From cbmvax!uunet!cuvmb.bitnet!LOJBAN Sun Aug 2 03:28:20 1992 Return-Path: Date: Sun Aug 2 03:28:20 1992 Message-Id: <9208011250.AA05792@relay1.UU.NET> Reply-To: Logical Language Group Sender: Lojban list From: Logical Language Group Subject: Apology X-To: lojban@cuvmb.cc.columbia.edu To: John Cowan Status: RO I kinda feel I should apologize to John and everyone else. My last posting in which I blew up over "de facto baseline" seemed excessive in retrospect. John has said that he understood what I meant, but I want to make sure{ others do. I instituted a policy of "baselining" portions of the language design because historically, Loglan has gone through many iterations of change, often major, constrained only by the judgeme{nt of the current design team (historically, this was JCB and JCB alone. pc, as editor of The Loglanist could also effectively make minor changes by his pronouncements and editorial asides, but these tended to be minor, and were passed-on by JCB who had the issues typed up and mailed). More recently, TLI set up the "Loglan Academy" which now has 3 people who can decide to change anything provided that it is unanimous among them. But the bottom line is that throughout Loglan history, people have invested a lot of time learning the language only to find that, for good reason or no, someone has{ come along and changed it, often drastically, with no notice or chance to comment. The baseline system is an attempt to change that practice, while still allowing us to complete and document the language design, and while doing so, to backfit and correct any 'bugs' that we find in things that have already been completed. In practice, as the almost sole publishing force in LLG, I have excessive power to make changes. If I were to suddenly, even through a typo, change a word or a grammar structure, then it "de facto" becomes the language. An example was the change of "gicmu" to "gismu" that resulted probably from my faulty memory on a word that I used often but had not learned well before then. Now the word IS "gismu". [John Cowan's now active role in backstopping virtually everything I do in language design, and his lead in the grammar YACCing now has reduced my power considerably, in that I am less likely to be able to slip in a change unnoticed. On the other hand, John could probably slip in YACC grammar changes almost unnoticed since they are relatively unchecked by me or anyone else until a baseline review comes along - at whcih time little changes are likely to go unnoticed. But big changes are at least unlikely.] Thus, to a considerable extent, the baseline policy is a promise from me to all of you to not change the language hap{_hazardly{, to consider everyone's investment in learning the language before making a change. A baseline is a place where I have accepted a community determination that the language is done - that determination having been made by a vote at LogFest - and where I have thereby committed to actively defend the community's decision. After a baseline, I consider any significant change to require a new community vote specifically on the issue, but even the most minor change requires an explicit community vote to approve a new baseline. ("The community" in this discussion is the dozen-odd voting members of LLG, but practically speaking, any signififcant objection in the community is brought to their attention by someone.) Baselining is thus a check on my power (and John's power in his sphere of the design) as well as a promise. As many people know, I consider promises rather sacred - in many ways Lojban exists because I made a promise to people to bring them the version of Loglan that they could use without constraint by JCB, TLI, or anyone else - a promise that I feel honor bound to implement even with the considerable cost to both my time and pocketbook that has ensued. Because I see a baseline as a promise, and promises as sacred, the concept of "de facto" baseline is anathema to me. It would mean that I am committed to keeping a promise that is attributed to me, but which I did not make or agree to in any specific way. (If I disagree with a community decision on a baseline, I have the option to resign my role as a leader of the community.) This does not mean that aspects of the language that are not baselined are completely uncontrolled. Discussions in my rafsi report presumably made clear that many{ rafsi in use are considered 'sacred' because established through considerable use in the community. But no decision has been made on the rafsi list as a whole, and I have specifically argued against such a decision because, as John pointed out, usage even including all the new data that I incorporated is not that extensive for much of the language. Since almost no one has learned or used most of the rafsi in the language (whereas my statistics on gismu use showed that{ more than half the gismu have appeared in Lojban text or related co{mments at least 10 times, and of course most of the rules of the grammar, excluding MEX, have been used quite extensively.{ Thus I have considered a baseline of the rafsi up till now unwise because the design hasn't been tested. I consider that the set of rafsi assignments that we had initially has now drifted far enough from ideal that a significant change was needed. because such a change means that a substantial amount of Lojban text is affect, and a small amount of people's learning is likewise affected, I feel obligated to o offer a formal baseline as the price for such a revision - in effect a promise never to do it again, or even to consider doing it again unless something turns out to be massively broken - unlikley in the case of rafsi assignments. After a baseline, I will be obligated to protect the design against unwarranted change, not merely to seek community comment before putting the changes in as I've done by way of this rafsi report. Before a baseline, therfore, a c change is permitted for optimaization - afterwards, a change is permitted only if something is WRONG. Before a baseline adding a rafsi is trivial, deleting a rafsi assignment to become unassigned is not much more offensive, but even changing a rafsi assignment is no small shakes provided that there is not a lot of usage. After the rafsi are baselined, in spite of what John said, I will consider it no more possible to add a new gismu, or a new rafsi for a new/old gismu than to make a change or deletion, without seeking full c consent by the community ON EACH SUCH CHANGE. You would not see a proposal like this rafsi report, but a series of 160 s4eparate messages, spaced out so no one is overloaded by the numbers, and giving written explanation and justification for each change. The formal grammar, the least stable of our baselines, had only 28 changes, all fully explained and documented before the baseline was updated. A 2nd rebaselining for the reference book publication will probably have about 18 more changes, again each fully explained and documented. This has rambled a bit, but I hope people understand why I treat the word 'baseline' so sensitively. I want people to feel that they can learn Lojban without having it change underfoot, but I also want the language, as defined, to reach as large an audience as possible. I can't do this while keeping part of the language open for the design to be coompleted and docuymented without the baseline concept, andthat concept must not be fiddled with in order for it to have meaning - if de facto baselines exist on things that have been published, then anything in the language that changes by decision as opposed to my evolution in usage is wrong or at least subject to some type of formal community vote. That will be the policy on the language during the 5 year baseline period after the dictionary and textbook are done. I don;t want it to be the case before then, or that 5-year period will be meaningless. And if the standard of change for the last few years is retained at the time of that language-wide baseline, then the 5-year period will also be meaningless. I don;t want that to be the case - to me, the ability of Loglan/Lojban to become a living language independent of both inventor and design team requires that 5-year period in which the apron strings are totally severed. lojbab