Half-Life 2: 20th Anniversary Documentary

0:00 (intriguing music)- I always have been doing

0:09 since my childhood a bit of urban exploration.

0:12 So I love to break to a window

0:14 of an abandoned factory or building or train station.

0:17 So that's one of my hobbies, used to be.

0:20 What's on the roofs of buildings.

0:23 What's in an abandoned building cellar and in a factory?

0:27 We did a lot of work on this.

0:28 (intriguing music) (intriguing music continues) At the time when

0:43 I did my first photo shoot of the Austerlitz station, it was under construction.

0:49 But then there was no security guards, no fences even.

0:52 You can just walk and see the broken

0:55 textures and layers of paint and brick and history.

0:58 And it was just so cool that I can get

1:01 and get this information before it got overbuilt with a new neighborhood.

1:06 I went to check it out last night and it

1:09 happens to be completely destroyed and under construction again.

1:12 And it just goes back, luckily,

1:16 coincidentally to the exact same state where I shot it 25 years ago actually.

1:22 (intriguing music) (slowed down gunshots) (radio beeping) (train

1:37 horn blaring) (intriguing music) (intriguing music continues) (intriguing

1:58 music continues) (intriguing music continues) (intriguing music continues)

2:20 (intriguing music continues) (intriguing

2:30 music continues) (intriguing music continues)

2:39 (music fading out) (compelling music)- So one

2:54 of the big challenges with "Half-Life" was just establishing

2:57 the company and sort of determining whether or not

2:59 we could actually make video games at all.

3:02 My background was operating systems and spreadsheets.

3:05 Mike Harrington, who was the co-founder, had a similar background.

3:08 He had a little more game experience than I did,

3:10 but with "Half-Life," we were pretty sure we were gonna fail.

3:13 And then "Half-Life" was successful.

3:16 It was like a lot more successful than we had any reason to expect.

3:21 And now we had to follow it up, right?

3:22 We didn't wanna have a sophomore slump We

3:25 didn't want to be a one-and-done kind of company,

3:28 so now we had the terrifying obligation to our customers

3:32 and to our fans to somehow advance our game design skills,

3:37 our technology all needed to level up for the second game.

3:43 I think a bunch of us were really shocked

3:45 by the success of "Half-Life 1." We were this young, unproven company.

3:49 I don't think any of us really knew what to expect when we finished the game.

3:53 I remember when we started working on "Half-Life 2,"

3:55 we had already set the bar pretty high for ourselves.

3:58 So during "Half-Life 2," what we were really trying

4:00 to do is take advantage of the visibility that "Half-Life" had

4:04 created for us in the industry to continue to hire

4:07 the best people that we could find in the world.

4:10 It just got easier.

4:11 The bar didn't get any lower, but the fact that we had, it went from, "Hey,

4:15 this weird dude who worked on operating systems wants

4:18 to convince you he can run a game company" to, "Oh,

4:21 'Half-Life' won a bunch of awards and now they're

4:23 working on a sequel to that."- It was either

4:25 my first week or my first day was actually

4:29 the going away party for Mike Harrington, the co-founder.

4:33 Yeah, I came to Valve when I was just tiny.

4:35 We had, I think, maybe 10 or so people.

4:40 Dave and I grew up together in Louisiana,

4:43 so he and I had always talked about maybe starting a game company at some point.

4:48 And during "Half-Life," that kind of changed

4:51 to a, "Maybe instead of that, the dream

4:54 could be you come and work here at this one with me."- Yeah.

4:58 And I interviewed the day that "Half-Life" went gold and I took the team photo.

5:03 [Jay] Yeah, there was like an old photo of us.

5:04 We broke a paper mache Headcrab and everything.

5:07 Yeah, I was freaking out.

5:09 Yeah.

5:09 (both chuckling)- I was actually working on a video game concept in high school,

5:15 and I was seeking a mentor and I got in touch with Marc Laidlaw at Valve.

5:22 I guess it was the right time and right place and quite a bit of luck,

5:26 and I ended up getting an internship that way.

5:29 I guess I was 17 at the time.

5:31 So you've got a whole company that's familiar with tools and a whole process,

5:36 and then we just started over.

5:38 We rebuilt the tools and we rebuilt the engine.

5:41 So we had this long period of time where

5:44 we could sort of take our time on design.

5:47 And so we spent a lot of time working on story

5:50 and just trying to figure out how to make this thing bigger,

5:54 like somehow more epic.

5:56 And those discussions eventually led to this idea that, so in "Half-Life" one,

6:03 you open this portal, some strange rift to another dimension.

6:06 And we thought, "Well,

6:09 what if that broadcast this signal to the universe that's like,

6:14 'Oh, there's this planet there.

6:18 It's free air and water.'" And so then we

6:21 sort of came up with this idea of, "Well, there's this mega organization that's

6:26 resource-starved and they're just gonna come

6:29 and take it if they can."- How many years passed between?

6:34 Six?

6:34 Yeah.

6:34 So I would say the first two years we tried a lot of things,

6:39 and a lot of those story arcs that we had initially

6:42 didn't have really any carryover from "Black Mesa" or any of that.

6:47 There was originally gonna be three different

6:50 alien races like the bugs or religious ones,

6:53 and the Antlions actually came out of that bug race

6:57 that was supposed to be a much bigger part of the game.

7:00 It had three alien races.

7:02 The Warrior Aliens, the Insect Aliens, and the Spy Aliens.

7:07 And it had a Prague-like city,

7:08 which ended up being the closest thing to City 17.

7:12 In a lot of ways, it reminds me of writing music.

7:14 Everyone was sort of riffing ideas on, "Well what if we did

7:18 this or tried that?" I remember we were really ambitious at the beginning.

7:21 I remember we designed, what, four cities?

7:25 At one point, Prague, Jerusalem, Chicago, Los Angeles, I believe.

7:32 We had an Arctic base, an underwater base, an icebreaker ship,

7:37 an airplane sequence that crashed into a high rise,

7:42 which we cut right after 9/11.

7:45 We had several Combine bases in the Wasteland.

7:48 We had the Air Exchange,

7:50 which is where they were taking the atmosphere from Earth.

7:53 We had a train depot and so on.

7:56 So we really obviously had to scale back what we were trying to do.

8:00 But that's okay to go a little bit nuts at the beginning.

8:05 I saw it almost as sophomore anxiety maybe, you know?

8:09 Like, "Oh, wow, we have to surpass

8:12 'Half-Life'," and "Half-Life" was important, you know?

8:15 And so it was this unbounded ambition.

8:20 And we kept asking ourselves and Gabe kept asking, "Well,

8:23 how is this 'Half-Life?'" When you have a silent first-person narrator,

8:29 you could tell people this is Gordon and throw

8:32 that character into literally any environment and say,

8:35 "Yeah, this is 'Half-Life,'" but unless the environment

8:38 is telling you that that's the case, it's very arbitrary.

8:42 So we wrestled with that a long time.

8:45 Slowly the sort of grab bag of characters we had,

8:49 we started to feel like there was connections between them.

8:53 And then at some point we realized, "Oh, we can make this out of the science

8:59 team the survivors from 'Black Mesa'." And suddenly,

9:02 for me it was the idea of it's a science team.

9:06 That suddenly, you could put those guys anywhere,

9:09 and because they have a familial interconnectedness and they recognize Gordon,

9:14 that will make you feel like it's a continuation of this story.

9:18 Obviously Alyx was totally new to that universe,

9:22 but she had a father who had been in "Black

9:24 Mesa" and he and the other scientists were friends,

9:29 and they had extended their relationship into this new world that you were in.

9:33 That was kind of the click, was the science team.

9:36 And that was a while till we got there.

9:39 Much like the first one where all

9:41 the contributors were really collectively deciding on the direction,

9:45 is we were really making a push to grow the team and hire new artists.

9:50 Dhabih Eng was an influential early contributor.

9:55 I think he and I were doing character designs for Gordon

9:59 and for Alyx and for some of the antagonist Combine characters.

10:05 At the time, I was looking at a lot of, I mean, really the helmets,

10:10 which I think are kind of one of their defining features,

10:13 was based off some gas masks from Europe that I had seen in some reference.

10:18 I remember doing early concepts of the new HEV suit, like,

10:22 "What's Gordon gonna be now?" And I had kind of just started, and so I was like,

10:27 "I'm gonna make this my own!" It looked I

10:29 guess kind of like the still suits in Dune,

10:31 but black and a lot more belt buckles and latex.

10:35 And Gabe was just kind of like, "What's this?" Like,

10:37 "It's Gordon." I was really proud and excited.

10:40 He's like, "Why is it not orange?" (Dhabih

10:43 chuckles) Or something to that effect of essentially,

10:47 "Why are you giving up this iconic part of the character."- People love it.

10:51 [Dhabih] Yeah, people love that, so what are you doing?

10:53 And it's like, I kind of was like, "Yeah,

10:55 I guess he's right." So we kind of went back

10:58 to the drawing board and ended up with what shipped.

11:01 We had, I don't know, so many creatures that we tried and abandoned, you know?

