ISO to Require Royalties?
September 24, 2003
As a computer geek I sometimes use technical abbreviations of common terms as a kind of conversational shortcut, especially when chatting on IRC or composing email. The set of abbreviations I use most often covers the names of countries, the same set which is used for national top-level domain names in the global DNS. It is not uncommon to see geeks write things like the following:
geekboy: Have you ever visited IL or MX? geekgal: No, but I'd love to visit CA and SA! geekboy: How about ISO headquarters in CH?
Most XML developers know that these two-letter country codes come from the ISO (International Organization for Standardization), particularly ISO 3166.
But now it has come to the attention of the W3C, as well as various other communities, that the ISO is thinking about imposing licensing fees for the commercial use of several of its standards, including 3166, the one which establishes country codes, as well as ISO 639 and ISO 4217, which establish language and currency codes, respectively.
In the rest of this article I provide some background to the present controversy as well as sample the reaction of web and other internet developers and development communities.
I have long been frustrated about the ISO; many of its standards, including the ones already mentioned as well as ISO 8601 (which covers the representation of dates and times), are fundamentally important to web and internet programming. This makes them the sorts of documents a curious developer might want to read or have handy as a reference.
But, unlike either the IETF's RFCs or the W3C's notes and recommendations (not to mention both organizations fairly transparent work group product policies), the ISO's standards documents themselves are generally unavailable. Well, it's not that bad, except that it is. ISO standards documents are available for purchase from the ISO site, but they are not cheap. In many cases they are prohibitively expensive for any but the most well-heeled organizations and individuals. As curious as I am, as a lone developer without organizational support, paying the following amounts for these documents is just not a viable option:
- ISO 639-1:2002 → $94
- ISO 4217 → $89
- ISO 3166-1:1997 → $109
- ISO 8601:2000 → $80
In other words, it will cost me $372 to download four PDFs from the ISO site, on the off-chance that I want to actually consult these crucial standards. This is less than ideal.
Also in XML-Deviant
These aren't, in truth, particularly telling examples of the ISO's bungling. As a Python programmer, I have at my disposal several libraries and modules which allow me to consume and produce identifiers for countries, currencies, and languages which have been standardized by the ISO. The same is true for C, C++, C#, VB, Perl, Java, Lisp and Scheme, Smalltalk, etc. It's probably true for most programmers, no matter their language of choice. (Of course, not having easy access to the standards documents themselves, I may not be able to fix bugs or mistakes in these modules.) But what about some of ISO's less popular standards?
Imagine that I would like to implement a system in Python for building multilingual thesauri. I want to build this system for non-profit institutions in the developing world, so I don't expect to be paid for doing it. My target audience needs to be independent of proprietary vendor solutions as much as, if not more than, anyone else. Standardization in such contexts just isn't optional. I already own a computer; my development environment, including a sophisticated programming language, extensive range of libraries, and a nifty IDE, cost me nothing.
I discover that there is an ISO standard, ISO 5964, which is relevant to my task. ISO 5964 provides "Guidelines for the establishment and development of multilingual thesauri". Precisely what I need. But the ISO suggests that I use ISO 5964 in conjunction with ISO 2788 ("Guidelines for the establishment and development of monolingual thesauri"), so I need that, too. Total cost to download these two PDFs? $198.
On second thought, maybe I will forego building my project on ISO standards. Maybe I'll just drop the project altogether? I can always find my way to a university library and consult the standards, if the library owns them, which is by no means assured. But not everyone is so lucky as to live within reasonable proximity of a public research institution.
The situation would be less frustrating if these prices reflected something like actual publication costs. After all, the ISO almost certainly arranges relatively small printing runs and may well, for all I know, insist upon a very high quality of document production and printing, befitting the use of these documents as standards. However, the ISO charges precisely the same amount for a PDF version of a standard downloaded from its site as it does for a printed version which it mails to you. That suggests that the fees are not about printing costs but are really information access fees.
I do not believe that information access fees are necessarily problematic. I often pay money to access information and regularly think I've gotten a good deal. However, the ISO is supposed to be motivated by achieving all the goods which can be achieved by widespread (global, in fact) usage of the standards which it develops. It costs money to develop standards, of course; but recouping those costs by charging people exorbitant fees to access the standards themselves has always seemed to me a pretty bad idea. Granted, my complaint, about not being to afford to pay the fees as a lone developer, may be an uncommon one. If so, then more power to the ISO; but it's hard to imagine that I'm the only person bothered by this or that I'm the only person who's adoption or promotion of ISO standards has been hindered by the ISO itself.
This complaint brings us to the present controversy. The ISO, rather than taking a step forward to make possible wider access to its standards, is taking the opposite tack. It's considering requiring license fees be paid for the commercial use of the information contained in ISO 639, 4217, and 3166. As Robin Cover has reported, the ISO is suggesting that "'software developers or commercial resellers requesting permission to embed the data elements contained in an ISO Code in their products for resale will be asked to purchase the Code in electronic format and pay either an annual fee or a one-time fee and any applicable maintenance fees required.'"
