Netizens-Digest Sunday, April 29 2001 Volume 01 : Number 388 Netizens Association Discussion List Digest In this issue: Re: [netz] Government and Science Was:FC: Ftc action Re: [netz] Government and Science Was:FC: Ftc action - part 3 Re: [netz] Government and Science Was:FC: Ftc action (part 2) ---------------------------------------------------------------------- Date: Sun, 29 Apr 2001 13:05:07 -0400 From: "Howard C. Berkowitz" Subject: Re: [netz] Government and Science Was:FC: Ftc action >"Howard C. Berkowitz" wrote: > >>In much of this discussion, I feel that Ronda and I may be talking at >>different levels. I split the Internet into layers, with the lower >>layers (my area of specialization) dealing with the internal movement >>of packets, not the user visible applications where directory >>services come into play. > >>At the lower layers, I simply don't see the corporate demons that she >>seems to suggest. I definitely do worry at the more user-visible >>parts. > >Well you may find it of interest to take a look at the May 2001 issue >of Wired. I don't often read it but this had an article by Larry Roberts >(one of the ARPANET pioneers) and he explains how he is creating >a new kind of intelligent router significantly change the routing done >on the Net so that there will be priority service for those who pay >more and lower class service for those who can't afford the higher >prices. Unfortunately, you are dealing with popularizations of concepts that have been discussed extensively, for years, in quite public engineering forums. See, for example, the IETF Differentiated Services Working Group at http://www.ietf.org/html.charters/diffserv-charter.html What's wrong with having economically differentiated premium service, as long as the lower class services are affordable? The "tragedy of the commons" is a classical economic paradigm -- unless there are consequences to using shared or premium resources, some people will take more than their unfair share. For health and productivity reasons, I fly business/first class on long flights. The availability of this premium service doesn't seem to empty out the economy class cabin. > >The changes that are being proposed would seem to affect the lower levels >as well as the upper levels of the Internet. > >And what is interesting is that these changes are not being discussed >and debated publicly. Instead there is an effort to install them by >fiat on the folks in the US and around the world. But these are widely discussed in public engineering forums such as the IETF and operational forums such as IEPG, NANOG, RIPE, etc. Where do you think they should be discussed? Incidentally, the economic impact of differentiated services also have been discussed for years. Geoff Huston is one of the leading experts there; see many of his papers at http://www.isc.org/iepg/ as well as his book, ISP Survival Guide. > > > >Well depending on who does the funding, that will help set the priorities >of what the research is that gets done. > >For example I have been looking at some of the technical articles written >during the early development of the Internet when the research was >funded by government. Then the aims of the research were to create >a "resources sharing network" and "fairness" of treatment of all packets >was an objective. Totally equal treatment for all packets has, with experience, turned out to be a very bad engineering idea. The most basic concern is that if all packets are equal, the control and management packets that actually run the routers and other infrastructure components may be blocked by user traffic. If the control traffic is blocked, the Internet as a whole cannot respond to changing conditions, and can and has suffer massive failures. There are reasonable differentiations among basic kinds of application traffic. Is it reasonable to treat an interactive session (e.g., web browsing, interactive games) with the same service quality of "freight" such as an overnight file backup? Internet voice simply will not work if it encounters significant delay. It is incidentally, surprisingly tolerant to loss and errors, but not delay. Prioritizing voice traffic, which actually doesn't take much bandwidth, is a reasonable engineering compromise. Military networks always have prioritized. While it is more folklore than reality that the ARPANET was designed for nuclear survivability, there are basic features built into IP to meet military priorities. It's also interesting to notice that while there is a very high priority level for Emergency Command Precedence traffic (i.e., the level of nuclear weapon launch orders), network control and internetwork control have even higher precedence. > >Now the article in Wired reports that the investment community funding >the research on charging more for service is eager to fund research that >will raise the cost for all to send packets. I haven't read the article, but I would disagree with the premise. Historically, the costs of mass electronic communications, as a function of personal income, have dropped