One of the entertaining things about business is that executives must sometimes, in the interests of their shareholders, say one thing for years and then reverse themselves and declare the opposite.

This can involve double-crossing, or at least angering, those who believed the original story. So it goes.

It often happens when a big company takes over a small one. At first, the acquirer insists that its new subsidiary will be left alone and its managers will be free to carry on as before without interference from head office. Then, one day, the long leash becomes a short one.

The question is not whether this is bad behaviour - in fact, it is life - but whether it is wise. Will those who work at the subsidiary be so annoyed that they walk out, leaving little behind? Or will the U-turn work?

I caught up with Franz Humer, chairman of Roche, the Swiss pharmaceuticals company, in the middle of this manoeuvre on Tuesday afternoon. Having just talked to fund managers at the St Regis Hotel in New York, he was on his way to the airport to fly to San Francisco.

He was due to have dinner there with Art Levinson, head of Genentech. Humer had just sprung on the independent-minded Levinson a $43 billion offer for the 44 per cent of Genentech that Roche does not own. It does not sound as if Levinson was pleasantly surprised.

Potential disaster

Humer had time for a quick chat. No, he had not wanted to upset Levinson by telling him on Sunday evening, only hours before Roche made its offer public. Yes, he hoped Levinson would stay. Still, Genentech was not a one-man band but "a pool of talented people".

Ouch. Behind Humer's elegant explanation of why the takeover he had always dismissed as a potential disaster now made sense, there was clearly some hard calculation. Roche could take control of Genentech and keep most of what made it the world's most valuable biotechnology company, with or without Levinson.

I have no objection in principle to Humer's switch. If it works for Roche investors, it is good enough for me. After all, other enterprises, from Hollywood to Wall Street, have pulled off this trick. My problem is that I still believe the old story.

Humer's former argument was that it was best to keep Genentech at arm's length from Roche in order to maintain the enterprising spirit of the US company. He did not want Swiss discipline and bureaucracy to override Californian innovation and free-spiritedness.

Roche acquired its stake in Genentech in 1990 and, in 1999, it first bought the rest and then floated a minority through an initial public offering. That gave Genentech both a degree of independence from Roche and share options with which to reward its scientists.

Genentech has invented some of the most important and valuable gene-based medicines, notably the anti-cancer drug, Avastin. It has been one of the biggest forces giving Roche an edge over traditional 'big pharma' companies such as Merck and Pfizer. Roche and Genentech have a promising pipeline of new drugs in an industry that is running dry.

So you would have thought that upsetting Genentech was the last thing that Humer would risk. Indeed, some have attributed Roche's change to Humer being succeeded this year as chief executive by Severin Schwan. (Humer says this is untrue; his views evolved.)

If Genentech were willing, Roche's move would be logical. As Genentech has grown from a start-up to an enterprise with 11,200 employees, it has acquired the trappings of a pharmaceuticals group. It conducts clinical trials separately from Roche, manufactures its own drugs and has an independent sales force.

The industry is trying to save money by becoming more efficient and Humer wants to eliminate some of the Roche/Genentech overlap. He believes it can be done without affecting the heart of Genentech's value: its research and development. Scientists will be allowed to carry on doing their own thing, he says.

There are similarities to Hollywood, another industry of complex relationships between big distributors - the major studios - and small production companies. Hollywood is also facing severe cost pressures and has been trying to shed unnecessary expenses and duplication.

Careful handling

Time Warner recently folded New Line, an independent studio with a big infrastructure, into Warner Brothers. That led to the departure of New Line's co-heads but was a price worth paying. Studios need a network of small producers to make films but need not pay twice for distribution and marketing.

A better parallel than New Line for Genentech, however, is Pixar, on which Walt Disney has come to rely for its best-selling animated films. Disney had a distribution deal with Pixar but Bob Iger, Disney's chief executive, persuaded Steve Jobs, then Pixar's chief executive, and John Lasseter, its creative head, to back a $7.4 billion takeover in 2006.

Genentech is as valuable to Roche as Pixar was to Disney and is also a Silicon Valley company whose employees are creative and need to be handled carefully. Yet Roche has gambled on being able to obtain the backing of minority investors and achieve a fait accompli.

This strikes me as risky. Not only may it provoke Levinson and others to walk out; it also undermines Roche's appeal to any other biotech companies it needs to cultivate. It will be harder in future to present itself as a trustworthy partner.

Humer is not being perfidious: it is a fact of life that acquirers do not stick to promises indefinitely. But he has given up something of value. I hope it is worth it.