Path towards a multilingalization IUse referent
Dear John, If I supported the IDNA2008 consensus it was because it suited my vision of the Internet architecture. This was a surprise. This meant that it was possible to read as the two faces of the same coin * the 2011 "inside Internet" that the IETF influences and documents. IANA glossary, PRECIS propositions * and the 2011 "outside Internet encapsulation" that we, its users, want to intelligently utilize . ICANN/WG/VIP, IUCG. Actually as the two sides of a smart interface that I call the IUI as: * the Internet Use Interface on its Internet side. * the Intelligent Use Interface on the User's side, because the IUI can also interface the same user with other technologies that will converge at the IUI. As you know this vision of mine raised two main problems:(1) where to document the IUI and (2) the need to experiment in order to correctly document it. ---- 1. My appeals to the IESG and IAB permitted the IETF to answer the first question. The IESG proposed for me to run a BOF about what it considered as research, but the IAB made it clear that the IETF was concerned but that it was not in its bailiwick. My evaluation is that the IUI is an "IUser" (Internet/Intelligent/independent, etc. user) community issue. This community of identified users, should be liaised with the IETF. I use the IUCG for that and the help/support/ideas of its small team. ---- 2. The particular IUI architecture that I investigate is in line with the architecture that I use, support, and have deployed for more than thirty years. 2.1. It is based on my extended systems understanding of: * the way systems, diversity, complexity, and simplicity are articulated (much in tune with the three basic principles of the Internet architecture: constant change (RFC 1958), simplicity (RFC 3439), and subsidiarity (as this results from the IDNA2008 architecture, once addressed the architecturally wrong location of IDNA in applications instead of at the IUI). * the particular case of communications where transmitted contents use static metadata in telecoms, metadata contained in the packet in datacommunications like the Internet, and separated metadata in what I call the metacommunications. IDNA2008 the way it implements the presentation layer in the Internet is at the border with metacoms. 2.2. From experience (Tymnet Extended Services), I think the location of the metacoms support is to be on the user side so that it is transparent to the network use. This is the Internet PLUS architecture, i.e. plugged additional OSEX layers (Extended OSI) on the user side. This layers can also be faked in a centralized manner, and result in the "+" approach: e.g. Google+. 2.3. The way IDNA2008 is designed calls for two necessary moves: * to use the DNS to deliver network oriented metadata, hence by synergy in order to, most probably, probably use DDDS registries for intersem (semantic internet layers above) referent data management and an IPv6 like registry addressing. In any case JTC1/SG32/WG2 is to be carefully considered and I miss the time for that. * a total separation from the Unicode typography that is not able to easily cope with network use requirements and to support human oriented algorithms. This separation does not need to be repudiation. However, the Internet MUST be supported by a network/human oriented universal semiotic system. I think it can start with a graphic (passive) and then a dynamic sign oriented approach in CLASS 0, as a common safe support of the other classes above, including the IN class. ----- 3. The danger of what we found and in such architecture is that it actually is the Internet technology architecture. There is not a single change in the existing RFCs or in a single bit in the existing protocols that is needed. However, there are a lot of opportunities for adjustments and, at the same time, there is major pressure being imposed on the whole architecture and on in particular on the ML-DNS (i.e. an IDNA2008 compliant encapsulation of the DNS). This pressure is the gTLD project of ICANN with its gTLD fee and delays. This may lead to a lot of individual architectural "tunings" to evade ICANN or to protect ICANN; in addition to these "adjustments", such as PRECIS. ----- 4. My first idea was to use the notoriety of ICANN's plans and Google Public DNS to experiment IDNA2008 compliant free gTLDs (like Projet.FRA). Then to proceed from there on, before anyone could try to make a business of it. A quick and dirty move to force everyone to understand where we are, how weak the Internet is IRT the market and sponsors, make people integrate diversity into their network thinking process. For personal reasons you know I was not able to dedicate myself to this. At the same time, I did better explore as to what the very initial visions of the Internet actually brought. Not only the "networking group", but others too. There is a solution I wish to try to revive, if possible in parallel (partial code integration). However, if that solution has been disregarded it is probably because of the complication that the IETF has reached. So the priority for me is to