Earlier this month, I published a couple rants about Microsoft's real plans for .NET, and the seeming inability of Microsoft competitors to, well, compete. My thoughts got a lot of attention within the computer industry, and especially spurred some internal discourse at Sun Microsystems all the way up to CEO Scott McNealy. But while Sun was trying to decide whether I was insightful, a pain in the butt, or an insightful pain in the butt, some readers far smarter than I am pushed me to thinking even deeper into the .NET mess. Now I see some quite specific roadblocks to Microsoft's progress. These are issues that I doubt Microsoft has even considered internally, preferring as usual to fail first, then come up with a fix. And the way they'll fix it is likely to be pure Microsoft again.

.NET, as you'll recall, is Microsoft's emerging suite of applications and services that are intended to build Windows into all levels of the Enterprise, not just the desktop. .NET leverages the Internet using XML to modularize software and create for Microsoft whole new sources of recurring revenue so Steve Ballmer can buy a new car. .NET is an architecture, a language, a suite of applications — it's a big project that is likely to succeed primarily because of the Windows monopoly. Attrition alone will build .NET components into nearly every desktop in the world within three to five years. This is a powerful advantage over Microsoft competitors.

That's why .NET should succeed, but why it might fail is because of the nature of big business. .NET as a services platform has the immodest goal of changing the way the world does business, but there is some real question how much of a change that really will be. Businesses are running right now without .NET, but are they faltering as a result? No. In fact, the very functions that Microsoft intends to offer in .NET are handily accomplished already using other software — maybe not in small to mid-sized businesses, but definitely in most large companies.

What incentive do those large companies have to switch to .NET?

.NET won't be cheap, with corporations paying about $350 per seat per year. Most big companies have spent the last five years installing SAP, Oracle, JDE, PeopleSoft, and Seibel at a cost of tens of billions, not to mention all the aggravation and stigmata associated with those installations. These companies are very unlikely to throw out their existing solutions just to buy .NET services. They simply can't afford it.

But that's what Microsoft is asking, for IT departments to throw out their past five years of work then pay good money just to get on the .NET treadmill. They would be paying Microsoft to build five years from now what these companies already have today (never mind what they will really need in five years). This worked well for Microsoft during the PC era, but what will be the business benefit of doing this for enterprise applications? I can't see one.

There are two likely outcomes here. One is that .NET will be successful with small and medium-sized businesses that aren't already using these other solutions. In other words, Microsoft will transform its business with traditional customers, but not with large businesses and government, where the big computing bucks are stored. The second likely outcome is that the Global 2000 are going to tell Microsoft, "Make this stuff work with the old stuff, and better not make us pay for the privilege, either — we've already paid you for the client side."

Just about everyone on the business side has figured out the game now, and they won't sign up for another round just because Microsoft thinks they should. The days of blindly upgrading the OS and office products are over. It is too expensive and there are few real benefits.

What is Microsoft to do, then, to make .NET the top-to-bottom success they hope it will be? In order for Gates to make this work, he will have to get into the same business as IBM, HP, Sun, and Compaq (and Hitachi, Fujitsu, Siemens, NEC, EMC, CSC — a whole long list) — providing integrated services for businesses, not retail PC users. Microsoft says they are already in this business, but they really aren't — not to the extent required to be really successful.

To imagine Microsoft competing at this point with, say, SAP (the giant German business software company) is laughable. To sell that class of software, they'll need a very serious amount of code and functionality. Many, many things have to be designed into the product (security, audits, business controls, etc. — stuff Microsoft does not value). The business process changes for customers are as big, if not bigger than the cost of a tool like SAP. When you ask a customer to make this level of investment, there must be a serious economic justification. Microsoft is not providing such a justification to buyers because they haven't had to do so before.

I'd bet Microsoft has not even thought of this. As far as I know, Microsoft executives have zero experience selling and implementing an SAP project.

At this point, if you are a bigshot at Sun or some other Microsoft competitor, maybe you are breathing a sigh of relief or patting yourself on the back. Not so fast. Just because .NET is cruising for a bruising doesn't mean that life is going to be easier for Sun competitors. My observations from earlier this month are still valid: Too many companies are complaining and too few are actually competing with Redmond.

It will be painful — especially for customers — but Microsoft will figure this stuff out. They will make the changes required for .NET to succeed, which is to say they will transform the company. HOW they will transform Microsoft is the part I find most interesting.

Several years ago, Microsoft made a deal to buy Intuit, home of Quicken and Turbotax. That deal eventually fell through for a variety of reasons, but I have my doubts that it was ever real. The primary strategy for Microsoft going afterIntuit was to get their retail business expertise. During their due-diligence visits, the Microsoft team figured out who at Intuit were the brains Microsoft wanted, and they recruited them. Those people are working today in important positions at Microsoft. Publicly the deal fell through, but Microsoft got what they wanted, saving $1.5 billion in the process. Those recruits weren't kismet, they were the point of the whole effort.

So when .NET hits a wall, expect Microsoft to pull out the checkbook and go shopping for a SAP. Just don't expect the deal to actually go through. It won't have to for Microsoft to achieve its goal.

Turning to old business, the readers have spoken and last week they definitely didn't like Bill Gates' claim to have deliberately created the platform that enabled Open Source software. The most compelling arguments against Gates, in fact, came from readers who pointed out that Richard Stallman, the true inventor of Open Source, didn't even write his software for personal computers. The point is not hardware standardization but software portability. I missed this completely.

Another highlight from last week was not just the appearance of IBM veteran Jack Sams, but the subsequent coming to light of Dave Bradley, who wrote the original IBM PC BIOS. "Jack has it pretty much correct," wrote Bradley after reading the column. "The Datamaster was the source of almost all of the PC design decisions — one of the primary ones being that IBM should buy rather than make the operating system and BASIC interpreter as we had done on Datamaster. We had begun a port of Datamaster I/O code to an 8086 design during the summer of 1980."

"Small corrections, since Jack was no longer involved after 11/80: I did the first version of ROM BIOS for delivery of the first prototype to Microsoft on Dec 1, 1980 (Ballmer answered the door when we showed up), but it was extended and revised through April of 1981. We published more than the interfaces for BIOS — we published the entire assembly code listing (it took me a week to edit it into a single file for publication) and the entire schematics for the hardware in the Technical Reference Manual. I've always thought we did that as a way to get done quickly, rather than trying to write an architectural specification of the interfaces."

Thanks to Dave Bradley's thoroughness, then, Compaq, Dell, and the other IBM PC cloners today have a $400 billion global business.