dramatically. That doesn't preclude the existence of premium services. > >Its interesting also that Wired hypes what Roberts is doing, rather >than offering any critical or social perspective of it. Wired is a profit-making magazine. Are they required to give perspective? Who determines how much perspective is sufficient? > >The social goals mean extending the resource sharing capability >of the Internet and making very low cost access available to all. I have to disagree. Not all users of all resources can or should be equal. One of the major social goods possible with IP networking is improved delivery of healthcare, including such things as telepresence surgery. Are you saying that the surgical blade can be delayed by the traffic of some kids playing Doom? Or are you saying that there must be so much network capacity that no matter what traffic is put onto the network, it will never be delayed? > >The aim being promoted by Wired is end the Internet, create a new >network that will cost everyone more and will provide those who >can pay access to the best service, and everyone else will be second >or third class netizens. So? As long as there are seats in coach, does that mean it's evil for me to fly in first? Is it wrong for me to send one package by Federal Express for 8:30 next day delivery, and send a routine bill by postal mail? > > > >>Ronda, I consider myself a legitimate researcher in the scaling of >>the Internet. I still have to make sure my feline research associate, >>Clifford, gets cat food. > >Are you saying that you can't get a job where the funding is public >and so researchers like you have a problem? The reality is I can't get a job with public funding which is remotely comparable to my private sector compensation. Years ago, I worked in government. The salaries weren't close even twenty years ago. > > >>>Without the scientific process to try to determine what is needed >>>for the scaling, it doesn't matter how much money is poured in. >>> >>>It will be wasted. >> >>>One the scientific research is done, then there should be a similar >>>scientific approach to determining what form the infrastructure's >>>development should take. > >>We may have some different definitions here of scientific versus >>engineering paradigms. Oh -- and there is definitely such a thing as >>engineering research. > >There is engineering research that is what I am referring to as >scientific research. I was just reading such a paper yesterday. > >The paper is what I am referring to as a scientific paper and what >you are probably referring to as an engineering paper. It is part >of a wonderful volume about Internet research "Proceedings of the IEEE, >vol 66, No. 11, November 1978) The paper is "Modeling and Measurement >Techniques in Packet Communication Networks" > >It describes the process of designing networks and designing ways >to test the protocols. > >This is what I am referring to as science :-) Then you'd probably be much happier calling the IETF the Internet Science Task Force. True, it does have a small research arm, and there are more theoretical forums such as SIGCOMM. But the same people typically participate in both. > >>>The paradigm of the Internet was to have a way to interconnect >>>dissimilar networks. It seems that that has gotten changed to > >>having a backbone that some company(s) create. Again, I'm confused. There are lots of competing backbones. When I design a large network, I have a choice of backbones. That's good both for price and for avoiding a single point of failure. > >>Unless it is a government funded utility, what is the alternative? >>And how are international backbones funded? > >This is a problem to be explored. I think in Austria the backbone >was built by the government to connect the universities and then >the public schools, and the private networks were connected to >it and offered service to the private sector. So there was a mixed >infrastructure. > >In the Netherlands there was a debate whether funding should go >into extending the train system so it crossed the whole country >or building a backbone for the Internet. Eventually I thought they >did something like build the national train system and use system somehow >to build the backbone for the Internet infrastructure on or connected >to it rather than as something separate. The two are more related than you might think. It's very convenient to run optical fiber over the sorts of rights of way that trains, electrical power, etc., utilities need. SPRINT historically is an acronym for, as I remember, Southern Pacific Railroad Internal Network Telecommunications, and was spun off from the railroad. > >By having a public discussion of the issues and the different points >of view new alternatives become possible, and ones that will be >more in providing for the needed public purpose.