Leaping Robot has been leaping a little less often. It’s summer and I have a big new research project – more on this some other time – that I’m getting off the ground. Also, I’ve been busy getting ready for an upcoming trip to Manchester to attend the 24th International Congress of History of Science, Technology, and Medicine. With meetings held every four years, it’s not quite the World Cup (or Comic-con) but it’s still the largest gathering for historians of science. One of the things I’ll be doing in Manchester – besides visiting Jodrell Bank and having a few pints of good beer with friends – is giving a paper called “Learning to Share.” Subtitled “Astronomers, Data, and Networks,” the paper gives an overview of a project I’m just wrapping up; the final results will be published next year in an issue of Technology and Culture.
My work on this topic started in 2011-12 when I spent a year at Caltech as a visiting professor. The main research project I did while there (besides finishing The Visioneers) was to explore the digitization of astronomy. Starting in the 1960s, astronomers’ view of the sky shifted from an analog perspective in which data was recorded using photographic plates and strip charts to one wholly mediated by digital technologies.Basically, I’m interested in how astronomy went from this…
Astronomer Roger Lynds doing some “old school observing”
Inside of telescope control room, Kitt Peak, a decade later.
By the early 1980s, astronomers expressed growing concern about having to deal with a deluge of data that was increasingly “born digital.” Data management became one of the modern astronomers’ necessary tasks as astronomy itself transformed into a particular form of “information science.” This transition presaged today’s debates about Big Data and the archiving of massive data sets in astronomy and other sciences which researchers mine.This process still has implications for today in the ways in which scientists share their data. For example…
In February 2013, the Obama administration announced a new policy designed to increase public access to scientific research funded by the federal government. In a memo that accompanied the White House announcement, John P. Holdren, director of the White House Office of Science and Technology Policy, directed U.S. science agencies to develop “clear and coordinated policies” so results from research they support will be publicly available within a year of publication. The new policy was motivated in part by the belief that shared science creates tangible practical and economic benefits. Another driver was the often stated complaint by scientists from astronomy to zoology who claimed they were simply drowning in data.
An overabundance of data, in fact, has long presented scientific communities with tremendous challenges. Today, astronomy is the scientific discipline that often appears in journalists’ accounts of “Big Data,” “data deluges,” and “information explosions.” However, it was during the 1970s, astronomers began commenting on an especially significant discontinuity in the amount of data they found at their disposal. Within a decade, astronomers routinely spoke with both trepidation and excitement about onrushing “floods” of data such that one might dare refer to research before this as “antediluvian astronomy.”
The key catalyst for this sense of crisis was the relatively sudden proliferation of new electronic and digital means for recording astronomical observations. This was not simply a matter of astronomers adding electronic computers to their toolkit. The operation of a competitive modern observatory also required a new workforce whose members possessed a different set of skills. Instead of (or in addition to) traditional expertise in astronomy and astrophysics, the digitization of astronomy required knowledge about solid-state detectors, digital circuits, and computer programming. Moreover, the digitization of astronomy helped reshape traditional norms and behaviors – what we can call a “moral economy” – in the astronomy community including those associated with sharing data and research tools.
Imagine it is 1976 and you are an observational astronomer. Regardless of what kind of telescope you use – optical or radio, public or private, orbiting in space or sitting on a mountaintop – if you wanted to share data you collected, could you? In the older analog tradition, astronomers loaned photographic plates to colleagues while observatories maintained physical libraries of the same. But, as more data was born-digital, the ease of sharing it posed an increasingly problematic issue.
In order for scientists to readily share their growing collections of digital data within the same observatory or – even more difficult – between institutions located in different countries it needed to exist in a common format. And, once digital data was in a common format, questions arose about how and when it, along with the digital tools to process and analyze it, might be shared.
At the Manchester meeting, I’ll be focusing on two episodes in this historical process using astronomy as an example. Both center around scientists’ wish to share data and data processing tools with their colleagues. The first episode concerns the emergence of the Flexible Image Transport System or FITS.
1981 paper describing FITS
This is a common data format developed by scientists at national observatories in the U.S. and the Netherlands and accepted as an international standard in 1982. Recently, the Vatican Library adopted FITS as its standard for storing tens of thousands of paper documents and manuscripts, some dating back more than 1,800 years, after they had been converted into digital images.
As critical as FITS was to calming scientists’ concerns about being overwhelmed by a rising flood of digital data, it only resolved part of the problem. The issue of how to interact with digital data still remained a pressing issue for astronomers. Even as FITS was accepted as a common data format, astronomers at institutions around the world still faced a bewildering assortment of image processing programs. When an scientist finished an observing run, they would often write their own “homegrown” software code to process their data. Consequently, here was little in the way of consistency when it came to image processing programs as astronomers came up with solutions that were local, disorganized, and ad-hoc.
This leads to the second example I’ll be discussing in Manchester. STARLINK was a sophisticated computer network for sharing and manipulating digital astronomical images that debuted in the United Kingdom in 1980. STARLINK was centered around a central node – this was at the Rutherford Laboratory near Oxford – to which several other sites were be linked.
Schematic of the STARLINK system.
Leased telephone lines (this is all pre-World Wide Web!) from Britain’s Post Office connected the STARLINK sites. Their initial data capacity was sufficient to allow the transmission of image processing programs but too slow to transmit large amount of data. At each of STARLINK’s sites, astronomers could access two image-display systems as well as peripherals like printers which allowed them to interact with their data in real time. But implementing STARLINK as a tool for sharing software proved more difficult than getting astronomers to share the data. As a result, STARLINK, while ambitious, never achieved all of its creators’ idealistic goals.
The Starlink VAX11/780 in the Atlas Centre, at the Rutherford Laboratory, August 1980
Astronomers’ development of new tools like FITS and STARLINK produced a new space in which different ideas, practices, and behaviors about data and its ownership could emerge. The digitization of astronomy and the question of “sharing science” was not limited to just one country or subfield of the discipline. Rather it was a process that all researchers – observers and theoreticians alike – experienced in some way. For astronomers, the question of how to share their science, was, in both senses of the phrase, a universal concern.