OSS projects have been able to gain a foothold in many server applications because of the wide utility of highly commoditized, simple protocols. By extending these protocols and developing new protocols, we can deny OSS projects entry into the market." (Microsoft, memo)
"The point that you seem to miss is that it is these simple, commoditized protocols and a culture of building freely on the work of others that brought us the explosion of innovation known as the Internet. And while the Internet has opened new areas of competition for Microsoft, it has also opened up enormous opportunities." (O'Reilly, response)
"The collaborative, massively distributed development process behind the Internet and Open Source projects is not your enemy. It is your friend, the source of basic research that you can turn into your next generation of products.
At bottom, the Open Source movement is an expression of the Western academic tradition, innovation and discovery through the free exchange of ideas. You rig that system at your peril. You have only to look at the stagnation of Soviet science and industry under a centralized autocratic system, versus the innovation that happened in our free markets, to see what fate you have in store for yourselves if you succeed." (O'Reilly, response)
\-----------------------
Here are some thoughts from a newbie. I haven't earned my red hammer and sickle beanie yet--so this is probably not propellor-ready.
Yes, I think the other posts are right about O'Reilly missing the point of the MS attackdog posture. Linux, and other unix branches are not friends and add nothing to the killer profits of the commercial world that MS dominates. But there is another point here too.
Re-read the MS quote, "...OSS projects have been able to gain a foothold in many server applications..." See? What's wrong here? There is no OSS foothold. OSS projects along with IBM, HP, and Sun absolutely dominate the larger server market. It is Microsoft NT that is atttempting to gain the foothold! The real question is can NT manage to gain any ground once it has to actually perform in larger commerical environments under heavy loads, high volumn, and overflowing bandwidth? From what I gather off the Unix news groups, the answer is, no.
So, NT will end up stuck in the closed, mostly off line small to mid-size business network market--a big and lucritative market, but not as big or as lucritive as large scale networks that actually run the internet and are the electronic backbone of multinational corporations, governments, and huge financial institutions. The only saving thought for Microsoft in that environment is the decentralization and re-distribution of those giants into smaller local envirnoments with isbn trunk connections to link up the smaller and distributed network pieces. It is within this exchange between local networks (running NT of course), that I think the MS memo is talking about when it mentions 'extending protocols'--i.e writing custom, non-standard, closed exchange routines, data checking and encrytion, security filters, etc. etc.
That development should be fun to watch.
There is another thing. MS follows Intel as Intel trys to wedge itself a bigger and bigger piece of the workstation level networks. But Microsoft then meets the idea of multiple processors running concurrently, 64-bit architecture and a bunch of stuff that is really over my head (I am already drowning here). I think this is where the whole concept behind Unix with its time-sharing modular design begins to really outstrip MS. And then there is the fact that all the protocols, and interfaces to and through the phone system were and are developed with the unix kernel design in mind--since that is in one form or another the primary OS used to manage and interact with the internet itself.
I have no idea what I am talking about here. But ISTM, that NT is not designed in anything like the same kind of time-share modules (the daemon idea run by crons ?).
On another thread, a question for Les Shafter--or anyone who knows. I just glanced at the I2O material on i/o device drivers and OSs. Does this spec require some change by Intel in its processor chip design, or can it be handled through a third party BIOS or is all the I2O spec directed to the OS itself?
Chuck Grimes
PS. I hope there are plenty of corrections on this post, since I am over my heard here.