\ Again: in what forum? > > >>>The original Internet architecture was designed so that it could >>>interconnect dissimilar networks under dissimilar forms of administrative >>>or political control. > >>I'm puzzled why you don't seem to think this remains the case. > >There didn't seem to be any effort with the creation of ICANN >to recognize the need to continue to support the diverse >administrative and political units and networks. I'm the last person to argue that ICANN was a good solution. Nevertheless, there were political and time pressures to do SOMETHING at the time. ICANN does have significant influence on the DNS, but its influence on addressing and routing is more theoretical than real. ------------------------------ Date: Sun, 29 Apr 2001 14:10:55 -0400 From: "Howard C. Berkowitz" Subject: Re: [netz] Government and Science Was:FC: Ftc action - part 3 >"Howard C. Berkowitz" wrote: >> >>>There is a need to learn from this development and build on it, >>>as the experience from IPTO shows that such a institutional form >>>is needed to support the Internet's continued development and scaling. > >>I think you need to quantify this. The rate of growth of the Internet >>has been increasing in a non-linear way since the demise of IPTO. >>Admittedly, certain mechanisms, never designed for this load, are >>faltering. > >But the seeds were nourished by IPTO. > >Now that one doesn't have IPTO it is much harder to have the long >term research that is needed for scaling the Internet in a way >that is in the public interest. How do you see creating such a guiding research function? Historically, the Congress hasn't been very interested, and dissolved its own technology office. There certainly have been focused technical organizations in the executive branch, both open (e.g., the Apollo program) and classified. But these did not have huge amounts of public input. > > >>>To me, netizens have to operate in such a politicized environment. >> >>>Yes that is true. The politics of science has to be taken on. >> >>>But that means recognizing that there are "vested interests" and in >>>the past there have been ways of inhibiting the damage they can do :-) > >>But there's also the issue of not demonizing them, and understanding >>what is and is not broken. I freely admit that the directory and DNS >>situation is in terrible shape. > >I agree it isn't to demonize them. > >And I agree that we want to understand what is and isn't broken. > >But after studying the incredible capability and achievements >of the researchers who created IPTO and the IPT community and >the important computer science and networking developments of our >time under its protection and leadership, it is hard to see the >kind of pressure that companies like MCI/Worldcom and others >exert on folks in the US Congress and other government officials. > >I went to a congressional hearing and saw that the congressman >from Mississippi was there to take care of MCI/Worldcom not >the public. I can believe that. > >And a staffer said that the Congress folk only hear from industry >and in such technical matters he claimed they had no ability to >evaluate what they were being told to do by the industry folks >pressuring them. There are ways to lobby. But I don't see many "netizens" organizing to do so. In general, by the time it gets to the hearing level, the positions have already been communicated, especially to staff. > >Also I went to a town meeting at Columbia about the Internet and >its future. The people holding it, it turned out, were CEO's of >companies. They claimed they were there to hear what people had >to say so they could tell the new President. Personally, I regard town meetings as purely public relations activitites. Just as an individual, I have gotten things I wanted from corporations and government organizations, but I did it through reasoned correspondence and focused meetings. By "reasoned", I phrased my goals in a manner such they were net positives for both sides. > >But mainly what people had to say was that they didn't like the >commercial activity and taking over of the Internet. Clearly >the CEO's at the meeting aren't going to tell the new President >that. I understand that "people" don't like commercialization. "People" also are demanding cuts in government spending. So where is the funding for new Internet capacity going to come from? > >So I agree that one doesn't want to demonize anyone. > >But the corporate folk that I have been around until recently >didn't think that there was anything else in the world but the >needs of their corporations. While I don't necessarily like it, the legal climate is that activity by companies that don't increase shareholder value leave corporate management very open to shareholder (or, in far too many cases, instigating class action attorneys) lawsuits. Corporations do have fiduciary obligations. Regulation is one way to divert corporate resources into things that don't yield immediate shareholder value, and may very well contribute to greater goods. But deregulation is Republican dogma. > > >However, with IPTO there was a general purpose objective. > >With Microsoft, there is a narrow self serving objective. > >And it seems that the general plans for scaling the Internet are >are being subjected to public discussion