system@condor.nmfecc.gov (10/25/90)
I want a deck now. I can't have one because no one has developed the intervening steps yet. We sit around in our offices reading books and postings and pretending that we are designing these things that do not exist. this is mostly a bunch of bs. If you are truly interested in doing these things, work on what is possible now. You can not jack into your brain, but you can make pretty pictures on a screen. What we need to discuss is what those pictures need to look like. By working ( not daydreaming ) we can begin to discover metaphors that do work in cyberspace. ( it is real, you are reading this message through it ). what we have now is akin to someone sitting on a skate board and holding a running gas engine with the flywheel rubbing on the ground to push the thing along. Wow! automotion !! In my smelly opinion, we can work on these things in several areas, one or two of which will need (perhaps unfortunatly) the consensual agreement of a significant group of participants which this interest group might provide. Before I go into the areas, i will present a quick view of one cyberspace possibility. I walk into my office and sit down at my workstation. jiggling the mouse, the screen lights up with a constellation of colored blobs. There are green blobs that I know I can log into, Blue ones that I can get interesting data from, Yellow ones that i get mail from but do not have other access to and red ones that get pretty hostile if I try to talk to them at all. I mouse over to a green blob in the corner and click. The blob zooms to fill my screen with a constellation like view of some local network. I can see hosts, some of which are green ( i have accounts on these ), and some others with just data that I can get - these are blue. Some of the objects appear brighter or bigger than others, and they perhaps have icon shapes that mean something to me about what type of host they are. if I click on one of the greenies, i am logged in and that host opens a window on my workstation. If i click on a blue one, a window opens with a directory of the files on that system that i have access to, and I can select files and they will be dragable to other windows, initiating data transfers. I hpe you get the idea. Now. what could make it happen? 1. a database of information about the hosts and networks 2. a tool running autonomously to fill in the database 3. a user interface that used the database to create the maps etc. 4. some standards for the inter-process communications 5. eventually a standard for multi-processing over distributed nodes Enough for now. If responses warrant, we can maintain this thread and work on the design of 1 and 3 on this group. Discrete implementation of 2 for various workstations is reasonable if there is a standard for the database. Jim Morton National Energy Research Supercomputer Center Lawrence Livermore National Laboratory JAM @ CCV.NERSC.GOV (415) 423-2374 These views are mine, and are not known or supported by my employer. If they knew I were doing this, they would probably shake their heads.
craig@com50.c2s.mn.org (Craig Wilson) (10/26/90)
In article <1990Oct25.070821.1@condor.nmfecc.gov> system@condor.nmfecc.gov write s: > [ ...description of a view of c-space...] > >Now. what could make it happen? > > 1. a database of information about the hosts and networks Sounds like Yello...er, I mean NIS. Except the user access rights part would have to be added. This could probably be implemented in a similar fashion. Or become an extension. Or maybe, that information would all be kept locally by the user along with preferences for display (color preferences and such). > 2. a tool running autonomously to fill in the database See man page for yppush. > 3. a user interface that used the database to create the maps etc. There probably is a tool out there already, but if not, it shouldn't be hard to write. > 4. some standards for the inter-process communications How many standards do you want? If you don't like the ones available today, just wait for tomorrow. > 5. eventually a standard for multi-processing over distributed nodes or distributed processing over multiple nodes? > > Enough for now. If responses warrant, we can maintain this thread and work >on the design of 1 and 3 on this group. Discrete implementation of 2 for >various workstations is reasonable if there is a standard for the database. > >Jim Morton sounds good to me. /craig
FC137501@YSUB.YSU.EDU (Paul M. Mullins) (10/29/90)
