Wish You Were Here (42 page)

Organizational life was something for which Douglas did not have the subtlest reflexes. He could be unreceptive to creative ideas from others—as if creativity was exclusively his job—and it was apparent almost immediately that the role of benign Godfather to other people’s projects (a role that all had hoped he might fulfil) was just not one that suited him, even if there were sufficient time. Douglas was out of practice in large meetings—after all it had been nearly twenty years since the BBC—and he would sometimes address his remarks just to one person in a way that could make colleagues sensitive to these nuances cringe inwardly. On the other hand, he often showed great warmth and kindness and never deliberately hurt anybody’s feelings; indeed he would be mortified when he realized that he had been a clodhopper. Douglas making amends was almost more embarrassing than the original slight. Despite the surface sophistication, in unfamiliar contexts the shy and gawky schoolboy could rise from his past.

The H2G2 site was very advanced technically for it was one integrated whole rather than lots of linked components, and was written in the computer language C++, which is unusual for a website. But despite its sophistication, Douglas’s dream for TDV was that it could develop some kind of artificial intelligence that would pass the famous Turing Test. That is still a long way off.*
 
¹⁹⁸
Instead he had to settle for something that looked like it, and thereby he came up with one of his engaging analogies.

“Think of it this way,” he told the team:

It’s a magic trick. When a magician saws his shapely assistant in half on the stage, we see her feet wiggling at one end of the box and her head, clearly too far from her feet, at the other end. We know she hasn’t actually been sawn in half, but the illusion is bloody good. We don’t want to know how the trick is done. That’s the kind of illusion of artificial intelligence I want with
Starship Titanic.
*
 
¹⁹⁹

Not being embarrassed by a surfeit of cash put all the more pressure on the timely production of
Starship Titanic.
Unfortunately this game, inspired by a joke in
Life, the Universe and Everything,
*
 
²⁰⁰
was hugely ambitious in scope, and almost beyond their resources to produce in the relatively short development time available to them, especially as the S&S money did not come through quickly. After a false start (Douglas’s fault, as he later admitted) over the buying in of some specialized software, the engineers wrote their own bespoke tools. Starting in the summer of 1996, delivering a giant game by September 1997 was always going to require a Himalayan effort.

The game itself is a remarkable achievement. From the outset Douglas had wanted a game with wit and humour that was also beautiful to watch. He had admired the graphics in
Myst
and loved the intriguing world it created, but he found it characterless and lifeless.*
 
²⁰¹
Starship Titanic
would be richly inhabited by characters, in this case robots, many of whom were obstructive in ways too baroque to be described.

Visually it was stunning. Douglas had been attracted to the work of the artist and designer team Oscar Chichoni and Isabel Molina, an Italian/Argentinian partnership of extraordinary talent. They were hired on a short-term contract and ended up working flat out for eighteen months. The look they created was that of an elegant but slightly sinister French ocean liner of the 1930s. The colours are unusual. Every surface is highly polished and elegantly decorated in an art deco style that is somehow suffused with an off-beat eroticism. The robots (the bots) with their smooth yet expressive faces and their sudden unfoldings, like mechanised bats, are both funny and disturbing. In fact there is something both comic and nightmarish about the whole enterprise. Isabel found Douglas very appreciative of the artwork, and receptive to suggestions that emerged from it to the extent of sometimes writing them into the script. The only problem with graphics so detailed and finished was the sheer amount of information they contained. Sometimes, Isabel says, the polygon count had to be trimmed because otherwise the screen would just take too long to refresh itself.*
 
²⁰²
As it is,
Starship Titanic
is a colossal game requiring no less than three CDs to run. Quite possibly it is the most elegant computer game ever produced.