and consideration so >that the research can be done in a general way, rather than >to increase some sector's profits. > >That is why I feel we need a research institution like IPTO again. > > >>>Well I didn't think that the regulated AT&T that developed the >>>world class telephone infrastructure in the US was any >>>"socialist system" though I am sure that MCI/worldcom might say >>>it was as that was their effort to end the deregulation and the >>>benefit that MCI got as a result. > >>To quote back, "regulated". Modified socialism. The 1913 Kingsbury >>Compromise gave AT&T effective control of long-distance >>communications, which provided the revenue stream for building their >>backbone and funding Bell Labs. > >Somehow to equate good government regulation with "socialism" >even "modified socialism" seems to be a lack of recognition of >the importance that regulation plays to make science possible. I'm not using "modified socialism" as a derogatory term. Once government takes a role in private industry, controlling rates of return, it's no longer a classic free market. Once arbitrageurs take a role in manipulating short-term financial instruments without reference to real value, it's also no longer a classic free market. > >The US has excelled in scientific activity when it has protected >and supported scientific research. This took good regulations. > >Bill Gates fights regulation tooth and nail and thus the world is >stuck with a windows operating system that perpetually crashes. > >Research is needed not only in the technology but also in the social >forms and institutions needed to nourish and provide the soil for >the technological development. And again: what vehicle can implement this? Who needs to take the initiative? Who pays? Who takes the legislative lead? ------------------------------ Date: Sun, 29 Apr 2001 18:11:58 -0400 From: "Howard C. Berkowitz" Subject: Re: [netz] Government and Science Was:FC: Ftc action (part 2) >"Howard C. Berkowitz" wrote: > >>I agree there is a definite concern in open access to broadband >>access networks (e.g., IP over cable, DSL) and third generation >>wireless access. But access networks and backbone networks are >>different things. They perhaps may need different models. > >One of the people I spoke with at the National Academy last week >told me that they are doing a broadband study there and >they have decided that it will cost $1000 to connect each household >in the US. > >The study is being done by a small committee in closed sessions >and with no public input. > >Yet it is a public question. > >In the article on this in Wired this month it seems there is basically >a plan to end the Internet and substitute this all purpose network >that will be to put tv, telephone and radio and computer data into >the same network. Again, Ronda, you seem to be using "Internet" as a model of nonprofit collaboration. It's also stretching things a bit to call everything "the same network." There are important distinctions between local access providers, IP service providers to end users, backbone providers, and many sorts of content providers. Right now, most telephone and a large part of data go through copper pairs run by the local telephone company (ILEC). Once the copper pairs enter the ILEC local office, they very well may split data to one provider and voice to another provider (I'm avoiding the term "network"). Copper pairs have physical limitations for many new technologies. Conventional cable TV systems have limitations as well, especially if high definition television comes into widespread use. In general, the technological imperative is to run optical fiber to subscribers, or at least to "curb" locations near them. Multiple services can coexist on these fibers. Admittedly, there is a great deal of controversy over "open access" to fiber and cable systems, and even to rooftops for wireless antennas. But there is a practical aspect to local access -- how many times do you want to dig up the same street to run fiber? Distruption of streets by uncoordinated excavations of competitors is a major problem in downtown Washington DC. Some cities are running "dark fiber" when they put in infrastructure such as sewers, and leasing the fiber, but that depends on there being new construction. I certainly agree that property developers need to put in cable duct and possibly dark fiber, right along with water, sewer, and roads. > >But radio and tv in the US and telephone are different kinds of entities >under different forms of public oversight. Radio and TV use a free resource -- the frequency spectrum. > >To put them all into one, and to subordinate the online data activity >fo computer folk to the princes of owning content is introducting >a serious problem and basically seriously jeopardizing the continued >existence of what has been the goal and dream for the Internet >since Licklider's early writings. > > >>>To the contrary it seemed that WWII demonstrated the need for >>>governments to support scientific research. And so after the war >>>there was the