Bill Gates once told me that the way to make money in the personal computer business was by setting de facto standards. This week, two very powerful companies each announced proprietary standards for Internet audio distribution. One of those companies still has a good chance of succeeding. The other has already failed.

The players, of course, are IBM and Microsoft. The possible winner is Microsoft. IBM is the one that has not only failed, but knows it has failed.

First, the opportunity. Driven by a combination of greed and fear, the music recording industry is rushing toward choosing a standard for the safe and sane distribution of copyrighted music over the World Wide Web. The greed part is simple: By eliminating the cost of manufacturing, shipping, and accepting returns from dealers (or allowing dealers to make a profit), the record companies stand to make much more money schlepping their songs over the Net than by pressing CDs. The Internet also opens up e-commerce possibilities like selling individual songs or even individual plays of songs or albums. The fear part is simple, too: MPEG level three (MP3) audio recordings are already flooding the Net with songs for free — songs of artists whose copyrights are being violated.

Not long ago, a recording industry executive equated MP3 with the atom bomb, so great is its potential impact on the music business. But it might have been more accurate to compare MP3 to a neutron bomb, since — like the N-bomb — MP3 just kills the artists, but leaves their prior work intact.

While the Internet opportunity was one record companies would inevitably have tried, MP3 definitely made the process run faster. This is in part because MP3 proved the whole concept of distributing near-CD-quality audio files efficiently over the Internet. But the copyright-blind nature of MP3 also forced the music honchos to take action since that's the way intellectual property protection works. Having a copyright doesn't mean the sheriff will shut down the violators, it means the copyright holder, himself, has the right to do the shutting down. It's the same way with all intellectual property: Ownership conveys only a legal right to protect the property yourself.

But MP3 bedevils normal copyright enforcement since there are literally millions of violators. It's not like shutting down an illegal record pressing plant, since tens, perhaps hundreds of thousand of people have downloaded free MP3 encoders and made illegal compressed copies of their favorite songs. You can't raid thousands of homes and offices. You can't even keep people from buying MP3 players like Diamond Multimedia's little pocket MP3 box, the Rio.

Since intellectual property rights that aren't enforced risk being lost, the music industry has to do something — and fast — or the limos and groupies and green M&Ms will be gone forever. The only thing they can do is try to agree on a single standard for encrypting and controlling digital music, and make the performance of that system even better than MP3.

That's what has unfolded for months in smoke-filled rooms as record producers, lawyers, and programmers hassled through many potential technologies. There are a half dozen candidate technologies for compressing, playing, and most importantly, controlling access to copyrighted music. This week, the two strongest players were going to make their moves.

Microsoft had announced awhile back that this would be the week it would show Microsoft Audio 4.0. The fourth version of any Microsoft product is bound to scare competitors because that's about time the Microsoft efforts finally start to get good — very good. That's exactly why IBM chose to try and preempt the Microsoft announcement with one of its own. IBM announced its alliance with Real Networks, combining Real's G-2 player and encoder with IBM's Electronic Music Management System, which has been well received by the music execs. Real has 85 percent market share in streaming audio, which appeals to IBM, and IBM has the content management code Real couldn't or at least didn't develop. It appears on the surface to be a perfect pairing of Microsoft enemies.

Only it didn't work.

I heard the IBM/Real deal coming more than a week before it was announced. But the deal I was hearing about wasn't what the two companies finally announced. The deal I heard was in the works was IBM buying Real Networks. I don't think I got this wrong. I think IBM got scared and backed-out, literally at the last moment. The two companies made an insipid and curiously undetailed announcement about intentions to work together.

Then Real turned around and bought Xing, the most successful MP3 software company. One could argue this was to bring Xing compression technology into the bigger IBM deal, but the more likely reason is to place Real in the burgeoning MP3 market where it can make hell for Microsoft when IBM loses its bid for de facto industry acceptance.

The Xing deal gives a second hint at why IBM may have got cold feet, since the Xing purchase agreement values Real at up to $7.5 billion. Maybe Big Blue just balked at the price.

While this intrigue was unfolding, Microsoft was playing compressed music files in the House of Blues in Los Angeles, wowing all critics and perhaps ending the whole discussion.

What Microsoft appeared to show at the House of Blues (you can never tell for sure at such events what's real and what's demotheatrics) was a whole new level of digital audio performance. Microsoft Audio can make CDs with half the file size allowing twice as much music at the same fidelity. At the somewhat sub-CD quality level of MP3 audio, the Microsoft Audio files are again half the size, so they'll move faster over the Internet.

Microsoft showed a Windows CE-based portable player from Casio that can hold up to four hours of music compared to the Rio's one hour. They did a comparison with with MP3, and with Real's G2 Player and the Microsoft code sounded significantly better to everyone present, including Mick Fleetwood, whom I'm sure was being paid to like it. And of course, there are also Microsoft's features to protect intellectual property, to enable e-commerce, file protection, crippling of files, digital watermarking, and other features to make any record company executive sleep easier. Lord save us from Microsoft's Pay-per-View Wizard. No wonder Microsoft was able to announce 63 commercial licensees at the launch.

The truth is that from the standpoint of the record industry, it doesn't matter which standard prevails, just that one — and only one — standard does. Record company sales have been flattening as baby boomers buy fewer CDs. A technical shift about now would be nice for sales, getting listeners to replace their CDs with the exact same songs in some new distribution format. That's what happened with CDs after 1983, but DVDs have too much capacity, and would have to cost too much just because they'd hold 10 hours of music. Sony's MiniDisk is a possibility, but it hasn't been endorsed by other manufacturers, and failed once already when it was introduced too early at too high a price. Nope, the Internet is better.

What works best from the industry perspective is if there is a new standard that is unopposed. After World War II, 45-RPM singles fought against 33 & 1/3rd-RPM Long-Playing albums, and the result was stagnation for several years as consumers waited to decide on replacing their Mario Lanza 78s in one format or the other. Everyone wants to avoid that same problem in this transition.

Of course, even if Microsoft wins, it could still lose. MP3 won't go away and with Real's money behind it, the contest could get very interesting. And on the horizon there's something more — MPEG level 4. IBM was probably smart to save its money.