Some of those CDs’ capacity was taken up with a plot which rested on narrative foundations as big as a castle. Making something as complex as a computer game is a team effort, so it may be unfair to list some individuals and not them all. However, the software engineers—Tim Browse, Sean Solle, Rik Heywood, and Mike Kenny—did an exceptional job delivering far more than seemed possible. Adam Shaikh worked heroically on the architecture of the game. Douglas could not really program, though he admired those who could, and sometimes he drove the techies mad trying. Robbie said “he wanted to be the Michelangelo who chips away the stone to reveal the statue.” TDV employed experts to do that. Wix, Douglas’s old friend, wrote some evocative music; Terry Jones was typecast as the parrot; John Cleese made a cameo appearance as the mystery voice of the bomb.

The story of
Starship Titanic
itself is deceptively simple and characteristic of Douglas’s view of the human condition: you’re on your own, the world doesn’t quite make sense and it operates according to rules that you haven’t yet worked out. The player has to negotiate an upgrade to a higher class of cabin while the desk bot politely but intractably closes off every rational route to that goal. Simultaneously a stroppy talking bomb is counting down to detonation, and the player has regularly to defuse it since the bomb has a perverse habit of rearming itself. The ship is deserted apart from a demented parrot and a weird collection of nicely observed bots, some of whom will drone on about their war wounds at heartbreaking length. The key to success in the game is talking to the bots—and this takes discretion and subtlety, for first-class minds have invested ingenuity in making the job horribly tricky.

There is a gigantic language engine built into the game. If you are so minded, you can talk to it for fourteen hours without it repeating itself. There are over three thousand sentences and phrases available to the bots, and they have a surprising range of cultural references at their disposal. It is an intellectual achievement of the highest order, one that required an endlessly inventive anticipation of player input and the subtle teasing of the audience along the route that the authors prepared for them. The language engine was very advanced—probably more so than the market required—but the culture of the company strove for excellence. Parsing natural language into recognizable components so as to generate an appropriate response from a library of possible replies is immensely complicated and it raises some interesting philosophical questions about meaning and the role of grammar in thought. An American baby and a Chinese baby are born with indistinguishable brains, and both set about learning their local languages thanks in part (according to Chomsky and others) to innate pattern generation and recognition talents built in by millions of years of evolution. A computer, on the other hand, is many orders of magnitude simpler than a baby’s brain, and its operating system (without which it is just an inert assemblage of components) may be all of ten years old. This is not the place to explore these notions, but one example of almost identical sentences with radically different syntax will suffice to illustrate the problem: “Time flies like an arrow” and “Fruit flies like an apple.”*
 
²⁰³

Douglas created the game concept and the basic architecture, and he and Neil Richards and Douglas’s old friend, Michael Bywater, wrote nearly all the character dialogue. Neil Richards also managed the text, and Douglas and Michael are credited with the game outline.

Not surprisingly such a groundbreaking game took longer to produce than they had planned, a grim outcome because the revenue from the game was effectively their only source of income. If they ran out of money before the game was on the market, they’d be bust. If they got more money from S&S Interactive, they would delay the receipt of their own income from the game—possibly for years. What’s more they would miss the Christmas 1997 market, and run the risk that the launch would be stillborn. It’s often impossible to rekindle excitement if something is announced, then delayed. Cash flow can kill a business even at its moment of triumph because it takes time to collect debts from the market whereas expenditure always increases before the launch of a new product.

In the end, after several missed dates, Robbie, who can be very persuasive, pleaded with S&S Interactive for more time and money—not an easy thing to do, for S&S prides itself on being tough. Gilles Dana, the Publisher of S&S Interactive who had taken over from Peter Yunich, flew over from New York to assess whether to pull the plug or extend some help. It was a frightening time.

Gilles Dana happens to be a strictly orthodox Jew, and the TDV management was not quite sure what to feed him. There were a couple of Jews on the staff who were consulted. Yoz remembers Douglas listening intently and without the slightest hint of satire while Yoz explained some of the unusual culinary regulations:

I laid everything on the table, and before Gilles arrived I lined up all the TDV directors and everybody who was going to be having lunch and carefully explained to them, “Look, this goes in here. If you’ve eaten that, you don’t eat this with that . . .” Enough for them to get the idea. Douglas was there. This was my first proper interaction with one of my all-time heroes, who happens to be the most famous, staunchest atheist in the world, and I have to explain to him how to keep kosher . . . It was deeply frightening. Douglas was very tactful, and asked a lot of questions. He knew how not to tread on people. Mind you, when Douglas was back in his office and, as he thought, I was safely out of earshot, he had this conversation with Alison [a colleague]. He was hitting his forehead with a cry of, “This is the twentieth century, for God’s sake.” And Alison said: “It’s
religion,
Douglas, it’s century-independent.”