11:06 Not because they were bad, because we had better things to pursue.

11:11 Like there were the Sackticks, the Cremator, the Crab Synth, the Stalkers.

11:16 And even stuff, like the Stalkers we had a little bit in the game, right?

11:19 [David S] Yeah, we shipped the Stalkers,

11:20 but not as an adversary that you fought.

11:22 There was definitely an evolution of how biological versus how

11:28 synthetic a lot of the creatures in "Half-Life 2" were.

11:32 And I was always hoping to get to a space where it was ambiguous, you know?

11:36 Where you couldn't quite tell if this was created or if

11:39 it had grown or if it was natural or synthetic.

11:41 There was some Synths I did that never ship.

11:44 Actually, we, we did ship 'em.

11:45 They were at the end of the game when you're going through the Citadel.

11:47 I had designed these Synths that no animator could figure

11:51 out how to actually move 'cause they weren't built to move.

11:55 And much like some of the environments that we've

11:58 seen released in the leak that were very dark,

12:01 some of the early "Half-Life" stuff was also very dark.

12:04 But at the time, video games were very dark.

12:07 In the interim before we hired Fletcher and before Viktor was there,

12:11 'cause it was the old environments, right?

12:13 I don't remember how it came about that they built this level,

12:16 but the idea was there was some looting going on and the idea

12:20 was you're in this little alley and you see a citizen,

12:23 you hear a voice saying-- "TV!

12:25 Free TV!"- They just got into the TV

12:29 store and the citizens and the cops are fighting,

12:32 and you see a citizen run past you holding a TV set and he goes, "Get one!

12:36 Get a TV!

12:37 Get yourself a TV!" And then he runs off down the thing and you

12:40 go in another courtyard and there's a cop with a night stick and a citizen,

12:45 and they're having a fist fight.

12:47 And it's interesting 'cause the NPCs are negotiating their bounding boxes

12:52 and there's some animation like how difficult it is to contact.

12:57 We captured everything.

12:58 We took pictures of whiteboard design sessions.

13:01 There were really crude drawings and ideas.

13:04 Some of the stuff from the very beginning made

13:06 it all the way to the end, you know?

13:08 Ravenholm, Antlions, Dog, Kleiner's pet Headcrab.

13:14 Those things were just so fun in the moment

13:16 that how could they not survive, right?

13:18 The other thing that was really memorable is

13:20 I remember we started making the world really dystopian,

13:23 really dark at the beginning, and some of the concept art shows that.

13:27 We didn't think players would enjoy spending 20 to 40

13:29 hours in such a dark and a kind of oppressive setting,

13:34 so the game lightened up considerably over time.

13:36 But still maintain the feeling of a security state.

13:41 So I was trying to get all these cool ideas

13:45 that the team had already when I arrived and streamline them.

13:50 Instantly I think we were friends with Marc Laidlaw,

13:54 we had a common sort of vision and fantasy to create Orwell-like city.

14:01 And then slowly, I brought context behind that, the idea

14:06 that we can place this in an anonymous Central Eastern European city,

14:11 which is a perfect setting because in these countries

14:14 there were so many waves of invasions.

14:18 We had the Ottoman Empire, we had the Russians, and now why not the aliens?

14:23 And one of the first maps I remember

14:25 working on was they called Test Room Standards.

14:28 In "Half-Life," every level designer kind of had their own idea of how big

14:33 stairs were or how big steps were gonna be or how big a button was.

14:36 Make sure we had some consistency.

14:38 Another thing that was sort of important was just getting more

14:43 people to go on field trips and going to do research,

14:46 'cause I think a lot of the level

14:48 designers on "Half-Life 1" just were building very abstractly,

14:51 and so we wanted to have some more

14:54 grounding in reality and have more realistic proportions.

14:58 And also just get, textures were going from hand painted to more photographic,

15:03 and so we would go out and collect a lot

15:06 of the pictures for those textures to start building for our new library.

15:10 I joined "Half-Life 2" probably a year after it had started.

15:16 I know that it was at least a year because I was put on Borealis,

15:19 which everybody else had disowned.

15:23 But then I came along and they said, "Oh, well,

15:25 I guess this is now your problem." So I started on Borealis,

15:28 and of course I failed at that too.

15:31 That eventually got cut.

15:33 Yeah, so I came in having had experience doing game AI.

15:37 I was at Looking Glass Studios,

15:40 probably the biggest thing I did there was lead programmer on "Thief:

15:42 The Dark Project," the AI for that too.

15:45 And actually my first thing is they handed "Borealis" to me.

15:50 After I'd finished?

15:52 [Tom] Yeah, to do the AI for the guy

15:56 that you're supposed to walk through the Borealis ship with.

15:58 There was a bunch of work put into it.

16:00 We even went on an icebreaker ship to get reference.

16:04 As soon as we started to put it together,

16:06 we realized how unbelievably tight the space was and you

16:10 could never make any kind of really interesting combat in it.

16:13 And so it just withered on the vine right there.

16:16 I mean, when I started there was that document that was the tech wishlist.

16:20 Yes.

16:20 It was just this very bare bones bullet list of tech features.

16:24 And I mean, I think it was six years of work to build that tech list

16:28 and then realize it into a product

16:30 and leverage all the implications of it, you know?

16:33 So this was a main feature, that the light felt very,

16:37 very realistic and intuitive because of the Source Engine and the work,

16:43 the collaboration between artists and engineers.

16:46 Ken Birdwell, he was a fan about photography and getting the lighting right.

16:52 The math that we were using was wrong.

16:55 And not only that, the math that everybody was using was wrong.

16:59 And then as I started to correct it and I realized just how

17:02 bad it was and then I fixed it and suddenly everything looked great,

17:06 I had to go tell the hardware guys, the people who made hardware accelerators,

17:11 that fundamentally the math was wrong on their cards.

17:14 That took about two and a half years.

17:17 I could not convince the guys.

17:19 Finally we hired Gary McTaggart and Charlie Brown,

17:22 and those guys had enough pull and enough- I have a fine arts major,

17:26 nobody's gonna listen to me.

17:28 Charlie and I met at college at the University of Florida.

17:31 We were both getting computer science degrees, yes?

17:34 I ended up dropping out to go work at 3dfx as employee 11 there.

17:40 And then I think Charlie, you actually graduated, didn't you?

17:43 I graduated, then I followed you.

17:44 And I was 17 there.

17:46 Yeah, yeah.

17:46 I was 17 there.

17:47 There were three kind of principles.

17:50 One was to make something that was immersive,

17:53 visually really rich and appealing, something more like you would see in a film,

17:58 and the art direction and the technology to make that happen.

18:02 Second one was to have characters that you cared about.

18:05 So basically that could react, emote,

18:09 that mattered in the world instead of just being signposts.

18:12 Third was to have a responsive environment with physics that mattered.

18:17 I think that I maybe more than other people realized

18:20 just how big of a task we were setting for ourselves.

18:24 So I was a little bit daunted by that.

18:27 We felt the pressure from Gabe actually

18:29 to double down and make something even grander than

18:32 "Half-Life 1." Therein lies the sort of little

18:34 bit of the trap of "Half-Life 2's" development story,

18:36 which went from being a pretty quick,

18:38 "Let's figure out how to do this" to a six-year slog.

18:42 [Tom] Yeah, and it was six, seven days a week, 10 to 14 hours a day.

18:46 [Ken] "Yeah, we're gonna cut your feature.

18:47 Sorry.

18:47 It's just not worth it."- [David S] So I ended

18:50 up being really the point person working with the FBI.

18:53 [Charlie] I remember coming in that morning and figuring it out,

18:55 and I just felt like someone had just punched me in the gut.

18:58 [Robin] What's this mean?

18:59 Can we still keep working on the game?

19:02 Is someone just gonna release the whole game tomorrow?

19:04 [Jay] I definitely was not shielded from that.

19:07 [Scott Lynch] I remember one conversation with Gabe,

19:09 which he was like, "So how fucked are we?" "Yeah,

19:12 we're kind of screwed."- [Gabe] The company was pretty close to going bankrupt.

19:16 I was pretty close to going personally bankrupt.

19:17 It's like, there was no money left.

19:20 [Karl] I can't think of another case when that happened.

19:23 Yeah, a real bullshit move.

19:25 [Marc] Are we allowed to cry in this documentary?

19:27 That's what I wanna know.

19:28 (tense music fades out) (gentle music) (gentle music continues)- I

19:43 remember we shipped "Half Life 1" and the world loved it,

19:47 and we were just super happy.

19:50 Like, "Yeah, we did this thing!" And we had a big meeting,

19:53 and Gabe looks at me and goes like, "So all the animations are terrible.

19:57 How are you gonna make 'Half-Life 2' better?"- In "Half-Life

20:02 1," we made what were at the time significant improvements,

20:05 but still they were fairly robotic and mechanical.

20:09 And so we had to go and do a huge amount of work on the animation system

20:12 and on the facial animation system and coming up

20:15 with authoring tools that allowed you to do it.

20:17 Well.

20:19 End of the line,- A lot of this was level designer design,

20:24 like figuring out how to build atmosphere.

