Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10 beta 3/9/83; site nbs-amrf.UUCP Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!ihnp4!zehntel!hplabs!hao!seismo!umcp-cs!nbs-amrf!hopp From: hopp@nbs-amrf.UUCP (Ted Hopp) Newsgroups: net.lang Subject: Re: Standardization Message-ID: <428@nbs-amrf.UUCP> Date: Sat, 16-Feb-85 13:33:05 EST Article-I.D.: nbs-amrf.428 Posted: Sat Feb 16 13:33:05 1985 Date-Received: Thu, 21-Feb-85 06:39:46 EST References: <283@gumby.UUCP> <5125@tektronix.UUCP>, <289@gumby.UUCP> Organization: National Bureau of Standards Lines: 85 > > ..... Whether you personally accept it or not, standards exist for > > Pascal for both the ANSI/IEEE and ISO versions. > Sorry. I am of the opinion that the plural of "standard" is an oxymoron, > unless you're referring to gas stations. It's similar to saying "multiple > universes" -- there is only ONE ("Uni"). If you have more than one "standard" > then you have none. To say, "This is ISO standard, that's ANSI" means that > neither one of them is really STANDARD. Sure, I can write my own compiler > and declare it to be O. W. N. Standard, but it will STILL be rejected by > some other "standard" compiler. > -Steve Patterson, UWMadison Steve Patterson expresses a widely held misconception of the nature of standards. The purpose of establishing a standard is to provide a common reference point for some purpose. In the case of languages, of course, it is so compiler writers can say "My compiler implements ANSI Pascal" or "... ISO Pascal" and people KNOW what is being offered. It can also be turned around, so I could say "I require ANSI Standard Pascal" and someone would know what I would buy. The fact that there are more than one standard for Pascal means that I have a choice of standards from which to pick. (Whether a choice is a good thing depends on the subject of the standard.) The fact that there are standards means that there is at least one thing I can point at to indicate what I want (or have). Standards are used to (1) foster competition, (2) provide guidance to vendors, (3) reduce marketplace risk, (4) protect consumers; they are (almost) never used to (5) advance technology. None of these functions of standards require that there be a SINGLE standard for something. At times, a single standard exists because there is only driving force behind standardization; Ada is an example. The Ada specification started out as just that, a specification by DoD of the language they would buy; it was not proposed originally as a "standard language" in any sense other than that. That it became an ANSI standard was because of the huge market impact any DoD decision like that has. In this country, nearly all standards are voluntary. The few that aren't are mandatory by explicit legislation by Congress or state legislatures. This is not true in other countries. In many European countries, for instance, if there is a national standard for X, you can only call something X if it complies with the standard. (Of course, "complies" may have a strange definition, but it is defined.) Standards have a legal standing in the international community quite different than in this country. ISO is where different countries hash out how their individual national interests will be reconciled. What makes a standard? Giving it a name (like "O. W. N. standard") doesn't quite do it. Standards are established in this country by organizations created by parties interested in the subject. These organizations are most effective when they associate under the umbrella of the American National Standards Institute. (Not a governmental organization, by the way.) A definition becomes a standard because there is (formal) agreement among interested parties to accept it as a standard. Formal are always developed through intense negotiations among interested parties in these organizations. There are also "de facto" standards, such as BSD 4.2. These are riskier to use unless you are Berkeley. They are called standards because the producers of the product have credibility in the community of interested parties. Obviously, Berkeley maintains its credibility by treating its products as if they were formally standardized as described above. If it shipped a different product every time they developed a new idea and still called it the same thing, people would rapidly get disgusted and either walk away or start formal standardization procedures. (Power to the people! (Many times, vendors standardize their products formally just to get people to accept them.) The formal standards setup may sound elitist, but it really is better (in my opinion) than having legislated standards. If you have developed the quintessential standard for X, you still have to (1) establish your credibility to the other interested parties and (2) sell them your idea. This is called "free market enterprise" and I like it. We are all better off if Sony has to convince its competitors that Beta format is a good standard for video tape, rather than if Sony just has to lobby an "independent" official standards organization. Evidently Congress thinks so to, since that's the law. (There are anti-trust laws to prevent exclusionary practices by standards organizations; one standards-setting organization [ASME, I think] got into trouble a year or so ago because they hadn't instituted safeguards to prevent abuse by its officers in interpreting existing standards regarding steam boiler safety valves or some such.) -- Ted Hopp {seismo,umcp-cs}!nbs-amrf!hopp