This is my major objection. Look, ethernet is a spec about the *physical
layer* of the communication. Quoting from Les:
> CSMA/CD refers to the extension 'invented' by Metcalfe et al to detect
> when other machines are broadcasting on a shared line, whereas the
> original Aloha system did not require a machine to listen for the
> presence of others talking, but only to listen for itself and whether
> its transmission was heard by the network.
You're right, Metcalfe's genius lay in the collision-detection of
non-deterministic transmissions (as opposed to Token-ring, e.g.); but this
has nothing to do with the alleged "closed" or "open" nature of MS's OS. Any
communication layer is on top of the physical layer (SPX/IPX, NetBEUI,
NetBios, or TCP/IP). And these are all *fairly* open specs (exc. TCP/IP,
which is totally open--check out some of the early disputes between MS and
Novell over SPX/IPX). A fair comparison would be, rather, the IBM PC.
Roughly, software (network protocols) / hardware (ethernet) == software (MS
OS, other OS) / hardware (PC).
> By the way, Microsoft and other select companies are campaigning for a
> new I/O bus system called I2O. The problem is that they want to
> establish a consortium to which you must be a paying member before you
> can learn of the specification for its operation. This would make
> development of open OS's like Linux more difficult if board
> manufacturers were to sign on to th eproposed I2O consortium, since
> Internet wide hackers would no longer have the free and unhindered
> access to the bus protocols neccessary to easily write device drivers
> for the operating system.
This is a non-sequitur. What about USB, or plug-and-play? You have a linux driver for a USB joystick? I2O is almost exclusively a server-oriented function, designed to match some of the features available in unix servers or mainframes. Server hardware = server OS = big boys. You pay to play. Linux may be great for a web server, or router, but no one (absolutely no one) is going to put it on their $300,000 database server. But if they do, they will *buy* Linux from a company that can afford to pay to license I20.
> in spite of leaving the specification open and public, Ethernet
> thrived and anybody with know-how can write code for it in a timely
> and efficient manner. This is in constrast to the games Microsoft and
I don't know what this means. Who writes code for ethernet? Maybe some protocol-jockey, or device-driver guru. Everybody I know writes (a) to a protocol-layer (NetBIOS was popular for a while) or (b) sockets (Winsock on the Win platform).
Sorry about the compression and tone. I don't really have a harsh opinion
about Metcalfe -- some of his arguments against the local bells are great.
As Les said:
> Finally, I do agree with you that more often than not, mainstream
> coverage of technology and science is pitiful.... i didnt share your
> feeling about the current article under discussion, however.
I totally agree with the former. The reason for my original question,
however, was not all the same old "MS is big 'un evil one. Break out big
gun." Rather, it seems to me that this is an important issue in business
coverage, one that it's *not* so easy to divide along typically right-left
lines, and that the case, as I've followed, has not been well-documented in
the press (both online and print). Cheers,
John
John St. Clair University of South Florida Department of Philosophy Cooper 107 Tampa, FL 33647
Office: CPR 267 Phone: 813-974-5896 Hours: M 3-5, T 10-12 http://chuma.cas.usf.edu/~jstclair/