20:26 We knew that we wanted to start you off seeing the Citadel right away and we

20:31 wanted to introduce you to the city

20:33 with this great voice acting from Robert Culp.

20:38 We wanted to finally do real characters.

20:40 We knew that from "Half-Life 1",

20:42 the way that players would respond to having any character

20:46 talk to them was way beyond the reaction that we expected.

20:51 When we figured this out in play testing,

20:53 if Barney would follow you around and shoot things,

20:56 suddenly you just love this guy and you

20:58 love the scientist and you love these characters.

21:00 So we knew we wanted to just bring that way

21:04 up to the next levels beyond what we could imagine.

21:08 (loud explosion) (frantic screaming)- And the theory was if

21:13 we could make a character who could act realistically,

21:17 that that would help the player make an emotional

21:20 connection to the other characters in the game.

21:23 And at the time we decided we were gonna

21:25 do that, not only did none of the technology work,

21:28 but I would go talk to industry professionals and they

21:30 would tell me not to bother 'cause you couldn't do it.

21:33 Luckily found a research work done in the '70s by Dr.

21:38 Paul Ekman, and he had just broken down how the face

21:42 works and how humans make expressions and he had this incredibly

21:47 wonderful formal clinical description of what a face does

21:52 and the sorts of things that normal people do with their face.

21:56 I tried this experiment with putting a grid

21:59 on Scott Lynch's face and shooting him from the front

22:03 and the side so that we could get points

22:06 from this grid in three space for the geometry.

22:08 And we got shots of his texture and we found a good way

22:11 to model a reasonable head of his geometry and then project the texture onto it,

22:18 and then abstract it or stylize it

22:20 enough so that it didn't look totally photographic,

22:23 but it looked like kind of a real person.

22:28 Occasionally when somebody hears the history, it will generate a light bulb,

22:33 but no, no, I don't get accosted.

22:36 [Producer] No one's asking you for beers or anything?

22:38 No, no.

22:39 Or buying me beers.

22:40 (both laughing)- We added Ken's eye tech to that, which was amazing because

22:47 he really modeled and he went deep into the research of how eyes work.

22:51 Eyeballs was about nine months.

22:53 Just on eyeballs.

22:55 (Ken chuckles) But you have to do each step.

22:58 You have to get the shape of the eye right, you have to get how the eye lid gets

23:03 pushed by the cornea and the ball of the eye.

23:05 So as your eyes move around, your lower lid, your upper lid,

23:10 all of those get sh shifted slightly and it's all automatic.

23:14 It was wonderful.

23:15 The day I got it working,

23:17 the characters went from looking like interesting dolls to, (gasps),

23:22 "Alyx just looked at me!

23:23 Oh my god, this is working."- They would ask for surprising things at times,

23:28 like they wanted really fine control over

23:30 looking at named objects in the 3D world,

23:33 no matter what direction the actor is facing.

23:35 So you had to figure out things about, "Oh,

23:37 maybe the upper body needs to turn and the head needs to turn and the eyes

23:41 need to turn."- So we're in the elevator

23:43 coming back from Starbucks to get morning coffee,

23:45 and we saw this face and it's like, "That's him,

23:48 that's the scientist!" And he was an accountant

23:51 in the firm below ours in Kirkland.

23:54 Yeah, definitely, and kind of blows your mind

23:56 when you've been working all day on this character

23:58 in your game and all of a sudden you get on the elevator and you're like,

24:02 "Holy crap, it's him!"- I once ran into him at the airport.

24:06 So we cast people that were real change vendors in front of supermarkets,

24:12 we'd cast waiters in restaurants.

24:14 [Ken] When I was in junior high school,

24:15 I took karate classes and that was my old karate instructor.

24:20 Randy Lundeen said, "Hey,

24:21 there is a guy holding a sign for work at this exit on 520 every day who has

24:27 this great face." And he took a picture

24:29 of his face and he stop and asked if he could.

24:31 And so it's like, "Oh, this is great.

24:33 I think this guy's Eli." We brought him

24:36 in and paid him a couple hundred dollars,

24:39 which is what we were paying people back then, and so that's where he came from.

24:43 Erdin Grcic was a Bosnian friend that I knew from school.

24:48 He was like, "We want the East European

24:50 contingent and stuff too," so it was pretty interesting.

24:54 We wanted anybody to feel a part

24:56 of the world when they're playing it as a player,

24:59 and so we found people from pretty much every skin color,

25:05 different ages, male and female, equal numbers of males and females.

25:09 Sadly for them, they all had to share the same voice actor.

25:11 (all laughing)- Sadly for us too!

25:15 We dropped a crate of rockets coming across the plaza.

25:17 We hoped for one more fighter to help storm that barricade.

25:19 We never dreamed it would be you.

25:22 I mean, Kleiner's Lab was the first prototype that we

25:25 did of how characters could work together in a scene.

25:29 Yeah, it was like a proof of concept.

25:30 We kind of had to make that work and put everything

25:34 we could into it at the time to test all the technology,

25:37 and then know from there what can we do, what else can we do with this.

25:42 Turning it into a performance that would be engaging

25:46 rather than just the audio radio play kind of thing.

25:49 Yeah, I mean that was the point where,

25:50 that scene was where it all started to come together,

25:53 all our theories and hopes for what the experience might be.

25:56 And it all went completely smoothly outta the box, it was amazing.

26:00 The most difficult technical aspects are really kind of boring.

26:04 They're the, "Okay, the character is running through AI

26:07 and now they need to jump in and hug another character.

26:13 Now they need to pick up something from the desk,

26:15 and that thing on the desk can move and where the character

26:17 is can move and where the player is can move and everything about

26:21 this whole scene can move dynamically." And to get to that stage

26:25 where it just looks right is

26:26 this huge technical challenge and artistic challenge.

26:29 But once it's done, it's just amazing.

26:38 Gordon?

26:39 Gordon snap out of it.

26:42 You're staring at me again.

26:45 [Kleiner] Ah, warming up nicely.

26:47 Are you sure you don't want me to swap out the polarizer?

26:51 I remember when we got Fletcher on and he would kind of teach us about

26:57 how they did things at Disney and they

26:59 would bring in acting coaches for the animators.

27:02 What does sarcasm look like?

27:04 Certain expressions, and there's a timing to those expressions.

27:08 When you have a scene with multiple characters,

27:11 it's not always the person talking that's doing the acting, right?

27:15 Everybody's acting.

27:16 Even though you may be looking at Kleiner because he's the one talking,

27:20 every character in that scene around you,

27:21 behind you, they're all doing stuff because not

27:23 everybody's gonna be looking at the same place.

27:26 Yeah, I remember even just like when Alyx teleports

27:28 and then you see her show up on the screen, right?

27:31 Oh, that was incredible.

27:33 Heads over, gives him a quick kiss.

27:36 It tells you so much about the relationship without having to say,

27:39 "Hi, dad, I love you." And all this extra heavy-handed stuff.

27:45 Jake was a random demo reel that was sent in.

27:48 It was incredible.

27:51 Jake was incredible.

27:53 Jake was incredible.

27:54 Miss that guy.

27:55 One of a kind.

27:56 Yeah, miss that guy.

27:57 Amazing technical animator.

28:00 [Matt] Yeah, I mean Jake was one of those guys that, whatever needed to be done,

28:05 he would figure it out and just make it happen.

28:09 [Douglas] It'll just get done correctly.

28:11 [Dhabih] I think also the context of when all this was taking place,

28:14 like the problems that he was helping

28:17 solve were just things that weren't being done.

28:20 Even something as simple as passing someone an object

28:23 between two characters sounds super simple but actually really,

28:28 really complex, especially back then.

28:31 And so he really helped push how do

28:34 we even do this in collaboration with the programmers.

28:39 Dhabih wanted to do a goo thing, so I worked on the shader for that.

28:44 I did the stuff for rendering that and all

28:46 the little monitors everywhere that showed other scenes.

28:50 Did the refraction shaders and the supporting code for that.

28:54 The shaders didn't really have conditionals or loops or anything like

28:58 that, so we would unroll it all and compile a bazillion shaders.

29:03 It got to the point where it was so many

29:05 shaders and it took so long to compile that we,

29:08 I think Mike Dussault implemented distributed building system.

29:12 Everyone's machine, the company was running

29:13 something that could sit there and compile

29:15 shaders all the time or do lighting for the levels or whatever.

29:21 [Dhabih] Any opportunity to kind of add

29:22 to the visual storytelling I think was...

29:25 Yeah, and I think we hid the first G-Man sighting there.

29:29 [Dhabih] That's right, he was one of the monitor.

29:31 It would flip to him.

29:32 Yeah.

29:33 Yeah.

29:33 And then he'd just walk out of frame.

29:35 The moment where you see him makes you realize there must be

29:40 a lot of moments where you didn't know he was watching you.

29:43 He's not Big Brother, that doesn't make sense to me that he would be

29:47 a Big Brother because that's sort of industrial and bureaucratic,

29:52 and this is much more intimate.

29:54 He's always over your shoulder, watching, waiting for the moment to speak.

