A Bit of History as GEDCOM Approaches Its 30th Year - Thu, 18 Sep 2014

I was looking for the early GEDCOM specifications prior to GEDCOM 5.3 from 1993, so I thought I’d contact some people I knew who I thought might have some of the early GEDCOM material laying around. One of those I contacted was Bill Harten, sometimes referred to as “the father of GEDCOM”, who I met and had some nice conversations with at RootsTech 2014 in February.

In our correspondence, Bill told me a story I had never heard that only he would know and it was regarding the genesis of GEDCOM. Bill gave me permission to publish this. Here is what he said:

“GEDCOM’s tag-hierarchy technology was re-purposed from a database technology named AIM that I invented in 1984 to provide both high flexibility and high performance for complex ‘big data."  AIM’s serialized representation was perfect for a data transfer format and GEDCOM needed to handle complex data in a flexible way. FamilySearch used the AIM database technology to distribute their products on compact disc readers and personal computers from 1985 until FamilySearch.org was deployed fifteen years later. You would recognize this technology today as an XML-based DBMS, although AIM includes advanced indexing, search and compression concepts that are still unavailable in commercial DBMS products now 30 years later.

GEDCOM 1.0 was published and delivered to about 30 attendees of the first GEDCOM Developers’ Conference in 1985 in the Family History Library in Salt Lake City. I sent out invitations to developers, media, and genealogical community representatives, who responded enthusiastically. GEDCOM 1.0 was a draft prepared for their review. We incorporated their feedback into GEDCOM 2 which became the first official release. GEDCOM 2 was first implemented by Personal Ancestral File and by Roots, a popular software produced by Howard Nurse. The original publisher of Genealogical Computing magazine took a strong stand in favor of GEDCOM and recommended that readers demand GEDCOM compatibility of any product they purchase. The combination of two initial implementations and favorable media coverage was by design. Their cooperation was secured in advance of the conference. Other vendors were supportive and eventually implemented 2.0 as well, and GEDCOM eventually rose to achieve critical mass.

At this conference I initially proposed that the developers join together under an ISO committee that we would form for the purpose of establishing GEDCOM as a formal international standard. I told them the cost in committee-time and money. They all expressed that the cost was too high given their small size. I anticipated this and had prepared an alternative that I described as a ‘de facto standard’ approach. In this approach the Family History department would commit to implement GEDCOM in PAF and Ancestral File, and share data with whoever would support the format. If others liked it they could follow suit. This approach was adopted unanimously. The list of GEDCOM-compatible products grew rapidly. In the first year a feedback-driven process was established for adding tags and evolving GEDCOM, and I managed this process through several releases, the last being GEDCOM 5. In 1991 I visited software developers in France, Germany, Great Britain and Russia to help them understand and adopt GEDCOM.

We learned that some developers were challenged by GEDCOM’s hierarchical data structure because it did not yield very well to traditional flat-record programming methods. Our AIM database was based on a powerful tree-based data structure and accompanying algorithms for representing, parsing, and traversing hierarchical data internally in memory, so we cloned this code and created the GEDCOM library, and made this available to the GEDCOM community in the public domain. Not all developers used it, but those who did seemed to have a better experience and fewer problems than those who used more traditional approaches. We see this same data structure today in the Document Object Model (DOM) trees that form the backbone of the modern web browser, and many of the names of the DOM structures and methods in browsers and DOM libraries today are the same as we used in the GEDCOM library a decade earlier.”

Let me also refer you to a few posts long ago by Bill Harten on the GEDCOM-L mailing list that explain more of the early thinking about GEDCOM. Very interesting reads:

Bill’s first post on the GEDCOM-L list:
LDS GEDCOM Report, 22 Oct 1994

Although database technology has since improved, the ideas at the time about the database model were both innovative and necessary:
Database: GEDCOM’s Genesis. 28 Oct 1994

Why wasn’t GEDCOM developed through a formal standards organization?
GEDCOM and Formal Standards Organizations. 24 Jan 1996

You have to conclude that there was a lot of thinking and effort put into the development of GEDCOM.

I think this material is a must read for the FHISO people, because to create a new standard today, they’ll have to rehash many of the same problems that Bill & Company had to address.

And if anyone still has any of the early GEDCOM specs prior to 5.3, Bill and I would love to see them.