I generally disagree with the "pick it apart line by line" response form, but here goes... > From: system@condor.nmfecc.gov > I want a deck now. Me too! > If you are truly interested in doing these things, work on what is possible > now. You can not jack into your brain, System input can extend considerably beyond what is common today. Voice, gesture, & eye tracking are just a few additional modes. It is also possible to "jack" in to human physiology (including brains) using non-intrusive psychophysiology. On a layman's level (I'm a moderately informed layman in psychology) you can compare some of these techniques to polygraphy. Polygraphy has a bad reputation because of a "high" incidence of false positives. When administered correctly by an expert the incidence rate is about 15%. That's bad in a murder trial, but an improvement on my typing skills! > but you can make pretty pictures on a screen. We can do better than just pictures with output also. Certainly audio deserves equal attention. For virtual realities we can easily include aspects of tactile interfaces (ala flight simulators). Can the other senses be that difficult? > What we need to discuss is what those pictures need to look like. A standard? I want my own interpretation, or at least something that _I_ find reasonable. > By working ( not daydreaming ) we can begin to discover metaphors > that do work in cyberspace. I think the meaning is "lets role up our sleeves," but I disagree with not daydreaming. I find that when I get caught up in details, the net.dreamers provide a certain amount of inspiration, after a lot of filtering based on my own predispositions of course. I would note that many people think the correct first step in finding the "best" solution to a problem is brainstorming. Chaff is winnowed out later. Besides, there is precedent: daVinci dreamed of flying and Vannevar Bush described a hypertext system in 1945. > In my smelly opinion, we can work on these things in several areas, one or > two of which will need (perhaps unfortunatly) the consensual agreement of a > significant group of participants which this interest group might provide. Or a big development grant. Any takers... I mean givers? > Before I go into the areas, i will present a quick view of one cyberspace > possibility. >.... > I hpe you get the idea. I know it's just a possibility, but it's one I disagree with. Fundamentally, it seems too limited. More specifically, 'blue' things that get mad if I contact them shouldn't be displayed at all. I don't like your choice of colors, etc. But what if I am trying to break into systems. Maybe all I want to see are the blue things. Different views will be necessary. It is highly likely that personalized interfaces will become common. In fact, adaptive intelligent ineterfaces may personalize themselves on your behalf. > > Now. what could make it happen? > > 1. a database of information about the hosts and networks > 2. a tool running autonomously to fill in the database > 3. a user interface that used the database to create the maps etc. > 4. some standards for the inter-process communications > 5. eventually a standard for multi-processing over distributed nodes > > Enough for now. If responses warrant, we can maintain this thread and work > on the design of 1 and 3 on this group. Discrete implementation of 2 for > various workstations is reasonable if there is a standard for the database. I agree that we have to start somewhere. The following is my variation on Jim's suggestion: These suggestions could _allow_ it to happen: 1. Each entity maintains its own knowledge base about itself. It is a knowledge base because it is dynamic. It is continuously updated by the "target" entity itself or some (set of) trusted entities. 1a. Entities include every application of significance. For example, an operating system, text editor, database, network manager, and a calucator tool are all entities. 2. The knowledge bases are updated by knowledgable entities. They contain only the necessary amount of detail. Detail may be temporarily or permanently increased. A network manager would have information about all network entities, for example, but may have to request additional information from a database (some entity) to determine access priviledges. 3. A user interface capable of utilizing the information available is a must. There are infinite possibilities for this interface. 4. A communications protocol to be used by all (cooperating) entities. Note that an entity may ignore information requests all together or determine an appropriate response based on what is known about the requestor. 5. Multiprocessing is no different at the level I am considering. Distributed processing would occur at a lower level also, and hence have its own protocols. The user might need access to this level, however. Standards are implicit in this kind of project, but they can be local standards for now (it was good enough for ethernet). If you would like more details on my view (including a description of a prototype system), get hold of "The Network User Interface Substrate (NUIS): A Task-Oriented Reference Model," Paul M. Mullins. PhD Dissertation, Department of Comp Sci, Univ of Pittsburgh, August 1990. Available through UMI soon. or wait for the pieces to be published separately - I'm working on this now. > Jim Morton > National Energy Research Supercomputer Center > Lawrence Livermore National Laboratory Copyright (c) 1990 Paul M. Mullins FC137501@YSUB.YSU.EDU or mullins@macs.ysu.edu