30:08 That's where I think he lives.

30:11 My name is Mike Shapiro and I am G-Man.

30:15 I am also Barney.

30:17 And Barney's, he's not that far from- "Gee,

30:22 Gordon!" He's not that far from where I live.

30:25 He's so fallible and he's always tripping over himself,

30:29 and he's just such a good guy, you know?

30:32 Not necessarily the most capable, but just really there for you.

30:38 Get that thing away from me!

30:40 Here, my pet.

30:41 Hop up!

30:42 I think early on, I had a sense that G-Man knew a lot

30:46 and was taking care of a lot that we didn't quite understand,

30:49 was kind of enjoying fucking with people and massaging reality

30:53 in a way that he had a unique capacity to do.

30:57 There were also aspects to him that I knew pretty early on, like

31:00 his relationship to time is very different than you or I would think.

31:06 In my mind, he could literally be in two places at once.

31:09 And so sometimes there was a kind of an implied hitch in his timing,

31:15 or he's experiencing two or three different

31:18 moments at once and that might be funny

31:21 for a reason that you don't know 'cause you're

31:23 only in one time with him in a moment.

31:25 The right man in the wrong place can make all the difference in the world.

31:32 [Marc] I think as we realized how well she

31:34 worked coming out of stuff like the Kleiner's lab experience,

31:37 we started looking for more and more ways to include her in stuff.

31:42 And of course she really came to life when we found Merle Dandridge.

31:46 Yeah.

31:46 We could feel her a already as a partner in creating this character,

31:49 I think, coming out of that.

31:51 She's just got a creative relationship with this character

31:54 and she's gonna bring a lot to it.

31:55 [Bill V] I feel like she was really able

31:57 to bring a warmth and humor into her deliveries,

31:59 too, that other people, it was more like a soldier in a game a little bit.

32:05 I mean, she did this great job of bringing a character warmth into it.

32:11 And now we're sitting ducks unless we can get this thing running.

32:14 Come on, Dr.

32:15 Kleiner.

32:16 Is it gonna work or not?

32:18 Now, now.

32:18 There's nothing to be nervous about.

32:20 Let's see.

32:20 The massless field flux should self-limit.

32:24 My name is Harry Robbins, I'm known as Hal Robbins.

32:28 I have a stage name of Dr.

32:31 Hal.

32:32 My only doctorate is in theology from the Church of the SubGenius,

32:36 which I'm involved in.

32:38 Well, I think perhaps Marc was familiar with the things I do

32:42 on the radio and he thought my voice would be good for it.

32:46 I have a monthly Ask Dr.

32:48 Hal show on Twitch and Zoom, and I am also on a couple of radio shows.

32:55 Well, I'm a graphic artist.

32:56 I've worked for DC Comics for their Paradox Press line.

33:01 Well, I have some patrons who I do painting for.

33:05 I've painted many things for them.

33:07 I played Dr.

33:09 Isaac Kleiner after I did the voice of a number of different scientists.

33:14 I try to pronounce all the words.

33:19 I make sure to aspirate consonants, for example.

33:23 I don't say "didn't" the way most people do, even newscasters today.

33:27 I always make sure to say "did-n't." Perhaps it sounds archaic that way.

33:32 Marc wrote archaic expressions into my dialogue,

33:37 and I would say things like, "Oh, fi" and so forth.

33:41 No, not up there!

33:42 No, no!

33:44 Careful, Lamar!

33:45 Those are quite fragile!

33:47 Oh, fi!

33:48 It'll be another week before I can coax her out there.

33:52 Yeah.

33:52 Longer if we're lucky.

33:54 Barney!

33:55 (chuckles) You're not an animal person.

33:56 (Barney grimaces)- It was always pleasant.

33:59 I was always well treated.

34:01 They were always patient, even if I mispronounced something which was obvious,

34:05 they would correct me and I would pay attention.

34:08 And I always try to be as close as possible

34:11 to a one-take wonder so they don't have to do it again.

34:14 But that enabled them to do it many times.

34:16 And so they had a lot of things which arose from that, and some

34:21 of these are the things I'm asked to say by fans,

34:24 which are inexplicable to people who don't know the game.

34:28 Like, "You want this guy to say, 'Have you seen my coffee cup?' 'And why do we

34:32 all have to wear these ridiculous ties?'" (Hal chuckles) (compelling

34:40 music)- So in the same way that we thought it

34:52 was important in "Half-Life 1" that when you shot a wall,

34:55 that we put decals on the wall to indicate that you shot it, we wanted to say,

35:00 "Okay, what happens when you throw a grenade at a bunch of boxes," right?

35:03 The world itself needed to be more reactive to everything you did.

35:07 We used some third party code, but Jay Stelly was the one who really

35:10 wrote the wrapper on that, that integrated it.

35:13 And physics was, even earlier in the project, was pretty successful, right?

35:17 Once we got over the initial thing of things

35:20 falling over and we started adding breakable objects,

35:23 things you could script in hammer.

35:26 We added vehicles, sound cues from all the objects and the AI,

35:30 so you could drop things on them and crush them and things like that.

35:34 All of those behaviors,

35:36 once we got just a few of those things in, that was really fertile.

35:41 For everyone who worked on "Half-Life 2," masses got broken down

35:44 into a set of scales that I assume someone's already talked about.

35:48 No?

35:48 Okay.

35:49 Well, you see!

35:50 So, yeah.

35:51 All right.

35:51 So the question is whether it's bigger than

35:53 a bread box or smaller than a space station, right?

35:57 When you want to debug something and the text says,

36:00 "Oh, this thing weighs as much as a racehorse-"- Yeah,

36:04 you knew it was more than a bread box and not as much as a space station.

36:07 "I think so, but I don't know in the scale

36:09 if it's closer to this or closer to that." A refrigerator.

36:13 Yeah, a refrigerator.

36:14 This all came from at some point, I think it was David Speyrer,

36:18 so you could look at any entity in the game

36:20 and turn on a bunch of debugging text about it.

36:22 And one of the most important things to get right with every

36:25 object in the game that was physically simulated was its mass,

36:28 'cause that dictated an enormous amount of things,

36:29 like the amount of force it would apply to something else when it impact,

36:34 when it hit them, its response when it's hit by something else.

36:36 All that stuff, had to get it right.

36:39 But of course, masses is just a number.

36:41 What does this mean to anyone?

36:42 And so I think it was David Speyrer

36:45 who found online some list of masses and objects,

36:49 and he just brought that into the game so that when you looked at an object,

36:53 it'd say like, "100 [racehorse]" or something.

36:57 And so over time, you all just got that locked in.

37:00 And because it was a logarithmic scale, it went from things like bread box,

37:05 paint can or something up to like space stations.

37:08 Yeah, the moon.

37:09 The moon!

37:09 You're like, "What?

37:10 What do we do with this?" At this point, it's just key frame at that point.

37:14 Yeah, so there would be many conversations

37:16 where we'd say something like, "I don't know.

37:17 Do you think that's more like a racehorse

37:19 or the space station?" (metal creaking) (water splashes)- The very

37:24 first physics puzzle that we put in the game

37:26 from a development calendar standpoint was the washing machine puzzle,

37:31 where you push the washing machine into the hopper

37:33 and raise the ramp for the airboat to jump over.

37:35 We developed that because we felt like we had

37:38 this section of relentless pressure with the attack helicopter.

37:42 Watching players play test, we felt like they needed relief from it.

37:46 You see the very first training of it

37:48 in the on-foot canals with the teeter-totter

37:52 that you have to put the cinder blocks on, which was added to the game later,

37:58 and it's sort of training of like, "Hey,

38:00 you can interact with levers and balance

38:03 and create paths for yourself." And then, yeah, later the buoyancy puzzle,

38:08 we added that one after the washing machine puzzle.

38:11 And it was just kind of like, "Oh, you can connect masses with invisible pulley

38:16 systems and then build little machines that way.

38:21 (barrel thuds on the ground)- I guess at first, right,

38:24 the gravity gun was kind of this, we weren't

38:26 even sure we were gonna ship it in the game.

38:28 It was like a development tool.

38:29 And so it was this beam,

38:30 you could reach out and grab things and pull 'em toward you or reposition 'em,

38:35 drop 'em, grab 'em from another point.

38:37 But then we talked about simplifying it and just making it

38:40 to where it was just pushing on things and pulling on things,

38:43 and where could we go with that.

38:45 [David S] And the first draft of Ravenholm

38:47 wasn't really about physics so deeply, right?

38:49 We ended up reorganizing the game to put

38:53 Ravenholm after you get the gravity gun.

38:56 Originally it was before you would park the airboat at the dock

39:00 of Ravenholm and then get out and go explore Ravenholm.

39:03 (eerie music)- We wanted a horror area,

39:12 since in a lot of ways "Half-Life" was a horror game and the horror was kind

39:17 of leaching out of "Half-Life 2." So we

39:19 at least went in one environment that was just scary.

39:22 And then we had a design meeting where Doug Wood said,

39:27 "Preacher with a shotgun." And it was like, "Oh,

39:30 yes!"- The important thing to know is we didn't sit

39:33 down and design Ravenholm as the tutorial for all those things.

