From cbmvax!uunet!Think.COM!gls Mon Jun 17 18:29:16 1991 Return-Path: Date: Mon Jun 17 18:29:16 1991 Return-Path: From: Guy Steele Message-Id: <9106172059.AA08885@strident.think.com> To: cbmvax!snark.thyrsus.com!eric Cc: gls@think.com, lojban-list@snark.thyrsus.com In-Reply-To: Eric S. Raymond's message of Fri, 14 Jun 91 18:44:16 EDT Subject: xebro Status: RO From: cbmvax!snark.thyrsus.com!eric@uunet.UU.NET (Eric S. Raymond) Date: Fri, 14 Jun 91 18:44:16 EDT ... > Now talk of the > future is necessarily approximate and speculative, so it's okay > to use "is", which has the same properties, in that kind of framework > for most purposes. In fact, I don't think your conclusion follows from your premise. If talk of the future is necessary speculative, semantic hygiene is *more* important, not less --- it is *more* important that `predictive' equality be marked as a semi-predicate, a slippery thing. It is important that it be understood, but if it is universally understood it need not be marked. A loose reading of Zipf's law is that only the unusual should be explicitly marked; the more unusual, the bigger a mark is merited. That which is universally true should be universally suppressed. (Application in programming languages: C has a unary "*" operator for pointer dereferencing. This is appropriate, because the use of a pointer is less usual in C than not using a pointer. But it is only somewhat less usual, so it merits a small marker. Compare this to Lisp; pointer dereferencing is /never/ marked, not because there are no pointers, but because /everything/ is a pointer.) --Guy