From OKSALA@po3.bb.unisys.com Fri Jul 21 06:58:30 1995 Received: from mail.unigate1.unisys.com by icarus.lis.pitt.edu (4.1/1.34) id AA11806; Fri, 21 Jul 95 06:58:30 EDT Received: from x500dir.bb.unisys.com ([192.63.146.10]) by mail.unigate1.unisys.com (4.1/SMI-4.1-1.1) id AA14332; Fri, 21 Jul 95 10:53:32 GMT Received: from it.bb.unisys.com by x500dir.bb.unisys.com (4.1/SMI-4.1) id AA11094; Fri, 21 Jul 95 11:07:02 GMT Received: by it.bb.unisys.com with Microsoft Mail id <300FB1AC@it.bb.unisys.com>; Fri, 21 Jul 95 06:54:52 PDT From: "Oksala, Stephen P [BB]" To: "Rutkowski, Tony" , "Garcia, Linda" , "Loughry, Don" , "'O'Donnell, Jon'" , "Spring, Mike" Subject: FW: FW: Next round of Discussion Date: Fri, 21 Jul 95 06:53:00 PDT Message-Id: <300FB1AC@it.bb.unisys.com> Encoding: 276 TEXT X-Mailer: Microsoft Mail V3.0 X-IMAPbase: 1249813036 1 Status: RO X-Status: X-Keywords: X-UID: 1 Tony (et al): More thoughts. ---------- From: Tony Rutkowski To: Oksala, Stephen P [BB] Cc: Garcia, Linda; Loughry, Don; 'O'Donnell, Jon'; Spring, Mike; rjs Subject: Re: FW: Next round of Discussion Date: Tuesday, July 11, 1995 10:24AM > 4.3 What are the critical aspects of the standardization process? -How about the question of ossification of processes? In particular, the -continued maintenance of groups that engage in ritualistic process -to maintain their own existence. In some organizations, the preponderance -of groups engage in this kind of activity. By contrast, the IETF groups are -all adhoc by definition - existing over a definitive time period only to -produce a specific standard. I agree completely; this is a problem to some extent in all the groups that I am aware of. However I don't think that it is limited to working groups; there is just as much, and maybe more, "love of group" at the top level as there is at the WG level. Perhaps we should talk about X3 or the IETF itself having a limited lifetime unless there is some way to ensure that these act in a way that doesn't provide the opportunity for freezing culture. (Or are there cultural aspects that should be preserved across processes, and who decides whether they're good?) >4. Resolving the problem of the cost of the process, and who ought to pay >for it - the real problem behind the "free standards" arguments. -In virtually all organizations, the preponderance of the costs are those -contributed by the participants. The institutional costs are comparatively -minor, and can generally be amply covered by nominal participant fees. -In today's world, standards whose dissemination are controlled for -the purposes or garnering additional revenue sow the seeds of their -own demise. To quote Malamud, "secret specifications are not standards." You're right as far as total costs, but that doesn't help much when one is trying to figure out how to pay some specific cost, even if it is "minor". (My car payment is a lot less than the national debt, but that doesn't help me one bit.) Most standards organizations today support their staff based on revenue from the sale of standards. I have no particular bias toward that as a revenue mechanism, but there does need to be one. The problems with "nominal" participant fees are (a) nominal is in the eye of the beholder; (b) it may not provide enough money; and (c) it allows people and organizations to get a "free ride" at the expense of the participants. This latter could be compared at some level to a system where there was no patent protection for inventions. The issue here is that somebody has to pay, and my experience is that the little guys think the big guys or government should pay. I cna't speak for government, but the big guys aren't willing to foot the whole bill anymore. > 4.4 What is the most important characteristic of a standard? >1. A majority of the participants in the developing group voted for it. -irrelevant, unless it encompasses a broad cross-section of providers and -users. >2. Public comment was solicited and responded to in a fair way, and there >was no significant opposition. -irrelevant >3. The three largest companies in the industry agreed on it. -significant >4. More people buy products with it than buy products with competing >standards. -highly important The idea here was to define several alternatives, each of which is valid in its own context. I would argue that broad consensus is important; otherwise the "rest of the world" (implementors and customers) don't know what to build or buy until the dust has already settled - which in our industry may be never. It all boils down to what 'customers' perceive is the meaning of the "standard" designation - and therefore what they think they are getting. (See also below.) >X3, IEEE, or IETF. I think that the topic ought to be narrowed to documents >which come out of some process where there is multiple responsibility for >development and/or approval and subsequent maintenance, and where multiple >implementations are found. -Perhaps it would be better to say the "discussion" should be narrowed -rather than the topic. As you note, vendor standards accepted in the -marketplace are still very much standards - and often highly viable ones -that may be far better than those produced by committees. There is a real problem with the term "standard" since it is used to cover specifications, products, and processes that are wildly different. Consequently we can take opposite sides of a particular position and both be right based on which part of the "standards" process we are addressing. That's why I think it's useful to try and define the context for any discussion. More to say on this in response to your comments further down the line. >Having said that, I would opt for a definition of consensus that requires >the absence of significant sustained opposition AND a sufficient level of >support. This is a very general point, but it encompasses most of the >current standards processes. -Nice way of describing this point. > 4.5 What needs to be fixed most urgently? >d. Somebody gets the bright idea that government needs to fix this problem. >e. The US establishes a government-based replacement for ANSI, with the >power to endorse standards and develop US positions for formal international >agreements. > >A similar scenario can be painted at the international level, particularly >in countries where independent thought and action are not necessarily >encouraged. I am concerned at this point that we may be close to the point >of collapse (that is, to a world where there is no perceived standardization >process) - and if so, we need to think about what ought to replace it. -Your self-described paranoia fails to encompass a learning process -that has ensued over the past 20 years that has encouraged a scenario -of multiple standards process competing in an open marketplace. That -direction has been highly beneficial to just about everyone, and it's -difficult to conceive of a collapse back into the darkness of government -controlled processes. Are there any indications otherwise? I believe there are a number of people in the Federal Government who believe that the government needs