39:38 It kind of kept getting things over time.

39:41 And then after we got more physics working, we had this idea that there would be

39:48 traps for the zombies in there with physics.

39:50 And there were a bunch of 'em, and they didn't all make it into the final game.

39:55 And then some of them did work though, and we had to build a system for cutting

40:00 the zombies in half once we added the propeller traps,

40:03 and we had to make 'em the right height so you

40:05 could crawl under 'em or the zombies could crawl under 'em.

40:08 And then someone had the idea, "Well,

40:09 what if the top half can keep going and it can

40:11 get you under there?" We tried that and that was great.

40:14 Someone had the idea of a saw blade,

40:16 and Ravenholm already existed and we ended up,

40:20 based on this behavior happening in another level,

40:22 built a saw blade and went and put it back in Ravenholm.

40:25 Ravenholm was probably some of the most fun I had working on "Half-Life 2,"

40:29 the whole idea behind Ravenholm and that we were going to starve you of ammo,

40:33 sort of force them in a position where they have

40:35 to rely on physics and the context of the objects around them.

40:39 It's like, "Oh, wouldn't it be awesome if this zombie,

40:41 it's gonna take him 30 more seconds to get to me,

40:43 but he's got a barrel right there.

40:45 What if he could swat that at me?"

40:46 So that was definitely something that was just like, "We have physics now.

40:50 This would be a cool thing to do with it

40:52 and it'd be a thing that would make sense and a thing

40:54 that would be scary and fun."- [Tom] One of things

40:57 that John was extremely strong at was little touches like,

41:00 I think we called them puzzlets.

41:03 We did call them puzzlets.

41:04 They weren't puzzles, they were puzzlets.

41:06 Yeah, it's this little thing that takes

41:07 you between five and 60 seconds to resolve,

41:09 and at the end you either get something or you feel a certain way.

41:12 And it was really a big part of making sure that minute-to-minute gameplay,

41:15 as we called it, was interesting.

41:17 Yeah, he was good at that.

41:19 (monster yells out) (gun emits whirring sound) (loud gunshot)-

41:23 We actually did this as an exercise of like, "Okay,

41:25 take all the elements in the game and have a grid of rows and columns of all

41:32 those elements and make sure that there's

41:34 a meaningful thing at each intersection of those cells,

41:38 so that everything has a meaningful interaction

41:40 with everything else." And where we saw empty cells,

41:43 we would actually design a thing.

41:45 [Jay] Yeah, like, "Is there something we could do there?"- Yeah,

41:47 barnacles plus grenades or whatever, you know?

41:49 "Okay, we need to make sure that they'll eat the grenade

41:51 and then blow up."- The reason we build these scripting level design tools,

41:56 it makes it accessible for all the people on the team if you can't write code,

42:01 to like put something together, try out an idea, see what we can do

42:06 with this, and the entity system's no different, right?

42:09 A lot of things that level designers just wanna like,

42:11 "I have this idea for something that could happen.

42:13 Is that fun?"- Look to your own salvation!

42:18 (Father Grigori laughing maniacally) (loud gunshot)- Yeah,

42:25 I was on "TF 2" at that time and I don't remember exactly,

42:30 I think it was during 2000 maybe?

42:32 Someone else has probably told you this and got the dates actually right.

42:36 But at some point we all jumped from "TF 2" over to "Half-Life

42:40 2." I feel like that marks the sort of, in some ways in retrospect,

42:45 the end of pre-production really of "Half-Life 2" and it becomes now,

42:50 "Let's build this thing and ship it." (exciting

42:54 music) (helicopter whirring) (exciting music) (helicopter whirring)- Actually,

43:05 the first thing I got told to work on was a jet ski,

43:08 and I spent three weeks trying to figure out a jet ski in first-person.

43:12 I was like, "Okay, none of this is working.

43:14 I don't see how you do this without making

43:17 people sick." So I remember one weekend out of frustration,

43:20 I just made the airboat.

43:22 I don't know, maybe it's the Florida bit in me.

43:24 So Charlie Brown had worked on the airboat,

43:27 and then we picked it up a different group.

43:30 We started working on the canals.

43:32 And we had this idea of it was gonna

43:34 be kind of like a lot of tricks under pressure,

43:37 where you're doing skateboarding-type moves.

43:41 I think it was Kelly Bailey who basically said,

43:43 "I'm playing the canals and I'm trying to do these tricks,

43:48 and I'm failing, and then I'm getting shot." And he's like,

43:51 "I wanna be like 'Smokey and the Bandit.' I just wanna be

43:54 smashing my way out." We did this whole other pass of the canals,

43:57 where we dialed down the consequences for failing

44:00 the tricks and we dialed up the feeling of power.

44:05 Some players absolutely refused to really stay in their airboats.

44:11 They wanted to get out and look under every rock, behind every tree.

44:15 For those players, we had to craft enough density through

44:18 the lengths of the canals to keep them entertained and happy.

44:22 And then there were people who just wanted to smash

44:25 through everything and treat it like a racing sim.

44:27 And so the game had to be fun and interesting for them,

44:30 and sometimes that would clash.

44:31 If you stop here, we're going to give you a reward, basically.

44:35 Either a narrative reward or an actual in-game reward.

44:42 The lambda caches, the lambda designs, we use them a lot more in the canals,

44:48 and part of that is literally because we did it afterwards.

44:51 I mean, I had worked on on coast maps and we didn't have that yet.

44:57 The canals are restrictive 'cause we want you to drive this airboat

45:00 through them and we've got all these fun things to do,

45:02 like you could jump over things, you could smash through things.

45:05 The worst case scenario is you're trapped with people who have guns and they're

45:10 shooting at you and you don't necessarily have a way of shooting back.

45:15 So that led to, "Well, we need to add a weapon to the vehicles."- Well,

45:22 part of it was the helicopter was such a persistent pain in your ass that we

45:26 really wanted to put a bow

45:28 on that whole experience and give you some vindication,

45:31 some vengeance, where you get to now hunt the helicopter.

45:35 So as soon as we attach it,

45:37 one thing we really intentionally did is you start to drive down

45:41 the tunnel and the helicopter pops down in your face and it's blocking you,

45:45 and you have no choice but to just shoot it if you want it to go away.

45:48 And then you see it flee.

45:50 And so it's like, "Oh, okay." The players could connect the dots, right?

45:53 "Now it's my turn." So the next sequence is all about

45:56 chasing the helicopter down to that final battle at the dam.

46:00 (loud automatic gunfire) I remember we worked pretty intentionally

46:06 to make it the ultimate showdown between you and the helicopter,

46:09 and some of it was by accident.

46:11 I remember Brian Jacobson, who did a lot of the code for that, at one point it

46:16 accidentally did this thing where it just spit out a ton of bombs.

46:21 And we're like, "That is awesome!

46:23 Keep that!

46:24 Make that part of the fight." We tried to make

46:28 it as action intensive as we could within reason, right?

46:31 (loud gunfire) That battle was actually,

46:34 in a way we designed it to be a final exam.

46:37 We unraveled those elements into the levels beforehand, right?

46:42 You learned how to dodge bombs, you learned how to shoot at it,

46:45 and you learned how to shoot while driving.

46:47 And then we throw those things together at you with a little bit of a twist.

46:52 Canals was sort of like a crucible because a lot of good

46:54 things came out of it 'cause it was such a difficult design problem.

46:58 The easy thing to do is to make a punishing design, where you just say,

47:02 "You can't go here and we're gonna shoot you and force

47:05 you to fail and fail and fail until you get

47:07 to see the next part of the game." It's much

47:09 harder to create something where the player feels there's an evolution,

47:13 they have some power and some agency, so that their skill is being rewarded

47:18 and they can be creative in that restrictive environment.

47:22 (loud crashing) (chatter over radio) (vehicle revving) (vehicle revving)- So

47:37 basically had a whole car engine down to through transmission,

47:40 down into gearing and differentials and turning the tires,

47:43 but then it turned out, "Oh, we can really only go 15,

47:45 20 miles an hour," otherwise you would blast through

47:48 the extents of the map in about 10 seconds.

47:50 which wasn't gonna do anybody any good.

47:52 People made jumps and made all these things, nothing was working.

47:55 So I was like, "Oh, you must be doing a jump.

47:58 I'm just gonna turn off gravity on the car for a little while

48:00 and let it just go and then gradually ramp it back in so that it

48:04 actually felt like you were jumping."- It was interesting how like the dynamics

48:08 of the buggy was so much about

48:10 the perceived experience versus how it actually performed.

48:15 It just sounds like it's going faster than it is.

48:17 So Jay and I worked really hard on these engine

48:20 sounds and this fake transmission and doing all this stuff,

48:23 but you're really going like 30 miles an hour maybe.

48:27 So it's like someone just driving around in first gear.

48:29 Got pulled into coast, one,

48:30 'cause I was interested in working on more game stuff, but two,

48:33 it was the biggest I guess group of people utilizing the displacement

48:37 tech since it was more of a terrany thing and outdoors,

48:40 and we were trying to figure out how we wanted to use the displacements.

