<p>The XMPP Standards Foundation is a continuing experiment. When we initially set up our policies, processes, and structures, we knew that our initial thoughts might not be our final thoughts, and that we would need to make adjustments as experience dictated. In this document, I argue that just such an adjustment is now necessary with regard to the Special Interest Groups (SIGs). <note>The proposal contained in this document formalizes some conclusions reached during a weekly discussion forum held by the XMPP Standards Foundation on 2002-01-23. A log of that discussion was formerly available at http://www.jabber.org/chatbot/logs/conference.jabber.org/foundation/2002-01-23.html. Further discussion within the Standards SIG has been helpful in clarifying the argument presented here.</note></p>
<p>&xep0002; defined a SIG as "a working group approved by the XMPP Council to address specific areas of growth or concern within the Jabber community", and specified that the function of a SIG is to "produce acceptable enhancements to XMPP within a reasonably limited period of time". In early January of 2002, XEP-0002 was modified to incorporate language about disbanding a SIG after a defined period of inactivity on the SIG's mailing list (normally six months).</p>
<p>Unfortunately, it is widely recognized in our community that the SIGs are not working. Ten SIGs have been approved by the XMPP Council, eight of them over six months ago in July and August of 2001. Two of the special-purpose SIGs (OOB and Presence) have seen no activity whatsoever and thus are clearly eligible to be disbanded. The other special-purpose SIGs (Conference, Formatting, Forms, Profiles, RPC, and Whiteboard) have seen extremely limited activity and it is a judgment call whether some of them should be allowed to continue according to the current standards defined in XEP-0002. Only the two "standing" SIGs (Security and Standards) have experienced significant and continued mailing list activity, mainly because the Standards SIG has assumed the role of discussion forum for specifications before they are submitted to the XMPP Council.</p>
<p>In perhaps the best measure of success or failure, only one SIG has produced a specification for submission to the XMPP Council, and that specification (&xep0009;) was essentially created outside the SIG structure at JabberCon 2001. Perhaps most ominously, no other SIG has shown any signs of progress toward completing a specification, or even starting work on one. With the possible exception of XEP-0009, all of the specifications created so far have come from individuals or small, ad-hoc groups -- not through the efforts of the SIGs.</p>
<p>In other words, an honest assessment forces us to conclude that the SIGs are not working.</p>
<li>"Crack the whip" -- encourage and cajole the existing SIGs into becoming more active, and energetically manage them so that they produce specifications.</li>
<li>"Wait and see" -- immediately disband the SIGs that are clearly inactive but keep the existing SIGs and hope that they will eventually produce something of value (over time disbanding any that are conspicuously inactive).</li>
<li>"Bite the bullet" -- recognize that, for whatever reason, the existing structure (many special-purpose interest groups) is not working and seek a better way to produce enhancements to XMPP.</li>
<p>Given the lack of activity in the SIGs so far (and the lack of time available to those who would manage them), I am skeptical that "cracking the whip" will produce results, and I believe the onus of proof is on those who would argue that the existing SIGs can be successful. Similarly, taking a "wait and see" attitude will simply let a bad situation continue unchecked, and in my opinion will at some point require us to choose between option 1 and option 3. Rather than postpone the day of reckoning, I argue that we need to address the problem head-on and take action to streamline the SIGs and find a better way of working.</p>
<p>But what is that "better way"? In order to figure that out, we need to understand why things are not working now. I don't think it's that the current SIG members are lazy, stupid, or incompetent -- after all, these are the same people who have in many instances created good XMPP-based software. Nor do I think it's that members of the XMPP community are incapable of creating specifications, because individually and in small, ad-hoc groups they have created quite a few.</p>
<p>I see several reasons why the SIGs are not working:</p>
<li>The XMPP community right now is too small to be split up successfully into smaller interest groups.</li>
<li>We have tried to overlay too much structure too quickly. The Jabber/XMPP community has traditionally been a fairly anarchic project (or set of projects), and creating ten SIGs right away was at odds with that successful lack of structure.</li>
<li>Good specifications, like good software programs, are usually created by at most a few interested people, not a formal group. Formal groups are not needed to move Jabber/XMPP technologies forward.</li>
<p>If we reflect on what is working, we see that specifications are being produced by individuals and small, ad-hoc groups. We also see that active discussion of those proposals is taking place in the Standards SIG, which contains everyone who is strongly interested in XMPP. Finally, we notice that the special-purpose SIGs have not played any appreciable role in our success so far.</p>
<p>My proposed solution takes into account everything we have learned to date about producing specifications and advancing the state of XMPP. Specifically, I propose that we take the following steps:</p>
<li>Immediately disband all but the Standards SIG. <note>In an earlier version of this document, I proposed that we retain the Security SIG. However, since there is a security aspect to all protocols, I now think it is best if security-related topics are discussed within the Standards SIG, not in a separate Security SIG.</note></li>
<li>Continue to use the Standards SIG as the preferred forum for discussion of experimental specifications before they are submitted to the XMPP Council.</li>
<li>If the Standards SIG cannot reach a working consensus on a given topic, let the document author(s) continue to rework their proposal informally outside the context of the Standards SIG. <note>One option would be to send interested parties off to their own ad-hoc mailing list (e.g., on JabberStudio, <linkurl="http://www.jabberstudio.org/">http://www.jabberstudio.org/</link>). Unlike the current SIGs, such a list would be established on the initiative of the document author(s) and would not require any formal approval by the XMPP Council.</note></li>
<p>There may be value in bringing back specialized SIGs in the future when the Jabber/XMPP community becomes larger. However, at this time I urge that we face the facts and proactively implement the solution I have outlined in this document. <note>Lest there be any concern that disbanding the SIGs is outside the power or purview of the XMPP Council, I note that Section 8.2 of the Bylaws of the XMPP Standards Foundation states in part that "The XMPP Council or the Members of the Corporation may, by resolution, ... terminate a Special Interest Group at any time for any reason." (An electronic copy of the Bylaws may be found at <linkurl="http://xmpp.org/xsf/docs/bylaws.shtml">http://www.jabber.org/bylaws.html</link>.)</note></p>
