Tuesday 4 October 2011

Driver: San Francisco - Post Mortem & Script Samples

So it's 18 months since I completed work, a month since release (to console, at any rate), and 48 hours since I finally got my hands on Driver: San Francisco. I have some thoughts on that topic.

First, it's my biggest and most polished release, so congratulations and thanks to the guys who did the real work over at Ubi Reflections. Thanks also to my agent, Sidelines, who put together the incidental dialogue team that I was a part of.

Despite being a big-bombast driving experience (ie the other end of the spectrum from my usual interests), Driver has my heart. I say this with what I hope to be a reasonable degree of independence - I never played the game before it's PC release, and I had nothing to do with game design or the central narrative thread - but Driver: San Francisco is that rarest of things: a game with a defined tone. I'm not the first person to say that the black humour, campy cop show theming and absurdist premise combine to deliver a world that works by its own rules and is a joy to inhabit.

But what exactly did I have to do with all this, and what did I learn? I produced a pretty significant portion of the Act 2 NPC dialogues - the flavour conversations (or barks) that occur when you shift into a car with a passenger. They were actually the topic of their very own bit of marketing, being as there are more lines in Driver than in Mass Effect 2. I'm not in a position to comment on that one, but I know I produced about 50,000 words across 30 characters and that others were doing similar. At any rate, RPS had some nice things to say about these dialogues in particular, so I hope they're worth discussing.

If the experience has reminded me of anything, it's that playing a game is essential to writing one. Stupid thing to say, I know, but it's amazing how often that principle is disregarded. In the case of Driver it was very much a priority job (which is a nice way of saying a rushed one), and though I asked for playable code I rolled over pretty quickly because sometimes a client just wants you to do the job you've been assigned, and are paying good money for freelancers so they have less to worry about, not more. I can't say I blame them, but I can't say I'd roll over so easily next time either.

The stuff I produced for Driver works, for the most part. If you're interested, a cursory play suggests to me that about half the in-car stuff around chapters 3-6 is mine, and you can find a full list here. A bit like putting a novel in a cupboard for six months, such a big gap between writing and playing lent me some objectivity, even to the extent that I was struggling to pick out my own characters. The darker ones work the best for me: the kid whose one day with his estranged father is ruined by Tanner's interference ("I think you broke my arm again, dad."); the hospital director who's terrified of winding up in her own intensive care unit ("You do understand our surgical staff are barely trained chimps?").

Where things falter a little is in the gameplay, or at least my lack of knowledge of it at the time. These scripts are supposed to be quick fire, simple to grasp scenarios that pack a funny / dark / atmospheric punch. Sometimes playing the game almost feels like a formality - after all, a scene is a scene, and a line entry marked 'Jump Land' or 'Scrape' seems pretty self-explanatory; but context is everything. How long is the delay between timed lines? How much more violent is a crash compared with a scrape? Does the dialogue reset if the player leaves the car?

The result is not that the good lines aren't good: the material still works; but a lot of it is lost in the mix. The ends are cut off of the wipeout lines; the story flow breaks in certain scenarios; lines you expect to be the mainstay are rarely heard.

I'm very happy with the work we did on Driver, and proud to have been involved in its development. Most of all, though, the experience cements for me what I've always suggested: that a games writer's job is still (and may be for some time) not so much to do what they're told, but to do what they ought. I failed to uphold that principle and at times, and perhaps only to me, it shows in the game.

If you're interested in the practicalities of writing for games you can check out a sample from the Driver: San Francisco script over at the Narrative Design Resource.


  1. I think these little snippets work really well, and the game as a whole I've found magnificent (http://www.sizefivegames.com/forum/index.php?topic=4207.msg56386)

    My only niggle with "your bits" is, again, gameplay based. Because you're often Shifting from car-to-car in quick succession, you get a pointless 2 second cut of the dialogue before you're off floating in the ether again.

    If it were me, I'd just add in a little timer and make sure the player's committed to that car to some degree before playing any VO, otherwise you get a weird "Hey! Wha... Well darli... Who a... Is it m...", which breaks the immersion because you know what's going on.

  2. Cheers mate! Agreed - would have been such a simple fix. Crazy thing is for each character we produced something like six different intro lines and it seems as if those lines are triggered in order in the game (ie variation 2 is played the second time you zap into the car). If we'd known that we could ahve written it appropriately so that the story didn't get so cut up.

    On closer inspection, I also realise some lines (mostly the intros) have been edited for length, and only three variations are being used out of the six. It's a good thing this happened since on the initial brief I know we were writing lines that were just too complex for the moment-to-moment gameplay, but it could have been a lot easier.

    To stress again, though, it's my position that the responsibility for this stuff lies with the writer. I write games every day, and I know that writing a script independent of the game is a risk. I don't know who hired or co-ordinated my writing team (it all goes through the agency), but they probably knew a lot less about games writing than we did. We were in a position to make things work better, but instead we did what we were told and took the money. The studio was happy (probably happier than they would have been if we'd come back with a load of suggestions and demands for code), but paychecks aren't why any of us are in the industry, and that's something worth remembering.

  3. I found San Francisco itself to be a touch flavourless but hopping in and out of cars and eavesdropping in on conversations - even if I were only getting a sentence or two without context - breathed life into the city. It made it a little more three dimensional.

    If I were hooked on the first line of dialogue I'd stick with a car and to the credit of all involved, I was frequently driving some atrocious vehicles. Had there been a three second counter or something to similar effect I'd have probably never driven a bus through a traffic jam or stuck around to hear much of the dialogue, content with hurrying about in a Lamborghini instead no doubt.

    There were some kinks but I think it effectively served a dual purpose.

    It was a great game and I say that as someone who went into it highly sceptical.

  4. Hey Tom, I was one of the guys on the development team, as was James Worrall (who started on the outside in Sidelines but became internal half-way through). I remember reading and enjoying some of your scripts.

    The reason you only hear 1/2 of the intro lines is because most characters were duplicated into two different chapters, and the intro lines split. Originally we had twice as many character scripts, and this got cut down for budget/time reasons. So you may discover that if Shift-in lines 1-3 are in chapter 2, the character reappears with their lines 4-6 in chapter 6 (though all their other lines will be identical in both chapters).

    The ordering of the line plays was something I really, really tried to get randomised but wasn't happening code-side with the audio guys. I think we did manage to randomise the other lines, but not the shift-ins. There were all kinds of arcane reasons why things weren't possible.

    Believe me, a lot of sweat and tears went into trying to make it as polished and non-glitchy/repetitive/cut-offy as possible; you should have seen what it was like 8 months before release! The reviews, at least, are testament to the fact that we ended up getting a few things right, thanks as always to collaboration, good material, and regrettably, plenty of long hours and the occasional 'spirited debate' (AKA me throwing toys out of the pram when the programmers say 'we can't fix that'!).

    I really enjoyed the game when it came out, too, so I generally feel the lessons to take out of the experience are to try and communicate, to plan ahead, to educate and debate with non-writing members of the team about why the writing matters, and how to make the implementation better.

  5. Good to hear from you, David, and good to get the inside story.

    Sounds like we're all on the same team :-)