48:44 Meaning we knew we wanted the things to look more natural.

48:47 There was a lot more geometry in those maps than our previous ones,

48:50 given just displacement in general.

48:52 We were always struggling with performance and trying to figure

48:55 out how to make things a little bit faster, easier,

48:57 or where we could make things do what we wanted them to do,

49:00 but not necessarily cost as much as they cost.

49:04 It's much faster than walking speed,

49:05 so you just signed up for building a huge amount of game.

49:09 People can just stop and get out anywhere.

49:11 So now you have this trade off of density versus distance.

49:17 We got a vehicle we wanna drive for a while, so make those levels long.

49:20 But from "Half-Life 1," we still had the philosophy of, "There should be

49:24 something cool around every corner." Not like a monster popping out of you,

49:28 but different reasons to be engaged in the world.

49:30 So they would dot the coastline with little

49:33 towns and you knew if you went exploring,

49:35 you'd always get some kind of reward for that.

49:37 It might be a health pack and ammo or it might be a character interaction.

49:42 So that would let the player set their own pace and feel

49:46 like they get to explore the world and they're always rewarded for that.

49:51 But if they wanted to drive straight through, they could.

49:54 I worked on the area where the roller mines first

49:57 come in and you're going along a curving path and then

50:02 they blow up all the rocks and drop stuff

50:04 in front of your path to try and slow you down,

50:06 and then you've got like a little house fight.

50:08 There were cases early on where people would

50:09 just try to drive past everything and be like,

50:12 "Why would I stop at this if I've got a car?"

50:14 And so we really wanted them to be fairly unavoidable,

50:17 and so you'd have to get out and deal with taking 'em off your car,

50:19 and then you'd be in the space.

50:21 So it was really fun just jumping in there

50:24 with the Combine soldiers and crafting little fight scenarios.

50:29 What makes AI fights interesting is not

50:31 that they're the smartest or the best or anything.

50:34 It's easy to make AI that can just perfectly shoot the player and kill them.

50:37 That's not what makes them smart.

50:39 It's putting on a show that makes them feel interesting.

50:42 So in that fight in particular, you go up inside of a building and you get

50:46 to the second floor and you're fighting guys in there,

50:47 and they kind of pull you up there,

50:48 there's guys up in there and as you go up and fight them,

50:51 that happens, but then more guys spawn outside.

50:53 There's a bunch of windows up in that top

50:55 section and there's a crate sort of near the window,

50:57 and when you go over to get that crate,

50:58 they start shooting through the window at you.

51:00 And it's not that the AI actually knows that you're there

51:03 and starts shooting at the window the moment they see you,

51:05 we have these things called bullseyes that you could place in spots

51:10 and they basically would act like a target that the AI wanted to shoot.

51:14 A lot of that isn't actually the AI doing this carefully crafted thing,

51:18 it's more of us thinking,

51:19 "What makes this a neat presentation for the player and gives 'em that sense

51:23 that they're kind of, they need to hunker

51:25 down?" Like later on in the same coastline,

51:28 there's a sequence where you're fighting a gunship and you're in the lighthouse.

51:33 And as you go up the lighthouse,

51:34 it's sort of predicting where you are and shooting through the windows

51:37 right in front of you and giving you that same sort of show,

51:39 and that's all bullseyes and things.

51:41 Almost every game you play, you're developing a sense of mastery,

51:44 and that actual sense of progress and mastery is the fun.

51:48 Piecing it all together is very core to what makes it fun.

51:51 (loud explosion)- Hello, Dr.

51:56 Freeman.

51:56 The car's all ready for you.

52:00 Hop in and I'll lower you down to the beach.

52:03 We did the player drivable cream first, I think, right?

52:07 Yes.

52:08 [Robin] And then we went back and used it

52:10 again at the start- [Adrian] As a perfect tutorial.

52:13 Like, perfect training.

52:15 "This is how that thing works.

52:16 Don't forget 'cause it's about to happen later and you have to do

52:19 it."- We did actually get very close to the base of a crane.

52:24 Obviously you wanna be in the crane, you wanna be driving and all these things,

52:26 but yeah, we did get up very close to this crane.

52:28 Again, I think it was asking somebody,

52:30 "We're just wanting to take some pictures for a video game,

52:34 could you just let us in?" Try to act as nice and non-threatening as possible.

52:39 [Robin] We need the buggy to get into a state where it was flipped,

52:42 so you could have to learn to recover and all that sort of stuff.

52:45 Like, "Oh, we're gonna flip it over and then to proceed,

52:47 you have to use the the gravity gun.

52:50 We're just trying to wrap that in some sort

52:52 of narrative instead of just spitting some text that tells you,

52:55 "This happens."- I mean, the crane was a hard physics problem 'cause one

53:00 of the worst case scenarios for physics is really

53:03 high mass objects moving around 'cause they're gonna apply

53:07 an enormous amount of force to anything they touch,

53:10 and we have a rigid body physics system.

53:13 So in the real world when a giant heavy mass touches something,

53:16 usually it crushes them.

53:18 We can't really crush anything.

53:20 So already we're in a tricky place.

53:23 The good news was we had a fixed space it could operate in, right?

53:26 We could control how far you could extend the thing,

53:27 and so as the level designers could look at it and be like,

53:30 "These are the only things that the crane will ever really be

53:33 able to interact with." And so we could be pretty careful around,

53:36 "What are the things that this magnet's gonna touch,

53:40 and let's make sure that at least the physics

53:42 works pretty well for that." (train horn blaring)- So

53:53 I worked on the bridge as one of the really

53:56 early things I worked on there, which was great.

53:58 Sawyer had already built the bridge.

53:59 It's actually based on Deception Pass.

54:02 So Deception Pass is nearby here in Washington.

54:05 Coincidentally, I actually proposed to my wife there.

54:07 But anyway, it's this really nice bridge that if you go there,

54:10 you'll be like, "Oh yeah, this is the 'Half-Life 2' bridge." The fight against

54:14 the gunship and everything wasn't really all put together.

54:17 It definitely had that scary fear of heights scenario of going under the bridge,

54:24 but we hadn't fully fleshed it all out yet.

54:26 One of the first things I did was like,

54:27 "What if this was a train bridge and you had to play chicken against

54:31 the train?" And getting the timing quite right took a little bit of of work,

54:36 but hearing the like train horn come in and the moment

54:38 of realization that players would have of- Probably some people die.

54:41 A lot of people probably die the first time there,

54:43 but you have this like, "Do I go faster or do I turn around?

54:46 What do I do?" I don't know, it was just a fun moment to create.

54:49 All credit to Sawyer on coming up with the going underneath the bridge scenario.

54:54 People to this day still are like, that freak them the hell out.

54:57 Like a lot of people would just, it's sort of seared in their memory,

54:59 which is just, it's what we wanna do when we're creating levels.

55:04 Trying to like capture the sense of height and playing with the ambient

55:07 sounds just to capture the sense of being precarious and high above the ground,

55:12 it was just a fun little design challenge.

55:16 David came up with that and we play tested it,

55:19 and the first time it play tested I think the person who was play testing,

55:22 their knuckles were white and it was really tense.

55:25 And of course then someone had the brilliant idea of like,

55:26 "Oh, they finally made it.

55:28 They felt great.

55:29 Now we're gonna make 'em go back." Oh, and the gun ship.

55:32 Yeah, that came in.

55:33 Yeah, so that was always fun.

55:35 (loud automatic gunfire) (distant helicopter whirring) (muffled beeping) There's

55:40 this thing we called "edge friction." So when you're moving,

55:43 you have a friction that just reduces your normal movement rate by an amount.

55:48 But what we would do is whenever players would get close to edges,

55:51 we would increase that friction a lot

55:53 so that people wouldn't accidentally fall off things.

55:55 It's just not fun for the player, right?

55:58 So there was definitely a bit of friction play involved there.

56:01 (wind blowing) (vehicle revving aggressively)- Gordon Freeman!

56:11 Hurry, get in the basement!

56:12 We're expecting gunships at any moment!

56:15 He became iconic for reasons I don't totally understand except the performance.

56:19 We were calling that town New Little Odessa or a Little New Odessa.

56:24 And then I got spam email from Odessa Cubbage or something like that.

56:28 It was definitely a name that came out of a email spam and I'm like,

56:32 "This name is amazing." So it went straight into the game.

56:36 and then John Lowrie came in and acted him with his fake British accent.

56:40 Damn!

56:41 Let me just send a warning to Lighthouse Point,

56:43 and then I'll come right up and lend a hand!

56:46 So Odessa Cubbage had this whole problem of training the rocket launcher.

56:49 That was a real pain.

56:51 The laser guiding aspect was just a nightmare to train,

56:54 'cause training in combat is always a nightmare by itself

56:57 'cause people are under duress, trying not to die.

57:01 They're focusing on some other thing that's threatening them.

57:04 It's hard to observe what's going on rationally

57:06 and draw lessons from it cleanly, yeah.

