risk.

To judge the payoff we adopt a simple model; we assume that the increase in research efficiency due to the ADS has increased linearly from zero in 1993 to 333 FTE research years in 1999, and that it will decrease linearly to zero over the next six years, after which there will be no difference in the technologies employed.

This yields a sum impact from the early creation of the ADS of 2 332 FTE research years, which is 23% of the astronomical research done in a single year, or 6463 (2760) papers (refereed papers). This is surely equal to the impact of the very largest and most successful projects. Doing this analysis for the entire Urania would yield a substantially increased amount.

Acknowledgements

  Peter Ossorio is a pioneer in the field of automated text retrieval, he gave freely of his ideas in the early phase of the project. Geoff Shaw provided the enthusiasm to keep the Abstract Service project going during the long period of no funding.

Margaret Geller gave crucial encouragement at the time of the original prototype. Frank Giovane long believed in the possibilities of the Abstract Service, and acted as a friend in high places.

Todd Karakashian wrote much of the software at the time of the public release, he left in 1994. Markus Demleitner joined the ADS project in April 1999, he has already produced much of value.

There are about a dozen individuals at the Strasbourg Observatory, and the Strasbourg Data Center to thank, too many to thank individually. The data services provided by them are at the heart of the new astronomy; their collaboration with the ADS has been both very fruitful, and a great joy.

Peter Boyce, Evan Owens, and the electronic Astrophysical Journal project staff have had the vision necessary to do things first. Their collaboration has been important to the success of the ADS, and crucial to the success of Urania.

Without the long term support from NASA, and Günter Riegler in particular, the ADS would not now exist.

We are supported by NASA under Grant NCC5-189.