Interview with Jeff Levinsky, Project Manager of ARAS Online
by Torben Gronning
The making of ARAS Online has been a bit like building a bridge over uncharted waters. In addition to the financing you need an architecture, a design, a crew, a chief engineer, a detailed plan of action, project management by the day or hour, and the skills to modify the project plan when you run into unforeseen obstacles.
Making the ARAS image archive available in electronic form owes its existence to a great many people,
but Jeff Levinsky — architect, chief engineer, designer, and project manager — was
key in making it happen. Jeff is a multi-talented computer scientist, who has been a true boon to the project. He graduated
summa cum laude from UC Berkeley, taught at Stanford University, and has worked on many computer applications in education, genealogy, and other fields. In addition to consulting, Jeff is also very active with the Friends of the Palo Alto Library and donates time to Native American tribes, where he teaches the use of computers.
Torben: Jeff, why does the world need ARAS Online?
Jeff: First, to make the collection available to a much wider audience; ARAS has many interested users from all over the world, most of whom cannot easily come to New York, San Francisco, or Los Angeles to work with the images. Second, to preserve the collection's paper images from discoloration and from tear and wear. Other benefits popped up during the project. We discovered new ways of presenting the collection for the users, and we made paths to many lesser-known records.
Torben: How did the digitization project start?
Jeff: Tom Singer, Chair of San Francisco ARAS, has in a previous
ARAS Newsletter article described the long journey to get this project underway. In late
2001 we were finally ready to begin the digitization project, which culminated
in the launch in July 2005 of the Internet-based ARAS Online.
After some study and initial meetings, in May 2002 I created a project plan and budget for the ARAS Board of Directors. First, we would scan
all contents of ARAS file folders — images and commentaries
— and store
digitized contents on a computer hard drive. Next, we would scan the catalog
cards and key in the commentaries. The last part of the project would be to
make it all available over the Internet.
At that time we only had
funds for the first part of the project, which we then started on.
Torben: What were some of the challenges in Phase 1?
Jeff: First of all, we had to undertake the scanning without disrupting daily operations of the collection in New York.
It's much like doing highway repair without disrupting the traffic. The
scanning of the commentary texts was done in the office, but we opted to scan
the images using outside assistance. We decided that we would not allow the collection's originals to be transported long distances over water, because of the risk of loss. That ruled out sending images overseas, and transportation had to be by truck.
We used two computers, a scanner and a screen color calibrator to verify colors on the high quality monitors. We worked out quality procedures and timed the operational steps per image and text record. During ten very hectic days at the start of 2003, we pulled the first 100 records
from the files, transported them to the chosen scanning company and ran the first trials. Every piece of paper was treated
with great care, because we didn't want to lose a single page, including hand-written notes. We had 16,748 images and 19,677 commentary sheets and related articles to scan.
By May 2003 one third of the scanning was completed. At the same time we learned
that a donor had given additional funds to complete the
A firm in India, which we selected from five bidders, helped us with the typing of commentary sheets. To ensure accuracy, the firm typed in every commentary sheet twice, coping with hand-written and faded lettering, and then we triple-checked their results. The same firm later typed the catalog cards the same way.
Within another year, all 23,000 catalog cards had been entered and all text commentaries typed. We created a prototype website for the archive in 2004, which made it possible to search for relevant images based on keywords. After yet another year of refinement and testing, we declared ARAS Online open in July 2005.
Torben: What were some of the surprises in Phase 2?
Jeff: We found that only two-thirds of the physical ARAS records were indexed by keywords.
We solved this problem in ARAS Online by letting you search for any word in all records at once, making it more likely you'll find what you are looking for.
Another surprise was that the parallel indexing in New York and San Francisco collections had created many duplicate and inconsistent spellings of the keywords on the catalog cards. For example, one card might say "3 Gods" and another "Three Gods" or "Gods, Three." To standardize the keywords, we wrote special computer programs and did a lot of manual cleanup.
We also had to transform the many styles of ARAS commentary sheets from decades
of researchers into one universal format.
Torben: What has the project meant to you?
Jeff: It has been an extraordinary experience to be part of a team so determined to preserve the incredible treasure of ARAS, and open it up for a much wider and appreciative audience. The team includes Allison Langerak, whose remarkable patience and attention to detail was essential to our success, plus very kind guidance from
ARAS board members Carol Sellers Herbert and Tom Singer.
My work with genealogy software and databases had already given me great respect for original historical documents in large volumes, but the archetypal nature of ARAS really opened a new world for me. I remember my excitement when returning
by bus to New York with the first 100 records scanned by our subcontractor. I knew my answer if anyone should ask what I had in the box: "Why,
that's part of the collective unconscious!"