David Speyrer was interviewed for his job at Valve on the day the original Half-Life went gold in late 1998. He started working there soon after, just as Half-Life 2 began its lengthy development. Since then, he's risen through the ranks from programmer to project lead.
Now, David talks us through some of the processes that led to Half-Life 2: Episode Two being the remarkable game it is. (Warning: there may be spoilers ahead...)
Has Episode Two taken you anywhere near where you'd thought the games would head after Half-Life 2? The escape-from-the-city buddy story?
David Speyrer: No. Well, during Half-Life 2's development, there were times when it was incredibly difficult to predict what the product would look like at the end. That's part of the huge monolithic development cycle - they feel very unpredictable.
At the beginning of Episode One, we did have a pretty good road map story-wise of where we wanted to go, broadly speaking. We knew we wanted to get to the event which concludes Episode Two. We didn't know all the details of how to get there, and a lot of that's driven by gameplay experiments - what succeeds and what fails directs the story quite a bit.
But we had a story to tell about Advisors and about Alyx and about the struggle of the Resistance versus the Combine. We had goals for things like setting - wanting to take players into more open environments.
The final battle is an example of a scenario we were interested in exploring. Really broad, non-linear combat with a freeform epic feel. But no, we couldn't predict Episode Two as clearly back then.
There seems to be a real depth of narrative structure. The narrative is far more focused than it was in HL2. Is that deliberate?
Speyrer: I think we're learning how to tell stories in this medium with every product we release. With HL2, we introduced the storytelling technology which governs our ability to do these scenes - the facial animation and all that.
We were building it as we built the game, so it was really hard to predict the quality of the scenes we could do by the time the technology was going to be done.
We targeted a pretty conservative goal with our dramatic scenes, mainly because of the uncertainty of what we could produce. Then on Ep One, we pushed it a bit further. Then with Ep Two, from the outset, we wanted to tell a more coherent and satisfying story - we set out to really craft an arc.
I mean, HL2 did have an arc. At the beginning of the game you see the citadel - you end up there. There's a fair amount of foreshadowing and what-not. But specifically in terms of the story arc of Ep Two, we really wanted to make it feel completely satisfying and somewhat inevitable, and reveal some specific meaty back-story and character details in this episode.
A lot of our customers feel we don't say enough in our story. They really want to learn more about our universe. We're trying to do more of that, and it really comes through in Episode Two. We had a lot of comments from people who say that the story of this episode is really cool.
Perhaps the most satisfying moment was reuniting Alyx and her father Eli. They were always separated by screens which fizzle out. You had a fear they'd never meet. And then it finally occurs in that oasis in the middle of the action.
Speyrer: We really wanted to create a calm in the story in that point in the game. So much of the episode pushes you forward. There's the urgency of getting to White Forest before the Combine, and getting the data package there and all that. And then the final battle demands so much of the player.
It's so intense and epic feeling. We really wanted to create a contrast to all that when you arrive at the base. Without the contrast in pacing, the impact of the high moments, the action moments, is less.
Did you always intend for Alyx to play such a major role?
Speyrer: Yeah. When she was created for Half-Life 2, she was designed as a focal character. But I think she really matured as a character in Ep One, which was when we really sat down and designed her. Alyx in HL2 was designed by the writing of her character - and she has a great character. She's smart and charming and witty and tough and scrappy and all those things.
But in Ep One, because she was with you through the game, it was the first time we really sat down and almost engineered her character. We figured out the things she couldn't be. She couldn't ever be annoying. She could never seem critical of the player.
People when they're playing the game are going to naturally make mistakes, and if Alyx calls attention to the mistake in a way which seems at all critical, players really react. It becomes a negative experience. They like her less, and they like how they're doing less. They feel like they're doing poorly because someone's telling them they're doing poorly.