clarify and simplify the model. You already did a lot with IDNA2008, but it is not enough as we stayed (Charter) within the IDNA concept. We have to come back to fundamentals. The Internet MUST be considered from the outside as a unique and simple system (service black box) that is able to support several, and more or less equivalent architectures, on the user side. I agree with John Day: we miss an Internet OS. To implement it calls for a clearer and coherent technical culture. The internet tomography is still to be done. The earthquake is the RFC 5895. It is from there that we need to proceed. In two directions: 4.1. to think, accommodate, test, and deploy a simpler, clearer, etc. Internet of today, i.e. multilingual and semiotic ready (support of passive [as today] but also active and localized content). 4.2. to make sure that the "patches" being discussed in parallel do not conflict and block that innovation. My understanding (bet?) is that, as usual, the single point of simplicity to obtain this is for everyone to use the same language and for the concepts underlying that that language to be open enough so they permit to clearly spell out the "post IDNA2008 possible". This is why I engaged the IUCG to compile all the data, ideas, demands, confusion, etc. on multilingualization (i.e. architectural linguistic neutrality) that are discussed. We are probably still missing many things currently, but we have compiled 73 dense pages. ---- Your misunderstanding about the way we use wikis (as a quickly visible to all mailing archives, that can be translated in the MindMap), has led me to work this afternoon for you. And to build the next step. This still is NOT a document; this is only a next phase IUse community working wiki, located at <http://iucg.org/wiki/Multilingualization_Glossary>http://iucg.org/wiki/Multilingualization_Glossary. This is rough stuff. My intent is to digest it through a clear network ontology and model (ontography) and tune it until its coherent with what the Internet IS and what our IUI MUST be. So that it can become a reference glossary. Obviously, there will be parts that do not belong to the IETF area, but to the IUTF area (the emerging IUse technical TF). That is because of our IUse center of the world is not your IETF/Unicode center of the world. However, we need our common world to entually be unique even if all this takes time to think, adjust; set-up. This is why I am making the place of that work visible to well-intentioned people and why I give my deep thanks to those who help. jfc
Dear colleagues, I did not read the rest of the message past what I quote below, becase it did not seem to me that the structure of the message would answer the question I pose. Therefore, I make a request: On Sun, Aug 21, 2011 at 01:08:29AM +0200, jefsey wrote:
However, the Internet MUST be supported by a network/human oriented universal semiotic system.
I would like a defence of that claim. Speaking entirely personally, I don't believe it. We shape our technology, but our technology also shapes us. There is nothing intrinsic about "green" meaning "go". There is nothing about the universe that causes time zones (which is different from saying "differences in the apparent time of dawn and dusk"). Yet we all, techno-literate that we are, take these bits of cultural dreck and make them true. Moreover, we adjust to them -- somehow finding it more convenient to say, "Darkness falls earlier," than to say, "The time on the clock is wrong." It is, in my personal opinion, completely foolish to imagine that Internet names -- which are, by their nature at the time of registration and lookup, even if not for the user, completely devoid of cultural context -- can ever be made completely user-centric. Users will have to learn some things, and some of those things will be "Internet names are a little different, & have strange restrictive rules." More importantly, while I am not one of those people in the thrall of the conceptual scheme (cf. Davidson), I think it is preposterous to suggest that we will come up with a universal semiotic system given the limitations of the DNS and the way that denoting works in different linguistic traditions. I believe we can come up with some useful conventions that will work most of the time, for most people. I believe that these will strain the rules of every writing system on the planet, in much the way that "ns1" strains English writing rules. I am incredulous at the suggestion that the Internet needs a "network/human oriented universal semiotic system," both because I don't believe it and because I believe that humans (and especially human language) are much more resilient than that. Best regards, Andrew -- Andrew Sullivan ajs@anvilwalrusden.com
At 05:24 21/08/2011, Andrew Sullivan wrote:
Dear colleagues,
I did not read the rest of the message past what I quote below, becase it did not seem to me that the structure of the message would answer the question I pose. Therefore, I make a request:
On Sun, Aug 21, 2011 at 01:08:29AM +0200, jefsey wrote:
However, the Internet MUST be supported by a network/human oriented universal semiotic system.
