What is a Standard at ISO?
by Rick Jelliffe
I've written before that there is something about s-words that make them difficult to use in conversation because everyone has different meanings: semantics is one, standards is another. I have previously argued against various views of standards, for example Microsoft's Mike Champion's Donut Line method of Standardization
(informality), or ODF's requirements creep for what makes an open standard
(though I liked Sun's
This time the merry-go-round goes back to Mike Champion: on his blog in January
I said Standards are a library of stable technological possibilities
. Regular readers won't be surprised: plurality is good. (If you attended WWW7 in Brisbane about a decade ago, you may have seen me debating -badly- against an MS rep, with me supporting the idea that XML needed to support an general infrastructure for binding resources to XML data in a layered approach to promote plurality while the MS guy thought just a big fat XML Schemas would be enough. Same deal.)
Mike replied I guess I have a different perspective: Standards are technological REALITIES that one can use with some confidence that they are supported by at least a critical mass of some audience.
I have two gripes with this idea. First, it unsoundly enthrones market success as the benchmark: but it is sometimes useful to have standards that are only relevant to a small sector of industry. The recent XML-DEV thread on the need for speed in parsing is an example, it seems to me, of a nasty side to standards: people who say "My need is important, your need is not" have no business in standards; whether I concur that your reasons for a standard are sound, if it doesn't affect me I should butt out. Second, it does not seem far from the first-come first-serve, exclusionary view of standards as if they are a kind of monopoly grant, like a patent. (This is the latest scam, I see.)
What do ISO say? Their website
is a good place to start.
Here are the first sentences of the ISO section "Hallmarks of the ISO Brand":
- Every participating ISO member institute (full members) has the right to take part in the development of any standard which it judges to be important to its country's economy.
- ISO standards are voluntary.
- ISO develops only those standards for which there is a market requirement.
- Although ISO standards are voluntary, the fact that they are developed in response to market demand, and are based on consensus among the interested parties, ensures widespread applicability of the standards.
- ISO standards are technical agreements which provide the framework for compatible technology worldwide.
Now I don't see anything there to connect ISO standards either with critical masses or monopoly grants: the market requirement aspect is the key: is there some group that (thinks) it needs it?
There is an especially interesting line concerning committee member's obligations: The experts participate as national delegations, chosen by the ISO national member institute for the country concerned. These delegations are required to represent not just the views of the organizations in which their participating experts work, but of other stakeholders too. According to ISO rules, the member institute is expected to take account of the views of the range of parties interested in the standard under development and to present a consolidated, national consensus position to the technical committee.
I see this as condemning the sectarian "My need is important; your need is not" approach and explicitly demanding an openness toward plurality. (Which is not to say that every crackpot scheme that comes along must be endorsed, of course.)
ISO standards are essentially technical agreements that reflect some market requirement, made by committees which are discouraged from sectional interest. The ISO process is geared to win-win, not win-lose: to encouraging agreement where possible but allowing separate technologies where agreement is not possible. And ISO clearly takes the view that regulation is not their business. That is why I think moving away from the view of ISO standards as a library of technical solutions will be unsatisfactory for those who attempt it.
I'm mentioned as a bad example of something, but I'm not sure exactly what. I don't disagree with anything in this post. I thought our disagreement in that blog thread was mostly due to my somewhat clumsy prose.
Here's a few things I believe that you may well disagree with, given previous conversations:
- There are plenty of "real" XML-related standards that Microsoft doesn't (yet) implement in any products, for a variety of good and occasionally bad reasons. Schematron is a good example -- it's a great spec, has some loyal fans in Redmond, is perpetually on our radar as something we might seriously get behind, but so far nobody has been able to make a compelling business case that it is higher priority than the other stuff on the TODO list. It's lack of critical so far doesn't mean it's not a real standard, but market success *is* a criterion we use in deciding where to invest. Granted there's sort of a Heisenberg effect with Microsoft and standards, but we've made the concious decision to thoroughly implement some core standards and to fully support those implementations for a LONG time, rather than spreading scarce resources more thinly.
- That't not to say "our need is important, your need is not." It's saying "we can't meet every important need", and we (increasingly) look to the larger community to fill those we can't. For example, we tell people who need Schematron support to use the reference XSLT implementation, those who need XQuery on the client to use Saxon, and invite people whose needs those approaches don't meet to let us know so that we can adjust our priorities.
- Nor is that to promote a "first-come first-serve, exclusionary view of standards as if they are a kind of monopoly grant". Perhaps you're referring to the fact that Microsoft doesn't implement RELAX NG (or directly implement Schematron)? That's not because XSD came first and has a monopoly on schema mindshare so much as a conclusion that the alternative schema languages don't solve enough of our real customers' problems sufficiently better than XSD does to be worth the cost to implement. If somebody invents a schema language that has a compact syntax that developers will use without extensive tool support, a useful type system that cleanly bridges the CLR, JRE, SQL, etc. types, offers a good way to constrain documents as well as data, can handle Schematron-style assertions, etc., we'd happily jump on its bandwagon.
- Informal ("donut line") spec de facto interop is a necessary but not sufficient condition for standardization. Whoever that "MS guy thought just a big fat XML Schemas would be enough" 10 years ago was, the wrongness of his assumption is appreciated in Redmond as much as, perhaps moreso than, in the rest of the world. The all-too-obvious failings of the committee-first standards process that produced XSD led directly to the implementation-first approach used for the web services spec standardization.
Mike: As I wrote before,
"Mike is a guy who opinions I respect enough to disagree with." I don't think that Mike is a bad example of anything, and apologize if it seems like that. Obviously I don't know why Mike persists in having different views to me, but that is true of the world in general.
