Last week I was at the CTIA show here in Atlanta where David Munns, who heads the company's operations in North America, announced a big initiative to serve cell phones with "ring tunes," that is, ringtones based on real recordings by real artists you know. (Picture from Janalaweb, a Portugeese management portal.)

The week was filled with optimism, talk of big bucks coming in from Asia and Europe, talk of that success being replicated here.

Then this week, fairly quietly, EMI canned 20% of its people, 1,500 in all, including 20% of its artists.

Lawyers cost money. Or as Munns himself said several times while in Atlanta, "It's complicated."

The big music publishers, including EMI, are in a race to the bottom of their pockets against a worldwide interest in trading music files online.

Who's winning? The lawyers. Who's losing? The artists.

And the lawyers don't always win, either. A Canadian federal court's decision that just putting music files into a directory that can be shared doesn't make you a pirate, subject to jail time, was the second blow to the industry in a few weeks. A Canadian judge had earlier ruled that the industry couldn't file "John Doe" lawsuits against music swappers -- it had to find evidence against individuals and file against individuals.

Now it's not like Canada is a haven for piracy. There's growing interest there in the idea of taxing everyone for music downloads. The result would be a "Canadian Principle" -- a paid-for copy is a legal copy.

I'm sure David Munns would have many objections to this principle in action. Some may be valid. Well, David...

It's complicated.

But direct negotiations between the industry and, say, government, is going to bring a lot more money to artists than your solution, which so far is only making lawyers rich. I suspect at least one-fifth of your present artist roster (those you're about to dump) might agree with me on that point. And it might bring more money into your pocket than the alternative as well. The bottom line, after all, is the bottom line.