It's not clear how far reaching this requirement might be. Would it require my Linux distribution maker to pay a fee for selling a CD that contains software, like the Python language and libraries, which uses ISO identifiers? What about all of the Web software, both client and server, which uses language and country codes extensively? The W3C's XML Recommendation makes reference to ISO 639 and 3166. Does that mean any product which uses an XML parser owes the ISO a fee?
For what it's worth, the ISO isn't seeking a license fee for the use of these codes in non-commercial software, and it makes some of the codes -- but not the text of standards in which the codes are established -- freely available on the Web. You can even get country codes in a (particularly awful) XML document. (Note that the country codes established by the alpha-3 variant -- which, according to the ISO 3166 FAQ, "allows a better visual association between country name and code element than the alpha-2 code" -- of ISO 3166 are not available except by purchasing the standard.)
At least three important institutions have responded to perceived change in the ISO's licensing policy: the W3C, the Unicode Technical Committee, and INCITS.
The W3C's Response
Tim Berners-Lee has written a letter to the ISO President, Oliver Smoot. The point of Berners-Lee's note was to express "deep concerns" over the ISO's proposed policy change, especially since the Web uses these ISO identifiers extensively, particularly in its internationalization efforts. At least three negative results are likely to follow if the ISO insists upon this policy change, according to Berners-Lee. These include "fragmentation, delay in deployment, and in effect a lack of standardization"; in addition, non-English speakers will be the ones who are harmed the most. Berners-Lee further adds a point which is congruent with my general frustration with the ISO:
Web technology today allows publication and reuse of data at a small fraction of the costs a few years ago. If it is the case that the costs of maintaining these databases is beyond ISO's capacity to cover, we would suggest that ISO open a discussion with the larger user community about how these services might be hosted in a manner that covers these costs.
If anything, Berners-Lee actually understates the centrality of the ISO's country and language identifiers, not only to the Web but to the internet itself (though I appreciate that's more the IETF's brief). It's not altogether clear, however, that, if the ISO insists on moving forward, alternative identifiers couldn't be arranged, perhaps even arranged easily. The problem, of course, wouldn't be forming new identifiers -- perhaps using URIs or URNs -- but the costs of retooling to use the new identifiers, as well as the transitional opportunity costs, which might be significant.
While the W3C-ISO jousting -- particularly between RDF and Topic Maps -- has often been healthy and conducive to innovation and technical diversity, in this case the ISO should tread lightly since the W3C is pretty well placed to create alternatives to the ISO's country, language, and currency identifiers.
The Unicode Technical Committee's Response
The Unicode Technical Committee (UTC), which makes decisions about the direction of Unicode, has also issued a response to the ISO's policy proposal. This response focuses on two points: the negative results of the new policy if its adopted by the ISO and its illegitimacy. The UTC suggests that potential users of the ISO identifiers may "avoid future contributions to ISO standards development"; "avoid using or referencing ISO standards"; "develop and use alternative, royalty-free, standards." Each of these seems likely, especially the latter.
As for the illegitimacy point, the UTC suggests that, since these identifiers are "based on contributions from other sources or duplicate pre-existing data," it's not clear that the ISO can successfully impose licensing fees for their use, commercial or not, without opening itself up to "debates over intellectual property rights and financial liability."
The INCITS Response
INCITS, the InterNational Committee for Information Technology Standards, has responded to the ISO's policy proposal by communicating with ANSI directly. The heart of INCITS's response is the suggestion that the proposed policy would impose an implementation fee for using the standard:
the proposal being discussed would in effect place a charge upon implementing a standard by enforcing a fee associated with each copy of a product built according to or incorporating the standard. In essence, therefore, this charges users of a standard, be they direct (in the case of manufacturers) or indirect (in the case of product consumers) to actually use the standard.
INCITS requests that ANSI adopt a threefold policy: that usage fees are "inappropriate" (alas, INCITS supports fees for purchasing standards); that ANSI oppose the policy; and that it work to "dissuade ISO and its members from this approach."
Does the ISO need a reliable means of funding? Absolutely. But it needs, at least in my view, a way which is independent of selling, at least at such exorbitant rates, its standards themselves. If it's a truly global standards body, it should be able to find funding from the UN (which might be able or more inclined to fund ISO if the US would pay its delinquent UN dues), from wealthy western nations (why not, since the G7 benefits the most from the ISO's work?), and even from philanthropically-minded individuals and corporations.
However, some things which the ISO has standardized -- and language, currency, country identifiers, as well as date-time representations, are among those things -- should be put immediately into the public domain. Some of its standards are simply too crucial and too much in the public trust to be tied in any way to the ISO's revenue model.