The agenda behind the mention is not to promote Schematron or to criticize anyone for not doing it, time is on our side: its not about schema languages at all, even though Mike is right in thinking that I see XSD as another example of the needs of some (esp. publishing) being squeezed out by the needs of big-bucked others (esp. DBMS vendors, the "typists").
The reference to first-come, first-serve monopoly does not refer to MS at all. (Perhaps people are too young to realize that other companies have effectively held monopolies. Or that patents are a form of monopoly. "Monopoly" is not a euphemism for Microsoft.) It relates to an idea floating around (the new "scam") that tries to make anti-pluralism a virtue by drawing some kind of analogy with patents. Standards where no overlap or parallelism is possible (the "monochromatic" world I mentioned yesterday) are a form of cartel. So the blog was to say that this is not what ISO standards are about.
The monochromatic or barrier-to-competition meme comes out of the investor communities (think, VCs, hedge funds, equity groups) etc. This is an entirely different group from the technical standards groups. The problem is the first group has learned to play the second group for their own advantage. That is what I refer to as 'the standards game'. It's a dodge to establish a hegemony made more palatable by the "We are the Web. The Web is All" meme which was usefully played to HTML's advantage and now sets the example others follow.
We screwed ourselves in the early years of the witless fielding of the WWW. We refuse to admit it so we keep on doing it even as the lubrication is thinning. The Internet is plumbing. The web is a common linking standard. Content has different standards for different technical requirements (see Rick Kimball's post on the speed issues in VRML browsers I copied to XML-Dev).
Not only is plurality necessary, it is increasing. Does that mean HTML goes away? Of course not. Will OOwhatsis this week trump OOwhosis next week? No. The investors will make sure that never happens.
And so it goes.
No need to reverse engineer a definition from the ISO web site. ISO plainly defines a "Standard" in ISO/IEC Guide 2:2004 "Standardization and related activities -- General vocabulary":
A standard is, "a document, established by consensus and approved by a recognized body, that provides, for common and repeated use, rules, guidelines or characteristics for activities or their results, aimed at the achievement of the optimum degree of order in a given context.
NOTE Standards should be based on the consolidated results of science, technology and experience, and aimed at the promotion of optimum community benefits."
IBM's Rob: That is a definition of standard in general, but it does not bring out the characteristics of (standards at) ISO, which is the topic. Your strategy is AFAICS to enlarge the definition of standard so they are not voluntary (but instead are patent-like and regulatory at the ISO level) and to restrict the definition so that they are nen ot responses to market demand (but instead must be exclusive); by tailoring expectations for what a standard should be, you then can argue that being pro-ODF requires one to be anti-ISO_OOML and then can blame ISO if Ecma OOXML does get through. Preventing ISO OOXML help sell your company's products, but it won't help the people who need to (i.e. have a 'market requirement' for) an ISO standard format for exact dumps of their Office documents. Being pro-ODF does not require one to be anti-ISO OOXML, which I bet is Novell's position for example.
"an idea floating around (the new "scam") that tries to make anti-pluralism a virtue by drawing some kind of analogy with patents. Standards where no overlap or parallelism is possible (the "monochromatic" world I mentioned yesterday) are a form of cartel. So the blog was to say that this is not what ISO standards are about. "
Oops. Sorry I misinterpreted, shouldn't comment on blogs too early in the AM :-)
Tying the schema languages permathread together with the docuemnt formats permathread at my peril, it's definitely a Good Thing that there are standard options for both. DTDs, XSD, RNG, Schematron all meet the needs of different groups; HTML, ODF, OOXML, and DocBook also cover the needs of different audiences. In both cases, those that try to cover the most territory are the most complex; the simplest may well meet 80% of the world's needs, but that last 20% is where things get fragmented quickly. Invoking the Pareto principle, or fantasizing about how simple the world would be if only everyone did things one way ("our" way, of course, whoever is doing the fantasizing) doesn't help those who live in the hard side of the 80/20 point.
Mike: Pareto yes. The issue is whether we have the 80 technology and the 20 technology reversed (i.e. using grammars rather than paths).
Also, akin to my "library" idea, I see that Ecma's response to the OOXML contradictions speaks of tools in a tool box.
1) Worked with weapons = COLT Weapons
2) Worked with aerospace = HUGHS Aircraft
3) Worked with automotive = GM Dies
I am 56 and have been well trained.
Tool & Die background.
Special Projects - Engineer.
Trained Programs = a Double (x2) Analytic Person.
Maybe you can prove that (I.S.O.) is a crap of "BULL".
--> I.S.O. +
I am at the wrong web site.
A money making system for SUITES !! --> Executives !!
I have implemented ISO 8001 --> Molding Shop.
Wrote the complete manual. and was approved.
I am an approved auditor for Q.S. 9000 --> Automotive.
I.S.O. = We Say What We Do,--> Do What We Say.
= A lot of gray ares.
= No Black + White (Bull Section)
Q.S. 9000 = Automotive.
= All Black and White.
= I.S.O = Not approved !!
Military = (was --> not sure now !!)
= MIL - ....
= A.Q.?.P. ??
Aerospace = STD / New , Must Be Approved !!
So what is I.S.O. --> For ??
- Automotive = NO
- Weapons = Approved Supplier - Only
- Aerospace = Approved Supplier - Only
Has taken over the manufacturing sector !!
1) Ignore any and "ALL" patents !! (Does Not)
Patent's do not exist here.
FREE enterprise only.
2) No Lab Testing Done = Safety --> Electric
No Research and Development required (R+D) = Stolen.
No Safety Factors Required.
3) CHINA has filled the North American Country with "NO"
standards at all.
1) Materials Used = Lead/Impurities (Toys)
2) Safety (Electric Wire Standards)
May I say more !!
I.S.O. = Old Dead Standard = For What !!
Best Regards, -ED-