As it turned out, Gilles brought his own food. He was, says Robbie, tough but totally straight, and he extended both their deadline and their funding, though on rigorous terms. Games are not like books which enjoy a market stable enough for there to be confidence that the “product” would still sell even if a year late. There was no leeway at all for further delay.

Emma Westecott was appointed as producer of the game, and she did an astonishingly good job in motivating the team and making up time. Nearly all the company’s efforts were now directed to getting
Starship Titanic
finished while a small team maintained the H2G2 site—though it was not commercialised. The website venture taking second place compounded Robbie’s difficulties in raising money, for his multi-media company was now engaged on only one product. There was a running joke in TDV that H2G2 was operating on the
South Park
“Three Phase” business model. Apparently in
South Park
(the anarchic cartoon series) the underpant gnomes abduct the infant heroes and take them back to a huge, cavernous lair in the middle of which is a gigantic pile of underpants. There’s a flip chart behind the pile, and on the chart it says, “Phase One—collect underpants. Phase Two—[a giant question mark]. Phase Three—profit.” The underpants business model, one can’t help thinking, basically sums up much of the strategy behind the dotcom explosion. Despite this gallows humour, everybody I have spoken to who worked on H2G2 was proud to have contributed to it regardless of whether it made a profit or not.

The pressure became overwhelming. The technicians frequently worked overnight and their room soon held mountains of print-out, tottering towers of discarded pizza boxes and changes of clothes. Robbie used to bring in food parcels for them when they were working on Sundays.

Under this crushing pressure, another of Douglas’s intense friendships became a casualty. Never employ your friends, they say, if you want to keep them.

Michael Bywater had known Douglas since Cambridge. Michael is a phenomenon. Once, Peter Bennett-Jones reports, after Michael had been consistently late for Footlights rehearsals, at which he played the piano, he had been threatened with the scrotal shears if he were late again. But at next rehearsal he turned up two hours after the appointed time looking wild-eyed and dishevelled. “Don’t ask,” he said. “I’ve just crashed a plane.” Michael does indeed have a private pilot’s licence (you will recall that he once flew a house party down to Douglas’s place in Provence). He plays the piano beautifully. He knows a great deal of stuff across an unusual range of subject matters. Once, at one of their smart dinners, Douglas and Jane, fatigued by Michael’s incontinent trickle of esoterica, especially of the medical variety, placed him next to Annie Coren, the wife of Alan, the humorist and writer. Annie is a highly respected consultant anaesthetist at a major London teaching hospital. After dinner she and Jane compared notes: “Michael,” Annie reported to Jane, “has a bluff that cannot be called . . .”

Michael Bywater is the most intellectually competitive person anybody is ever likely to meet. He is ferociously bright, and not inhibited from letting others appreciate the fact. He does not wear his erudition lightly. In an age when the uneasy journalism of self-revelation is popular, it was nevertheless surprising when Michael told his readers in the
Independent on Sunday
about an incident in his childhood. His parents, believing on medical grounds that he would be a singleton, adopted a baby girl. Michael adored her, but for whatever reason she did not bond well with his mother, and after a short time she was returned with great regret to the agency. All his life, Michael confessed, he had been at some emotional level convinced that if you did not pass muster
you could be sent back.
Human beings are so complex that the idea of a single motivational key to their behaviour is obviously crass—or best reserved for movies as with
Citizen Kane
’s Rosebud motif. Nevertheless, there is something driven about Michael’s need to impress. For God’s sake,
relax,
one wants to say: every sentence need not be a winner.