I would like a defence of that claim. Speaking entirely personally, I don't believe it.
Dear Andrew, I cannot object faith, but I can explain why I believe we think the same things.
We shape our technology, but our technology also shapes us.
Correct. But the way we shape our technology is diverse. Up to now IETF accepted that the user's response to their options was the "market". The IUCG is the acknowledgment that the response and the possible counter-response may first come from lead users, i.e. the people who know in advance what they want and are able to implement it by themselves. The IUCG mission is to welcome those who are ready to cooperate with the IETF, i.e. considering themselves as IETF users. IDNA2008 has significantly extended that role. This results from the IUI as I just explained to Patrik. This is because the IUI is an IETF/Intelligent Users interface, where users are not only IETF users but also IETF technical partners. In this it is necessary for the IETF: - 1. to understand and accept what the IUsers want to implement. - 2. to support it. In the IDNA2008 case, Point 1 is in part lower key because Point 2 has already been carried out: this is IDNA2008. However, it is also important because: - the IDNA architectural concept does not scale : nothing guarantees that two applications on the same machine will resolve the same DN into the same IP. - the guidance of IETF people, like yourself, is called upon by groups of users (e.g. VIP) - Post-IDNA2008 internal protocol updates are advisable (PRECIS) and RFC 1958 calls for the same problems to receive the same sollutions. - it is most probable that the outside-IUI choices will sometime interact with the inside Internet deployment. This is where we shape our IUI technology that may shape our Internet technology. By essence the Internet technology is resillient to our shaping. By essence the IUI technology wants to support and adapt itself to the Users' ways.
There is nothing intrinsic about "green" meaning "go".
It does not mean "go". It means its OK. If you steered at sea in the dark and poor weather you know that only three colours are really distinguishable: white/yellow, red and greed (blue can come from the moon and some cultures confuse it with green). - white means "I am here", - red means "furnace, heath, danger", - and green means "peace, cool, no risk" as in landscapes.
There is nothing about the universe that causes time zones (which is different from saying "differences in the apparent time of dawn and dusk"). Yet we all, techno-literate that we are, take these bits of cultural dreck and make them true. Moreover, we adjust to them -- somehow finding it more convenient to say, "Darkness falls earlier," than to say, "The time on the clock is wrong."
I will not dispute that! The same as you adjust to your wife, to others, to your own weight and size, your personal gifts and weaknesses, etc. Organization and technologies become part of us. This is the reason why we prefer they are smart enough and in tune with our own ways. Not easy with the Internet (e.g.: ASCII DN), but we want to ease that (e.g. IDNA2008 we discuss here) supporting people's languages. Why just in part, when we can consider the whole thing?
It is, in my personal opinion, completely foolish to imagine that Internet names -- which are, by their nature at the time of registration and lookup, even if not for the user, completely devoid of cultural context -- can ever be made completely user-centric.
I feel this comes fom your Internet DNS responsbility bias. Please remember that users are not interested in Internet DNS but in naming what they want, something they do all the day long. The difference between engineers and users is that: - you talk about "internet names", we talk of "domain names" that can me on resumes, TV, books, etc. - you talk of registration and look-up as if they came first. We observe that what comes first is the person asking for the name he/she wants. - you rightly consider that for the computer process the DN string is devoid of cultural context. This is the opposite for the human choice to have registered that particular string. - you seem to consider user-centric as user-exclusive. I only talk of "network/human oriented" semiotic system. Please also remember that semiotics includes three (actually four on my PoV) disciplines: - syntax - includining orthotypology - semantics - meaning - pragmatics - context influence - variances - multilinguistics - practical coexistance of languages - including computer languages: what we discuss in here.
