Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uwm.edu!psuvax1!psuvm!BITNIC!ROBINSON From: ROBINSON@BITNIC.BITNET (Andrew T. Robinson) Newsgroups: bit.listserv.policy-l Subject: Specs vs. Standards Message-ID: Date: 5 Feb 90 23:28:01 GMT Sender: Discussion about BITNET policies Reply-To: Discussion about BITNET policies Lines: 29 Approved: NETNEWS@PSUVM Gateway In-Reply-To: Your subject -- Re: Standardized tools Michael, There is a distinction between standards and specifications: Specifications that are well designed can live through one or more standards. Standards are required to get real work done, but if you truly work from the top down your specification will be transportable between protocols. The reason it's so hard to come up with good specs is because most of us know too much. We base our ideas of how things should be done on the way we know they are done, which is limiting in and of itself. That is where the discipline part of this process comes in: People doing the specification have to divorce themselves from any preconceptions about how the work will be done, and concentrate on what work should be done in the first place. One of the best ways I've seen to accomplish this is to poll people who ARE ignorant of the innards of the various protocols and established tools and ask what they expect from the network. The days of a highly technical group of people dictating how software development should proceed are coming to an end. If BITNET is going to suceed we have to manage this network to the benefit of the end user who doesn't know or care about NJE, TCP/IP, RFC822, or x.400. Those same users will take their business elsewhere if we can't extract ourselves from the same mire. Yes, I'm taking an extreme viewpoint here. I am fully aware that tradeoffs are inevitable. However, as I said before it is important to approach future development efforts with the idea of making the perfect networking world, even if we know we can't acheive that goal (yet :-). Andy