One day in 1996, 10 men filed into a conference room in Shibuya, Tokyo's trendy shopping district, and mapped out a plan to restore their company to its former glory. At the heart of the scheme was a top-secret product known initially to employees as Black Belt or Katana, the ancient Japanese sword, a weapon to put the humiliating defeats of the past three years behind the company. "It was a fresh start," said one.

In fact Black Belt - more prosaically known as the Dreamcast video game console - brought Sega nothing but misery. Five years later, most of the men at the Shibuya meeting have quit. The company has been rescued from bankruptcy by Isao Okawa, the wealthy benefactor who donated more than ¥135bn ($1.1bn) to Sega before his death earlier this month.

How Sega veered off course is not like those tales of Japanese management woe in which engineers overpower financial planners or senior executives bungle negotiations with a foreign partner. Instead, it is the story of a management that debated a project to death - and nearly a company, too.

"It was a management problem. Nobody was looking at the big picture," says one former executive.

Sega had every reason to make Dreamcast a success. By 1996, the former jukebox importer and the creator of Sonic the Hedgehog, Frogger and Virtua Fighter had spent a decade failing to conquer the home video game console market.

The Saturn, its previous machine, had launched to lukewarm reviews in the US in 1995, selling 9.1m units. It had been upstaged by PlayStation, made by rival Sony, which was already dropping hints about PlayStation 2.

"We wanted a return match," says one person involved with the project. "We said, 'we'll get them next time'."

Everyone in the room in Shibuya that day could agree that Sega urgently needed a successful product. But the problem then - and for most of the next five years - was that nobody could agree on much else.

Executives bickered over the launch. Some wanted to start selling the console in the US, Sega's largest market outside Japan; others insisted on starting at home. Then there was the network connection: a modem would raise manufacturing costs by about ¥3,000-¥4,000 a console but most executives agreed that US gamers, at least, would pay no more than $199 (£140).

"We wanted to make the cheapest, best product on the market. It would have been fine to add the network connection but to sell that machine for $199 was extremely difficult," says one person involved in the discussions. Managers dithered over the modem for months.

Executives also disagreed about the specifications for the all-important graphics chip - semiconductors represent 70 per cent of the cost of a console, and its popularity depends heavily on its graphics capability.

"We wasted three months on the chip debate," says one insider. While executives bickered, memory prices jumped 30-40 per cent, raising Dreamcast's production cost higher than anyone had imagined.

The long debates reflected a management undergoing a lot of change for a

Japanese company. Three presidents - Hayao Nakayama, Shoichiro Irimajiri, and Isao Okawa - oversaw the project.

Mr Nakayama, who was recruited in 1984 partly because of his English skills, knew the industry but failed to gain employees' respect. "He didn't listen to what people said," says one former executive.

Mr Nakayama enjoyed a reputation as an internationalist and a reformer. He promoted executives who spoke English well and cut 20 per cent of the workforce. Although this impressed the local media, it prompted an exodus of disgruntled employees. Yet during his tenure, Sega paid little attention to marketing. "The idea was, 'if the product is good, it will sell itself'," recalls one former manager.

Although Mr Irimajiri hired more marketing specialists and tracked product development, his authority was undermined when Mr Okawa, a leading shareholder, began to exert his influence over daily operations.

Mr Irimajiri and Mr Okawa disagreed on such issues as the price of Dreamcast, according to former executives. Their animosity was institutionalised after Mr Okawa hired nearly a dozen executives from Nomura Securities just as Mr Irimajiri was recruiting a smaller group of his old colleagues from Honda. Instead of an ideal blend of Honda's marketing and Nomura's financial expertise, the mixture proved toxic: the Nomura camp pushed for cost cuts while the Honda team defended spending on marketing.

Sega gave NEC the chip specifications at least three months late, holding up game development. Other developers - specifically Square and Enix, the industry darlings - refused to make any games at all, despite a year-long campaign by senior Sega executives to bring them on board.

In mid-November 1998, about a week before Dreamcast was to hit the Japanese market, NEC Kumamoto, the part of the electronics group making the graphics chip, told Sega they could only supply 25 per cent of the order by the launch. When stores opened on November 27, 1998, there were only 200,000 consoles, not the 600,000 originally planned. Instead of 15 games, there were only four.

In the games industry, a product's destiny is decided within one month of its launch. Even before it had started, Dreamcast's game was as good as over. "Nobody ever knew whether Dreamcast would succeed as a business, anyway," admits one former employee.

In the months that followed, the debate among top management became even more rancorous. Senior executives visited Mr Okawa's office in Tokyo's prestigious Ark Hills building nearly every weekend between February and April, armed with proposals to prop up the company's finances.

The company started talks with telecommunications carriers, including France Telecom and Deutsche Telekom, and US cable television providers to convince them to purchase Dreamcast technology for use in set-top boxes. One senior executive went as far as to arrange a meeting with Mr Okawa and Electronic Arts, the US software developer, to discuss a possible capital alliance. Still others argued that Sega should withdraw from Dreamcast and start afresh.

But Mr Okawa, who was ill with terminal cancer, "didn't want to talk about it", according to one executive who visited him during that period. Instead, Mr Okawa named Hideki Sato, a close ally and a strong supporter of Dreamcast, as vice-president and titular chief. In a restructuring announced in late October, the company formally appointed Toshimichi Oyama and Tetsu Kayama, formerly of Recruit, the job search agency, as special advisers.

The new appointments proved too much for some senior managers, who resigned. Mr Irimajiri had stepped down in June to become vice-chairman at Mr Okawa's insistence.

With Mr Okawa, who had bailed out the company once already, in poor health, the board finally agreed to withdraw from Dreamcast and focus on software development instead. In a country where the lifetime employment system makes tettai, or withdrawal from a market, relatively rare, that was a difficult choice. But there was no other.

"It was inevitable - there were no other options left," says one former executive.

But the move has only eliminated part of the problem for Sega. Many of the men who gathered in Shibuya in 1996 were reunited at Mr Okawa's funeral in Tokyo. The executive's death "closed the door to an entire business model", one says. The challenge for Sega will be opening the door to a new one.