In Ep Two, when Alyx is riding along with you in the car and you slam into the tree, she laughs about it. She notices it, because it would be strange if she didn't. But we had a few reactions from her in Ep Two which came off as exasperated. "Oh, she's getting annoyed with my driving. I'm already feeling bad because I ran into a tree."
Do you think she's a muse for Gordon?
Speyrer: In the sense of being an inspiration? You need a character like Alyx to interact with others in the world so that their character can come through more richly. Eli shows a lot of himself when he talks with Alyx - that he's a caring, tender father.
Though a lot of his character comes through with his interaction with you as well. But really, I think Alyx is important as a motivator, to provide the player with someone to feel strongly about. She's kind of a good reason to fight, and she's a great vehicle for bringing out other characters.
That seems like quite an achievement. Companions tend to be infuriating. How did you avoid this with Alyx?
Speyrer: Alyx doesn't generally dictate the pace of your forward progress through the game. When she does that, it feels like she's playing the game and you're the tag-along. We learned this in Episode One, and applied it in Ep Two as well. The player should dictate the pace of moving through the game.
And we're very careful that it's the player who solves most of the problems. Alyx can state a problem - "We should really figure out a way across this bridge" - but it's important the player should be the solution. It's not much fun watching an NPC play a game for you.
We deliberately keep the player as the prime mover, and Alyx's job is really to augment the gameplay, create interesting gameplay scenarios, and reward you for a job well done. There's a whole set of rules - the player dictating the pace and Alyx not stealing the fun - that we make sure we follow. It was a lot of work achieving the Alyx you see in the game today.
It took a lot of iterations, and even all of her responses to the events which happen in the game took a lot of fine tuning. She can't talk too much, and she can't be too quiet either.
There's a middle ground where she's talking enough to show awareness of what's going on around her. We try to make everything she says interesting or entertaining, so we really pay attention to how people respond.
The car was a real surprise. It gives the impression of incredible speed, but you always feel in control.
Speyrer: Well, you have two groups of players. We ended up striking a balance between what I guess you could call performance drivers and casual drivers.
The performance drivers want to say, "If I floor it and cut the wheel really hard to the left, then I should fly off the road," and casual drivers want to think of it as a really fast player, which does whatever they say. There was a difficult balance to achieve satisfying both groups of players.
But there's quite a bit of code helping you out as you drive to keep you on the road as much as possible, despite you giving conflicting orders to the car - steering it and flooring it, for example. We change your field of view in the car, which accentuates the sense of speed.
We also improved greatly how turbo worked. In Half-Life 2, turbo was pretty much the 'Crash The Buggy' button. You'd press it, and pray it goes in the right direction. In Ep Two, you can hit turbo and really drive - you stay on the road, the steering works really well - it's a fun way of driving around by hitting turbo a bunch.
Games are terrible at endings. In fact, game endings are always shit. Why aren't Valve's?
Speyrer: We're very self-critical of our past work, so we constantly want to get better at things like endings. And with this ending, we took a bigger risk than we had in the past. The payoff is there. I think it's going to tell us that we need to be more and more willing to take big risks like that.
There was a moment when Alyx noticed the zombies were Combine and then made a joke calling them "zombine" obviously Gordon doesn't laugh and she responds with a sort of "oh well ok then, not so funny" and I thought that was brilliant the way they put that in.
Games are terrible at endings. In fact, game endings are always s**t. Why aren't Valve's?
Speyrer: We're very self-critical of our past work, so we constantly want to get better at things like endings. And with this ending, we took a bigger risk than we had in the past. The payoff is there. I think it's going to tell us that we need to be more and more willing to take big risks like that.
I really liked the ending to episode 2. It was a bit of a suprise at the time (and that's all I'll say to avoid spoiling it for those that haven't seen it yet), but I think it really helps set up episode 3 nicely... I'm looking forward to seeing it.
Copyright 2006 - 2009 Future Publishing Limited, Beauford Court, 30 Monmouth Street, Bath, UK BA1 2BW England and Wales company registration number 2008885