Users will have to learn some things, and some of those things will be "Internet names are a little different, & have strange restrictive rules."
This is where we agree: actually they will necessarily have to learn the DNS orthotypographic ways we will have decided together. Every semiotic system is necessarily conventional (protocol) between interlocutors. The difficulty is to support the orthotypography people expect and/or are able to technically impose (e.g. Project.FRA and French majuscules) and that will most probably evoluate with time, new technologies and contexts. - http://tools.ietf.org/id/draft-iucg-afra-reports-00.txt - IESG indicated that they did consider it before approving IDNA2008. - http://tools.ietf.org/html/draft-iucg-idna2008-orthotypography-00 - http://tools.ietf.org/id/draft-iucg-punyplus-03.txt
More importantly, while I am not one of those people in the thrall of the conceptual scheme (cf. Davidson), I think it is preposterous to suggest that we will come up with a universal semiotic system given the limitations of the DNS and the way that denoting works in different linguistic traditions.
I only talk of "universal", i.e. "used, intended to be used, or understood by all", not of anything exhaustive (in such a case you would be correct). What is on the table is a "network/human" relative need. How network systems and human expectations can fit together.
I believe we can come up with some useful conventions that will work most of the time, for most people.
Sorry, as a user of the deliverables, I cannot accept the human "most of the time/most people". Mecanology calls for an algorithm that works or fails without hesitations (you were the one to call for an algorithm and you were right). 1. As far as the technology is concerned the matter is over. This is IDNA2008: IDNA2008 does not consider variants. 2. So,the remaining issues are: - how do we define, protect, register, etc. variants at a community level (algorithm)? - how do we support their operations: aliases inside the DNS or user domain name massaging at the outside ML-DNS? To clarify our job let have a look at the current definition of "algorithm" in http://iucg.org/wiki/Multilingualization_Glossary Algorithm A term used in a broad sense in the Unicode Standard, to mean the logical description of a process used to achieve a specified result. This does not require the actual procedure described in the algorithm to be followed; any implementation is conformant as long as the results are the same. (source: Unicode glossary, author: undisclosed, text listed to be reviewed and possibly changed). This means that we have to specify the result first. Before deciding if this is possible or not. So, if we want to succeed we have to specffy something we can achieve. As far as I am concerned I would be happy with an algorithm preventing TLD homographic confusion and supporting reasonable TLD variants. I think it is possible.
I believe that these will strain the rules of every writing system on the planet, in much the way that "ns1" strains English writing rules.
To clarify this, let look at the current definition of "orthotypography" in http://iucg.org/wiki/Multilingualization_Glossary orthotypography The term 'orthotypography' is seen from the viewpoint of readability. It means the correct use of typographic signs to convey the intended semantic or context. Also known as "Typographical syntax", orthotypography defines the meaning and rightful usage of typographic signs and cases. Orthotypographic rules may vary broadly from language to language, from country to country, etc. (source: IUCG, author: Jefsey Morfin, text listed to be reviewed and possibly changed). "ns1" is correct orthotypography in the network/human oriented universal semiotic system and the DNS resolution algorithm can apply to it.
I am incredulous at the suggestion that the Internet needs a "network/human oriented universal semiotic system," both because I don't believe it
The only way I can oppose faith, is to show it wrong. i.e. to implement a prototype solution. Under way.
and because I believe that humans (and especially human language) are much more resilient than that.
Absolutely yes (and this is true for each of the 22.500 language units, and more for the billions of idiolects and avalects) This is why Unicode, which is designed to support typographic workshops may not be adapted to support the algorithm we look for. For the time being my algorithm includes: - support of scripting metadata through ML-DNS since this was (probably rightfully) denied by IDNA2008. - a graphcode solution to oppose phishing. - particular cases agreed upon on a community basis, as TLDs and root-names (equivalent to TLDs outside of the Internet DNS - the ML-DNS being a multitechnology oriented project). jfc
participants (3)
-
Andrew Sullivan -
jefsey -
JFC Morfin