It was stunning. 10 days ago, a few days after the voting period began on XRI Syntax 2.0 and XRI Resolution 2.0 becoming an OASIS Standard, the W3C TAG (Technical Architecture Group) came out with a statement recommending that members of OASIS (a completely separate and independent standards body) vote against it.
Despite 20 votes already being cast in favor of the specifications at that point, almost immediately a handful of negative votes were cast with comments referencing the W3C recommendation, starting with Sun Microsystems (especially ironic since Eduardo Gutentag of Sun is chair of the OASIS Board of Directors).
Although it’s not unusual for a proposed OASIS standard to have some opposition, what is strange is to have that opposition led by another standards body. What took the XRI TC even more by surprise, however, was the W3C’s sudden vocal opposition to XRI 2.0. When the W3C TAG submitted a comment on the last day of the first public review period of the XRI Resolution 2.0 specification in March, the XRI TC responded with a detailed 5-page answer to the three questions posed by the TAG. We never received any response.
In a subsequent final public review of XRI Resolution 2.0 held the following month, we didn’t hear anything more from the W3C TAG. Nor did the TAG minutes show any further discussion.
In any case, the XRI TC posted a response to the TAG’s position, and OASIS members responded by casting more positive votes, enough so that only a few days later the vote passed the minimum threshold of positive votes (15% of OASIS voting members) required for an OASIS Standard to pass.
That’s when it became apparent just how badly the W3C didn’t want that to happen. The second rule with an OASIS Standard vote is that no more than 25% of the votes cast can be negative. That’s never happened in OASIS history â€“ in fact the highest percentage of negative votes ever cast against an OASIS standard was 9% (for the Management Using Web Services v1.0 specification in February 2005, for which a whopping 6 negative votes were cast).
But within hours of the XRI 2.0 ballots reaching the 15% positive threshold, suddenly more negative votes began appearing. Almost all of them referenced the W3C TAG recommendation. By Thursday of last week, with three days left, enough negative votes had been cast to reach the 25%-of-all-votes-cast negative threshold.
Naturally XRI supporters responded by contacting other OASIS members, informing them of this unprecedented situation, and asking for their support. In one case, a company’s OASIS voting rep had cast a negative vote apparently without knowing his company was planning to adopt XRI and XDI technology. After a hastily arranged meeting to explain the details, the company reversed its position and voted in favor.
Many more OASIS members responded likewise, and by Friday morning there were again enough positive votes to safely pass both specifications.
But it wasn’t over. More emails, phone calls, and even blog posts from W3C TAG members went out. More negative votes appeared. By Friday evening, the negative votes were again just above the 25% threshold.
Given that the final day of the vote was a Saturday (OASIS Standard votes always run the final two weeks of a calendar month), it took an extraordinary effort by XRI supporters to reach out once more to OASIS members for help. But once more they responded, and by noon on Saturday, 8 hours from the close of the vote, 72 positive votes had been cast, enough to pass both specifications.
But I had a sinking feeling as I left to work on a birthday project for my youngest son. Sure enough, when I came back for dinner that night, with only four hours left in the vote, two more negative votes appeared â€“ just enough to cross the 25% negative threshold and defeat both ballots.
As I watched the voting period end on Saturday night, one thought kept echoing through the back of my mind: “What is the W3C so afraid of? Why do they care so deeply that OASIS members not approve XRI 2.0 as an OASIS Standard? Why on earth would they turn out such an extensive and unprecedented lobbying campaign for something they have so long ignored?”
In other words, if they thought XRI was such a bad idea (or in their precise words, “We are not satisfied that XRIs provide functionality not readily available from http: URIs.“), why don’t they just let it die a natural death in the marketplace?
In any case, we’re about to find out. A number of OASIS members who voted No at the TAG’s urging noted their reluctance in doing so in their comments. They explicitly asked the OASIS XRI TC and the W3C TAG to sit down together and iron out our differences. The most eloquent was Ray Denenberg of the Library of Congress, who said:
First we reference comments of Wells Fargo, who said:
“Although we support XRI’s objectives, we urge the XRI TC to consider W3C’s comments seriously and add a non-normative clause explaining key differences between XRIs and URIs, and detail how the former address specific deficiencies of the latter.”
And comments from Nokia, who said:
“Although the XRI TC and W3C TAG have exchanged some e-mail regarding the XRI spec, it appears the engagement has been mostly superficial. Consequently, we recommend these two groups engage in detailed technical discussions (including use cases and deployment scenarios) before OASIS formally adopt this spec.”
We further observe that the body of existing documents on XRI, though abundant, all seem either too high level or too detailed. It is very difficult to get the whole picture.
The Library of Congress urges OASIS to:
1. Consider W3C’s comments seriously, explain differences between XRIs and URIs and how XRIs address deficiencies of URIs, and respond with substantive explanations (rather than existing promotional text) to W3Cs concerns.
2. Prepare a paper (or non-normative clause), perhaps 5-10 pages, that includes the above, as well as a comprehensive description of XRI, including use cases, deployment scenarios, and real-life examples.
The Library of Congress supports the XRI objectives and we are prepared to change our vote if these steps are taken.
This is highly constructive feedback by Ray, and in my personal opinion, it lays out precisely the path the XRI TC and the W3C TAG should take together. Although I obviously would have preferred other ways to get here, those who know me know that I prefer to focus energy on how to solve problems, not how to create divisions.
The past is past. With this blog post I’m personally holding out an olive branch to the W3C TAG (and encouraging other XRI TC members to do the same) and asking to begin the dialog that will hopefully result in a mutual understanding about the role a layer of abstract structured identifiers will play in the Web.