Diversification is a powerful concept, and every portfolio needs to have a generous dose of it. But when individual companies start to cover new markets like a giant kudzu, they'd better have a good reason to do it. A badly planned or executed strategy of diversity can do a lot of damage.

Some folks call it "diworsification," a term coined by Peter Lynch to describe how companies can go wrong when they acquire unrelated business to "diversify" their revenue.

Examples, please!
The blueprint for industrial conglomerates is General Electric (NYSE:GE). Today, GE is a collection of media assets, heavy industry, consumer goods, and even a banking operation big enough to use up a quarter of the government's Temporary Liquidity Guarantee Program.

Keeping a heterogeneous operation like that going for an entire century is no mean feat, but the strategy has required GE to reshape itself every once in a while. GE has moved in and out of broadcasting, public utilities, computer technology, and financial services during its long history. Buying, selling, starting, and closing business segments is standard procedure for GE.

But that tactic hasn't been working too well recently. The banking operation is suffering under the same duress as every other bank. NBC, once the indisputable king of network television with megahits like ER, is now struggling to keep up with CBS, Walt Disney's (NYSE:DIS) ABC, and Fox. GE wants to sell its household appliances division, but can't find a buyer.

And GE's blue-chip status is getting chipped. Investing in an S&P 500 tracker two years ago would have caused you 28% of pain in that time, even if the results are adjusted for dividends. GE would have hurt you more than twice as badly during that time.

What should I worry about now?
When Oracle (NASDAQ:ORCL) decided to buy Sun Microsystems (NASDAQ:JAVA), the buyout left a lot of people -- myself included -- scratching their heads. Why is software giant Oracle interested in Sun's storage and server hardware?

Likewise, Cisco Systems' (NASDAQ:CSCO) designing its own server-class computers is kind of like Brett Favre playing for the Vikings. Cisco is stepping on the toes of many of its biggest customers and fans by playing ball in a new market.

What should Oracle do, then?
These moves worry me. Oracle is a superb software business with sky-high margins and crushing market shares in several important sectors. But the company has no experience or exposure at all in the server and storage markets, and hardware traditionally offers lower margins and return on investment than software, anyway. Diworsification at its best -- or worst.

Now, Fortune says that Larry Ellison might be thinking about handing the hardware side of Sun off to Hewlett-Packard (NYSE:HPQ) -- and I think he should. HP does have the infrastructure and expertise to sell Sun hardware in an effective manner, and its EDS subsidiary was Sun's top customer before Hewlett-Packard scooped it up. There are lots of efficiencies for HP in a move like that, and Oracle could see a decent payoff, too. Ellison really wants Sun's software and services -- and that's all. Larry is not known to settle for mediocrity, so I can see this play happening as soon as the Sun buyout is closed.

How about Cisco?
I often hold up Cisco as a shining example of corporate vision and long-range thinking. But it's hard to see how the upsides of Cisco's "unified computing system" could outweigh the negatives.

CEO John Chambers seems excited about early sales of this new product and says that "customers are giving us absolute permission to come in here." Sure, your customers want to unify their hardware providers as well as service and support contracts a little bit. Networking and computer systems do go hand in hand like that.

But Cisco's biggest resellers include (oops!) server giants HP and IBM (NYSE:IBM). Those are some powerful OEM sales channels. Is Cisco really prepared to lose the loyal assistance of HP and Big Blue over a product line that might contribute a fraction of Cisco's sales, in a couple of years, and assuming that everything works out according to plan?

Not worth the gamble, I say. Cisco, too, is diversifying in a bad way.

What now?
I still like both Cisco and Oracle in general terms, but these bone-headed decisions are making me rethink that position. When I buy a single stock -- instead of a mutual fund or exchange-traded fund (ETF) -- I kind of like to know that I'm getting the focused management I was looking for. The closest thing to a conglomerate in my own portfolio is probably Disney, and even there the whole theme park, cruise ship, media puzzle is built on the same rich foundation of applied movie history.

Until Oracle sells the Sun hardware and Cisco forgets about servers, my money is going elsewhere. How about yours? The comments box below is dying to hear all about it.