to supervise the process. Some of them have tried to start up efforts to do that in both the IT and other areas. I don't know of any current strong pushes, but the sentiment does exist and can be bolstered by concerns over trade and other economic activities. I refer you to recent Congressional hearings on standards and conformity assessment. -It's not clear if anything "needs to be fixed most urgently." I would agree - until a disaster or perceived disaster happens, or some "outrage" that activates the Senators or Representatives. > 4.6 What practice is most damaging to standardization? >1. The value of a standard is that it is a standard; when there is more than >one, its value diminishes. -Just like central economic planning is in theory more valuable. It don't -say this pejoratively. It's just that we're trading off near-term versus -long-term effectivity here. I don't think the comparison is at all valid. Consider back some years ago to the days of proprietary architectures - when every company in the industry had its own processor architecture, I/O structure, and communications networking protocol. In each case there was a "standard" (we actually published them as internal technical standards at Burroughs) that had a following - known as a customer base. So this was a set of standards - but not the same in many meaningful ways from single industry standards. I have no objection to the competitive model of industry - as you note later, it has done great things for us. It just isn't "standards", and I don't think that you have to force everything to be part of the standards community. There are lots of mechanisms by which things get done, all of which may have their time and place. Where we may disagree is that you seem to be saying that the true cooperative model - the elimination of competitiion in design at certain levels - is inappropriate. Obviously I don't agree. In fact I would contend that if such competition arose inside the IETF, then the membership would be quick to deal with it in a way that would eliminate wasted resource and non-productive activities. At the very least, the IETF management structure would make sure that incompatible standards did not get the same approval. I should also not that I am not in favor of central planning - I thought that was clear in the previous note. The US people in the JTC 1 process have generally not been in favor of 'directed' activities, and this has been a bone of contention with the Europeans who would like to see strategy boards determining what gets worked on. Central planning is an entirely different mechanism from cooperative behavior, even if it is 'forcd' a little. >2. Standards groups are not (as some would like to think) independent >companies; they are mechanisms through which real companies come to >agreements with their competition and customers. -I would suggest that your model for what constitutes a good standards -process in today's IT world may be broken. You seem to ignore that -there are real people and cultures involved here - as well as highly -significant "technology transfer" process among a broad array of -participants. Perhaps the single most important characteristic of the -IETF or W3 Consortium are their ability to bring these diverse (some -would even say perverse) parties together in ways that result in copious -numbers of really creative new approaches, products, techniques, code -...and standards in very short order. First of all, this comment was intended to address precisely the issue that there are real people and cultures - and to point out that some of those people and their cultures are employed by standards developing organizations, and their views may conflict with the views of the volunteers who are supposed to be doing the work. This has been an issue in a number of instances over the last few years. I don't disagree, however, with your coments on what IETF or WWW have done - they are not unique in this regard, but they are clearly more inclined that way than the regular 'formal' bodies. One could take your comment and extend it to make the claim that an individual company's best strategy would be to let all this turmoil take place and utilize whatever comes out. I don't think that would be a sensible strategy for a large multi-national, but it would be interesting to pursue the ramifications. >3. The existence of competitive standards essentially puts the consumer into >the same practical position he would be in if there were no standards groups > - trying to guess whose products to buy based on their architecture, and >hoping that in a multiple unit environment that things will work together. -Considering that the existing semi-chaotic processes have produced one -of the most dynamic and successful marketplaces, as well as technology -evolutions in the history of humankind, it's not clear what you're trying to -fix. See above. I am not saying that everything has to be standardized (my definition), but that when standardization is the intent then the goal ought to be cooperative behavior rather than competitive behavior. There's a place for both. > - and with many companies belonging to more than one cartel! The whole >thing seems 'not sensible' in terms of rational economic behavior. (A >thesis topic for an economist?) -So you're doing to find an economist that favors central planning over -the marketplace? Not at all. My comment has nothing to do with any type of planning. Rather, it addresses (and a psychologist might be better than an economist) the fact that many companies spend large amounts of money to participate in organizations that compete with each other, and that participation is usually not done as a coordinated action with a single goal in mind. I don't consider this rational behavior by the company. > 4.7 Who should take responsibility for moving forward? >messy situation, particularly when some of those players believe that >'messy' is the way to go - at least for their organization. -As Forrest Gump would say "messy is as messy does." Messy works -spectacularly. And as one who lived through the dark ages of nice tidy -monopoly standards processes, I don't think the IT industry is ready for -the necessary collective pre-frontal lobotomy to return to that era. I agree that we can't and shouldn't go back again if for not other reason than the fact that there are several orders of magnitude more players, and the mechanisms that worked before don't work now. -Hope we can pursue this in good humour - and maybe get someone else -to join the discussion. You should get someone like MIT's Richard Solomon -to join the fray. He can tell you all about the monopoly standards processes -in the HDTV realm. :-) Amen to both the tone and the need for more voices. I was really hoping that Linda Garcia would interject some of her thoughts - based on that part of her work that I am familiar with, I might find myself in the middle of the intellectual spectrum rather than at one end. And now that I've finished this round, I can head for the Maryland shore for a week with a clear conscience. Best regards, Steve Oksala