57:09 So you always know it's gonna be hard if you're trying to teach something,

57:12 you're trying to teach in combat.

57:14 When you watch a play tester,

57:15 they step out and they fire a rocket at the gunship,

57:18 and then they duck behind cover.

57:19 And so the laser dot goes onto the fence in front of them.

57:23 We all hear the rocket turn around and start coming back,

57:26 and so you hear like-- It's basically, "Here we go.

57:28 Five, four, three-"- As it gets louder,

57:29 and then the player looks out again at the gunship,

57:33 and the rocket goes (whooshing).

57:34 So he moves around, back at the crate, gunship.

57:35 And we're like, "Ah!" And then he ducks back again and we're like,

57:39 "Oh my God!" We hear the rocket getting louder and louder.

57:41 And he peeks at it again, it whips off again,- An integrated part

57:44 that I think has really always been important,

57:47 something that Gabe introduced in "Half-Life 1"

57:49 one was the way that we play test.

57:52 And then iterate really rapidly on the play test results.

57:55 So you have dozens and dozens of people come through

57:59 one section of the game and you sit there painfully silently,

58:04 taking notes while they struggle and get

58:08 frustrated and totally misunderstand your brilliant design.

58:13 And then you go and you fix that over and over and over again.

58:16 If your title is game designer and your design sucks,

58:20 then it can be threatening, right?

58:21 Like, "Maybe I'm bad at my job and I should get fired," right?

58:24 And so a lot of the games industry never

58:25 did any testing at all with this play testing process.

58:29 That's what basically taught me design in a lot of ways.

58:32 We would just come up with an idea

58:34 for the helicopter and then we'd be like, "Okay,

58:36 let's put it in front of somebody and watch

58:37 it happen." And then we'd very quickly start realizing

58:42 that our mental model of how people are reacting

58:44 to this stuff is wildly different than what we expected.

58:48 (loud automatic gunfire) (beeping) (compelling music)-

58:55 I joined the team in January, 2000.

58:58 I had met Gabe when he was starting Valve and I was at Sierra On-Line.

59:03 But pretty much the first thing was, Gabe was like, "So when you're at Sierra,

59:08 we did this deal with you for 'Half-Life'

59:12 and 'Half-Life 2.' You need to go fix that 'cause

59:15 we don't like the deal that we did."

59:17 The intellectual property for "Half-Life" was owned by Vivendi.

59:21 The "Team Fortress" IP was owned by Vivendi.

59:24 So it was get back our IP, get a better royalty rate,

59:28 and then have the ability to experiment with online distribution.

59:34 We knew we were gonna grow the team,

59:36 and then part of the deal was also that we were

59:38 gonna take on all of the funding responsibility for "Half-Life 2",

59:42 "TF 2," all those things.

59:44 There was cashflow from "Half-Life," we were figuring

59:47 out how to do things here and there, like expansion packs.

59:50 A large part of it was Gabe tapping into his assets.

59:56 He had done fairly well at Microsoft.

59:58 Not crazy well, but things started to get a little thin.

1:00:02 (suspenseful music) (loud banging on door) (suspenseful music

1:00:15 continues)- So I assume you guys have talked

1:00:17 to Scott or Eric or somebody about sort

1:00:21 of the context of the whole-- [Producer] Yes,

1:00:24 we talked to Scott, Eric, and Gabe about it.

1:00:26 (Karl chuckles)- Sierra, by that point, had become acquired by Vivendi,

1:00:32 had worldwide distribution rights, but only for retail package product.

1:00:37 Vivendi was licensing "Counter-Strike" to cyber cafes.

1:00:42 We went to Vivendi and said, "Hey, this is not a big deal,

1:00:46 just agree that this is outside of your license.

1:00:50 And we've spent a little bit on attorney's fees,

1:00:53 some tens of thousands just covers for that and we'll

1:00:58 move on and just keep going." And they wouldn't do it.

1:01:03 I think some people at Vivendi did not

1:01:06 like the new deal that they ended up signing.

1:01:09 And I think there were some hurt feelings in some quarters about that deal.

1:01:15 When we said, "Hey, the cyber cafe thing,

1:01:19 you really can't do that" coming on top

1:01:21 of a year negotiations and torturous negotiations,

1:01:25 I think that just struck an angry cord.

1:01:28 We ended up filing the lawsuit,

1:01:31 but really just on this narrow thing of like, "Hey, court,

1:01:36 will you go just read this language and say are they outside

1:01:41 the license or not?" And Vivendi decided to go World War III.

1:01:48 We filed our case in Seattle, federal court in Seattle,

1:01:51 and they hired a law firm to defend it, of course.

1:01:54 There was some initial skirmishing about discovery,

1:01:58 there were some disputes and motions that had to be decided by the judge,

1:02:02 and those went all Valve's way.

1:02:03 And then suddenly the law firm disappeared.

1:02:07 They hired a big scary law firm from San Francisco.

1:02:11 They thought, "Okay, Valve's had some success in this case,

1:02:16 and so we need to change their mindset from 'We think we're

1:02:19 gonna be successful' to, 'We're gonna be destroyed.'" The next thing we knew,

1:02:24 here comes this big stack of counterclaims,

1:02:28 everything from canceling the 2001 agreement to obtaining ownership of all

1:02:35 the "Half-Life" IP to keeping us from doing Steam, you name it.

1:02:41 Where they included Scott and Julie Lynch and Gabe and Lisa Newell,

1:02:45 who was his wife at that time, personally as defendants.

1:02:50 And so it was kind of like, "Well,

1:02:51 we're gonna put Valve out of business and then

1:02:54 we're gonna bankrupt the two of you." I think that was a lot of what they

1:02:59 felt like was their path towards winning this fight.

1:03:02 They decided that it would have more effect and be scarier if process

1:03:06 servers showed up at everybody's house in the night and handed- I mean,

1:03:12 so yeah, a real bullshit move.

1:03:17 It's not a legal strategy, it's basically trying to intimidate you.

1:03:20 In other words, they're saying,

1:03:21 "Not only are we going to take all this money from the company,

1:03:24 but we're gonna take money from you personally as well."-

1:03:28 I can't think of another case when that happened.

1:03:31 They were certainly a lot bigger than we were.

1:03:33 Vivendi was a multinational conglomerate.

1:03:36 As compared to us, their funding was infinite.

1:03:39 It felt very David versus Goliath.

1:03:42 Publishers at that time in the industry were

1:03:44 used to being able to bully developers, right?

1:03:47 And so this was really about an assertion of power

1:03:50 as much as it was optimizing for a financial outcome.

1:03:54 The tactics that they were using were trying to just run us out of money, right?

1:03:57 So they knew how much money we were making

1:03:59 because they were doing retail distribution of our games,

1:04:02 and so they were just trying to crank up our legal

1:04:04 costs as just another way of draining resources from the company.

1:04:08 We were protected from that a lot internally,

1:04:10 so we didn't have to think about it.

1:04:12 They did a pretty good job of letting the team concentrate on what

1:04:15 the team could do without worrying about the future of the company so much.

1:04:18 I mean, that was my perspective.

1:04:21 So yeah, I definitely was not shielded from that.

1:04:24 The lawsuit was something where it created all

1:04:27 this uncertainty around whether the game's gonna come out.

1:04:30 Vivendi was definitely trying to use that as leverage.

1:04:34 And so even if we finished the game, they're not gonna let us release it.

1:04:37 There came a point where Gabe had pretty much

1:04:40 used up his liquid assets and he was like,

1:04:44 "Should I put the house on the market?" And I was like,

1:04:48 "Yeah, I think it's the time where you put the house in the market

1:04:51 if we want to keep going."- The company was pretty close to going bankrupt.

1:04:56 I was pretty close to going personally bankrupt.

1:04:58 We went all in.

1:05:00 It's like, there was no money left.

1:05:02 [Producer] How anxious was that for you personally?

1:05:06 I don't know.

1:05:09 There's certain things that I just,

1:05:11 I'm kind of a weird person in a number of dimensions.

1:05:15 I don't really think, "Oh, this is super scary." I don't have

1:05:18 an emotional response to it, I just say, "Well,

1:05:20 this is what we're doing and we'll see how it plays

1:05:24 out." So I don't think it was super stressful to me, right?

1:05:27 I mean, it didn't really bother me.

1:05:29 I was diving in South Africa recently and a shark

1:05:32 tried to bite me a couple of times,

1:05:34 and the people around me were way more freaked out than I was.

1:05:36 I was like, "Oh, a shark's trying to bite me.

1:05:38 I should get away from the shark." Whereas

1:05:40 other people were having like, "Oh, a shark!

1:05:42 It's trying to bite somebody!" And I just think that's how I'm wired.

1:05:47 I don't think it's anything that speaks to my character or anything,

1:05:50 it's like I just seem to not get particularly agitated around risk,

1:05:56 which probably means I take on more risk at times than most sane people would,

1:06:02 which can be a positive or you can wreck a bunch

1:06:05 of other people's lives being in the neighborhood of your risk indifference.

1:06:10 (suspenseful music) (suspenseful music continues) (eerie music)

1:06:27 (monsters shrieking)- They're this rigidly structured organization.

