fbaube@NOTE.NSF.GOV ("F.Baube") (11/12/88)
Steam-powered klunkers like C-64's and TRS-80's and Atari 800's got the revolution in data *manipulation* going for households. Can't some mass-marketer get the ball rolling in data *transpor- tation* by putting basic Fido in ROM ? Perhaps with a tiny, ergo inexpensive, non-volatile RAM to store parameters for node/BBS numbers and auto-logon scripts ? Maybe include a provision for negotiating protocols ? And for keeping a running total of con- nect charges, both to the phone service provider(s) and to the node/BBS ? How large (in K) would a basic software config be to manage and exchange email with Fido, MCI Mail, and anything else that's big enough to stir interest ? K.I.S.S. ! #include <disclaimer.h>
kent@WSL.DEC.COM (11/12/88)
Wait a second. Is FidoNet *really* what we want to build a revolution on? Yah, it's neat, and it's simple, and it might even be clean. But do we want to be stuck with it and whatever limiting assumptions it has for the next 10 years? Look at the uucp world. It started out as a quick hack for distributing software, then grew into mail distribution, then news was born, then came the anarchy we all know and hate. We're stuck now, for better or worse, with a stupid naming scheme, explicit routing, and a very limited (and limiting) functionality. Exchanging mail is great. But there's a lot more that I want to be able to do -- and expect to be able to do from my home communication node in the next 10 years. I worry about standardizing yesterday's solutions. Yes, it's better than standardizing tomorrow's (bad) ideas (cf the ISO protocols), especially before they're built. I believe in building something that works and then promulgating it. But let's promulgate something that allows us to expand. chris
doug@isishq.FIDONET.ORG (Doug Thompson) (11/13/88)
>From: fbaube@NOTE.NSF.GOV ("F.Baube") >Steam-powered klunkers like C-64's and TRS-80's and Atari 800's >got the revolution in data *manipulation* going for households. >Can't some mass-marketer get the ball rolling in data *transpor- >tation* by putting basic Fido in ROM ? Perhaps with a tiny, ergo >inexpensive, non-volatile RAM to store parameters for node/BBS >numbers and auto-logon scripts ? Maybe include a provision for >negotiating protocols ? And for keeping a running total of con- >nect charges, both to the phone service provider(s) and to the >node/BBS ? > >How large (in K) would a basic software config be to manage and >exchange email with Fido, MCI Mail, and anything else that's big >enough to stir interest ? You can probably get a fidonet mail interface, an editor, a conference processor, and a mail router all packed into one Megabyte. Then you have the nodelist :-). Basically, without a huge amount of storage, the current incarnations of home computers could not handle a fully connected network consisting of more than a few tens of thousands of nodes. And that, if they do any amount of conference mail, will keep the first 100Mb of hard disk quite busy. Not to mention the CPU cycles needed to sort and process that all and the high speed modems needed to move it economically. Admittedly that's the worst case. 10 Mb will support a modest system with a handfull of conferences. Nodelist information is not suitable for ROM distribution. The Fidonet nodelist is updated weekly, and has doubled in size every 12 months. A network is more than just some number of computers with compatible software and each other's phone numbers. That's a start, but message routing requires the coordination of a heirarchy of mail hubs. And that requires organization and some pretty substantial hardware at points. And Fidonet is a volunteer organization! What is needed to put a mailer in every household is a world-wide system of servers and hubs to do routing. Connection to the nearest server should be quite cheap, since the hardware and staffing costs are not really all that great -- although they are greater than volunteers can provide for the whole population! Not everyone is going to want to install a dedicated line for their computer, so they *can't* be in the nodelist, since they can't take machine calls, they can only place them. They need store-and-forward sites to take the calls for them and store the mail. To keep it economical, it's also desirable to batch all mail from, say, London to New York in a single call with expensive high-speed modems, instead of having every 1200 BAUD user sending every Trans-Atlantic message in a separate long-distance phone call. Fidonet does offer a versatile and efficient way for computers to do mail, and it is public domain. But I can see lots of room for improvement. It's basic message header structure doesn't have enough room for most internet addresses so our gateways are a series of kludges. If the header (and subsequent message manglers) were enlarged and a good, flexible world-wide address standard were adopted, and an international organization developed to create the infrastructure for simple connectivity, we'd have the necessary