recognition that this was now an important >>>need. For example Vannevar Bush and the important report he >>>and others at the National Academy of Science did "Science: the >>>Endless Frontier" > >>WWII rather than Cold War--you are right. I would point out, however, >>that massive investment in basic research was more post World War II. >>There's no accident, for example, that the discipline of operations >>research is named what it is -- it's the use of quantitative methods >>to improve military operations. Many of the early OR problems dealt >>with antisubmarine warfare. Norbert Weiner's cybernetic research was >>given a push because it was useful in antiaircraft fire control. The >>first primitive computers generated artillery ballistic tables >>(Harvard Mark I), broke enemy cryptosystems (bombe/Colussus), or did >>hydrodynamic calculations for atomic bomb design (IBM). > >But Licklider's background was in brain research, in how complex >systems, both natural and artifical function. And your point is? Norbert Weiner is generally considered the founding father of modern control theory. For that matter, he did also work in neuroscience. > >In studying servo mechanisms whether in the human brain or in the >technology that was being developed. > > > >I was referring to the Soviet people and their desire not to have >another devastating war and their pressure on the government to support >science in hopes of preventing such another war. What Soviet people's pressure on their government, before Gorbachev and perestroika? People who were not in the Party, or at least apparatchniks in the nomenklatura, who pressured the government would get personal government attention. From the Second Chief Directorate of the KGB. Indeed, they might get up close and personal with Soviet psychiatric science--in which dissent was considered an illness. > >>>So it is not only the cold war, but the effort to prevent another >>>world war that has motivated the events that have led to the >>>public support for science that helped to make possible ARPA (1957?)and >>>the Information Processing Techniques Office that Licklider started >>>at ARPA in 1962. > >>ARPA. Later DARPA. Always in the Department of Defense. To say that >>this agency's motivation was public support of science, and not the >>recognition that advanced research supports military development, is >>ridiculous. > >After WWII there was the realization among a lot of people that the US >had won the war because of science. > >You should perhaps look at my papers on this starting with >http://www.columbia.edu/~rh120/other/arpa_ipto.txt > >Licklider was brought to ARPA in 1962 to start an office of information >processing techniques. His goal was to catalyze the development of >an information processing science. > >Look at what happened when Licklider was brought back to IPTO in 1974. > >Then the pressure from industry on the US Congress had had its affect >and Licklider could not go on and develop the scientific work he >wanted to develop. > >Putting scientific research under the pressure of applications >meant that the research was harmed. This happened in the Air Force >Office of Scientific Research (AFOSR) in the Office of Naval Research >(ONR) and at ARPA when it was changed to DARPA. Every one of the organizations you cite routinely supported basic scientific research. Obviously, when there was an applied benefit, they exploited it. Back in the late 60s, I assisted in a study for ONR, tracing the applications of basic research. It's been a long time ago, but one thing that comes to mind is that ONR sponsored basic research in chemistry, including the feasibility of nitropolymers. Pure physical chemistry. Subsequently, however, some of the new compounds discovered became the basis for rocket fuels that made submarine-launched ballistic missiles practical. There was lots of support for pure mathematics, which turned out to have all sorts of applications unforeseen at the time. Somewhere on my shelves, I have a symposium from the early sixties called "Computer Augmentation of Human Reasoning," which was very very early blue-sky work. It was sometimes hard to tell what was basic and what was applied. One very controversial study was called "Witchcraft, Sorcery, Magic and Other Psychological Phenomena in the Republic of the Congo." Done at the Army Special Operations Research Office at American University, and roundly criticized in Congress as silly science. Turned out that it was for a very specific military purpose, and done on a crash basis. US forces were going to intervene in the Congo rebellion -- the ground troops involved were Belgian paratroopers, but flown in on US aircraft. There was a very real chance US ground troops might become involved, and the Simba rebels were operating under a quasi-religious belief that their leaders made them immune to bullets. The psychological study was intended to find a nonlethal way to convince the Simbas they had no such immunity. > ------------------------------ End of Netizens-Digest V1 #388 ******************************