minimum. Existing fidonet software has been ported to many kinds of micro hardware (it is even possible to do netmail on a C-64). > >K.I.S.S. ! > Right. If we think about the post office, and just imagine an electronic version of it on the same scale, we've got a pretty simple and obvious way of doing it. Instead of the letter carrier dropping letters in your box, you have your computer call the neighbourhood postal station. That station in turn calls the city station every few minutes, etc. etc. Whenever you want to check your mail, you just trigger a poll to see what's come in since the last call. With cellular phone technology, the computer can be portable too. Businessmen are marketing this idea and hobbyists (Fidonet) are doing it free and showing the phenomenal *viability* of the idea. With proper routing and batching it only costs about 5 cents to get a 1000 byte message from North America to Australia. Costs of moving mail locally are almost infitesimal. And Fidonet does it all on micros! It's happening! I think we'll also see a situation of exponential growth continue, first in numbers of people connected, and later in variety and scope of information resources available over the wire. Right now it costs about $5 to send the average paperback across North America over the modem. It also costs about $5 to buy the average paperback. So if you are a write in New York, and you write the "average paperback", and distribute it electronically, it should be obvious you can offer your audience the product for less money and you can make more money per copy if you disseminate it electronically. Which introduces the problem of copyright. It's easy enough to keep track of someone printing your book in violation of your copyright, but it's a lot harder to keep friends and co-workers from passing around electronic copies that one of them bought. Oh yes, it's going to get very interesting as each and every one of us applies to be classed as a "public library", allowed to "lend" information to anyone for free while publishers holler bloody murder. Information is rapidly being re-defined, as is copyright. Basically it is as hard to effectively copyright computer information as it is to copyright the spoken word in a ballroom. If soemone utters a quotable quote, it will be repeated many times, regardless of how many copyright actions are launched in court. All these issues will have to be confronted by a genuinely large-scale public electronic data network. =Doug -- Doug Thompson - via FidoNet node 1:221/162 UUCP: ...!watmath!isishq!doug Internet: doug@isishq.FIDONET.ORG
dave@celerity.UUCP (David L. Smith) (11/15/88)
In article <649.237CDE8F@isishq.FIDONET.ORG> doug@isishq.FIDONET.ORG (Doug Thompson) writes: > >What is needed to put a mailer in every household is a world-wide system >of servers and hubs to do routing. Connection to the nearest server >should be quite cheap, since the hardware and staffing costs are not >really all that great -- although they are greater than volunteers can >provide for the whole population! > >Not everyone is going to want to install a dedicated line for their >computer, so they *can't* be in the nodelist, since they can't take >machine calls, they can only place them. They need store-and-forward >sites to take the calls for them and store the mail. To keep it >economical, it's also desirable to batch all mail from, say, London to >New York in a single call with expensive high-speed modems, instead of >having every 1200 BAUD user sending every Trans-Atlantic message in a >separate long-distance phone call. > I worked for a company that was planning to set up just such a scheme (it's defunct now, for any number of bad reasons.) We had a variety of devices from pocketsized terminals with acoustic couplers (in production now in the Netherlands!) to office mail-servers (projected) to Unix-based hub machines. One of the problems we came up with very early on was the sheer volume of communications required to reliably send, track and bill messages. We were counting on users sending spreadsheets, database files, etc. for the bulk of our business. 100K spreadsheet files can saturate a network very quickly. There's also the problems of addressing, keeping the databases up to date, having mail follow users with hand-held terminals skipping about on airplanes. All of these problems are surmountable, but it requires a very large investment of equipment and the willingness to take some large losses before the network is used extensively, if it ever is used extensively. ============ David L. Smith FPS Computing, San Diego ucsd!celerity!dave, ucsd!amos!sdeggo!dave
kent@WSL.DEC.COM (11/16/88)
I strongly urge everyone who is interested in issues of nationwide networks and dissemination of copyrighted information electronically to read Ted Nelson's writings, especially "Literary Machines". Whether or not you agree with his particular vision of hypertext, he has thought out what seem to be workable solutions to most, if not all, of these very hard problems. chris