1:06:39 Of course they're gonna have this really terse language for everything they do.

1:06:43 There's sort of an announcer voice that follows

1:06:47 you around and kind of comments status essentially,

1:06:51 like how alert the world is to you.

1:06:53 (chatter over radio) (gun reloading) I sat

1:06:58 down and just basically wrote a language

1:07:00 for that so that the soldiers and the Metro

1:07:03 Cops would have jargon that they would use.

1:07:06 Then I just used that to kind of encode.

1:07:09 So that's where like Anti-Citizen One and things

1:07:11 like that came from, rather than saying,

1:07:13 "The guy with the orange suit is causing trouble," it would be,

1:07:19 "Anti-Citizen One breach.

1:07:21 Scalpel something, sector something." So I

1:07:24 basically just had this little dictionary.

1:07:26 Then I overdid it on the vocal processing.

1:07:29 because they're all wearing these masks and I

1:07:32 wanted it to sound like they're in a mask.

1:07:35 And it was very deliberate to never have

1:07:37 to shoot somebody that you can see their face.

1:07:41 Like, we didn't want you killing humans.

1:07:43 I mean, the Metro Cops are humans, but we covered all the faces up.

1:07:47 I think it was all part of this calculated and deliberate way of saying like,

1:07:51 "Look, games don't just have to be a certain way.

1:07:54 There's so much room in what we can do here." (gun clicks, reloading)- Well,

1:07:59 the metro police and the Combine soldiers share a lot of AI,

1:08:03 and so there's an internal communication among them.

1:08:06 And one of the things we did, both for the sake of difficulty and exposition,

1:08:10 was limit the number of squad members that could

1:08:12 be firing at the player at any given time.

1:08:14 I think it's two.

1:08:15 So if you have a squad of four,

1:08:16 a guy who has an opportunity to shoot with the player

1:08:19 grabs one of those slots if it's available and starts to shoot.

1:08:22 And if someone else, another squad member,

1:08:25 wants to attack the player but no slots are available,

1:08:27 they have to find the next best thing to do, which is,

1:08:28 "I'll get outta the scene for a second,"

1:08:31 or, "I'll go reload" or, "I'll throw a grenade."

1:08:33 And that was all stuff that we learned building

1:08:36 "Half-Life 1" and I think got a lot better at.

1:08:39 (rapid gunfire) (loud explosions) (chatter over radio) (rapid gunfire)

1:08:43 (loud automatic gunfire) "Half-Life 2" used a node graph system,

1:08:47 hand placed nodes by developers in the levels would describe a really

1:08:51 super low fidelity version of the environment for the AI to use.

1:08:55 It enabled the level designers to sort of inject and puppeteer a little

1:09:00 bit the AIs in ways that were just easier for them to sort

1:09:04 of be in control as opposed to having to grab an animator

1:09:07 or a programmer to be able to get really cool things into the game.

1:09:11 (loud whirring) (wood thuds on ground) (monster growls) (loud

1:09:23 automatic gunfire)- Floor-is-lava is a really obvious piece of gameplay.

1:09:28 Physics is really fun to manipulate.

1:09:30 We just thought like, "Hey, you could build your own path." We'd

1:09:34 already seen players who enjoyed moving stuff around,

1:09:37 like building bridges and stuff, and we're like,

1:09:40 "We don't have enough moments where I just pick up a plank

1:09:43 and lay it across a gap and walk across it." Like,

1:09:45 obvious stuff, we've barely done that in the whole game.

1:09:49 And the fact that you're laying stuff down on the floor

1:09:51 means that the finesse required to do precise placement wasn't necessary, right?

1:09:56 You're like, "I just wanna place it here," right?

1:09:59 And doesn't matter the orientation,

1:10:00 you just kind of jump on it and you do the next one and so on, right?

1:10:02 And also training was free.

1:10:05 We didn't have to train it.

1:10:06 Yeah, everyone got it.

1:10:06 "Stay off the sand" and everyone was like, "Oh,

1:10:07 the floor is lava."- Laszlo, the finest man of his generation.

1:10:10 He couldn't figure that one out.

1:10:11 He tells you exactly how to do it.

1:10:13 Someone's coming.

1:10:14 You there!

1:10:14 Stop where you are!

1:10:15 Stay on the rocks!

1:10:17 Don't step on the sand, it makes the Antlions crazy!

1:10:20 Laszlo, don't move.

1:10:21 No!

1:10:21 Help!

1:10:22 (monsters shrieking) (loud gunfire) Dear God!

1:10:30 Poor Laszlo!

1:10:32 The finest mind of his generation!

1:10:34 To come to such an end.

1:10:36 We were heading for the Vortigaunt camp,

1:10:39 hoping to pick up some Bugbaits so these damn things would leave us alone.

1:10:42 Without Laszlo, what's the point?

1:10:46 You just give these NPCs a name at the moment they die and let

1:10:51 the fans fill in their whole life story and give them something to play with.

1:10:56 The Freeman will have needs of these Pheropods on the paths ahead.

1:11:00 Gather them now.

1:11:02 Nova Prospekt and the Antlions with the pheromone balls were really

1:11:05 fun because it lets you experience them in a different way,

1:11:08 and it was another classic example of getting

1:11:10 the most value out of technology investments that people could.

1:11:14 I remember the beach thing, the design intention was definitely to make it

1:11:18 feel like a sort of D-Day beach landing.

1:11:22 You didn't come from the water,

1:11:23 but you're storming up towards these old gun placements, if I remember right,

1:11:27 with Combine in them and you're invading

1:11:29 the bunkers and you're bringing your hoard with you.

1:11:31 You're overwhelming these in place defenses.

1:11:36 It was surprising how much chaos you could get away with.

1:11:41 I remember we did a lot of tuning to make sure the Antlions spawned.

1:11:45 We had difficulties from time to time

1:11:46 with design some choke point or challenge or whatever,

1:11:49 and the Antlion flow rate wasn't enough to make it feel chaotic enough.

1:11:54 We had an atmospheric goal too, it wasn't just about, "Mechanically,

1:11:58 can you overcome this?" You wanted to feel your forces have stormed the walls,

1:12:04 they're everywhere, there's chaos all around you.

1:12:06 And so a lot of the tuning was about achieving that feeling,

1:12:09 not just about mechanically getting past the enemies that we'd set.

1:12:14 Nova Prospekt was different than a lot of the maps that I worked on at least,

1:12:17 is that in that we had some early art prototypes that were done.

1:12:22 We have this whole concept of zoom maps we call 'em,

1:12:24 which are basically little art test scenes,

1:12:28 how they would fit together and some basic lighting.

1:12:31 But by the time our cabal got to Nova Prospekt,

1:12:34 we knew we were coming in with Antlions and with Bugbait.

1:12:37 And so we already knew that a bunch of stuff was gonna change

1:12:40 because we had to support you

1:12:43 and your army of Antlions fighting Combine soldiers.

1:12:47 Antlions look great whenever they were being shot by anyone.

1:12:51 Particle flecks flying off, and they always look great.

1:12:55 And so setting them up against Combine soldiers looked great,

1:12:58 setting rights against turrets looked great.

1:13:00 Those were super fun scenarios to work on though because having these Antlions

1:13:03 that you could just use in any way you wanted was really fun.

1:13:07 And coming from the coast,

1:13:08 just the contrast of shrinking everything down and then just

1:13:11 piling it full of stuff was a lot of fun.

1:13:14 (loud gunfire)- The first time I saw one

1:13:21 of those, I think it was Ted Backman's concept.

1:13:25 I just thought, "I can't believe that I get

1:13:27 to use a creature this cool in level design.

1:13:30 These guys are really, really, really fun to use.

1:13:33 [Ted] I remember Kelly Bailey had created a short list.

1:13:37 Things like he wanted Flocking Flyers and he wanted a moving guard tower,

1:13:41 which was the first description of the Strider, was a mobile guard tower.

1:13:44 So I saw that and immediately understood

1:13:47 that, kind of orchestrating the enemies in that way,

1:13:51 trying to get the most variety and the most breadth of shape

1:13:55 and movement and experience was the correct way to move forward.

1:13:58 The design for that first strider map,

1:14:00 it was something I came up with and I was really passionate about.

1:14:03 It was all about running across and just getting one shot in your tush, right?

1:14:09 Just as you made it through.

1:14:11 [Dario] Even just hearing the sound, the thump,

1:14:13 thump, thump of them walking, you know, "Okay,

1:14:15 there's a boss nearby." You just see the legs

1:14:18 walking by, but you can't see the body.

1:14:19 There are so many things you can do with these guys.

1:14:22 Honestly, there were still people on the team I

1:14:23 think that were still thinking that we were gonna bomb,

1:14:25 that was people are just gonna not be impressed.

1:14:28 When did you know it was gonna be good?

1:14:31 E3.

1:14:32 Yep.

1:14:32 E3.

1:14:32 I'd say the same thing.

1:14:33 (loud rapid gunfire) (monster growling) (loud rapid gunfire)- Oh my god!

1:14:43 (loud rapid gunfire) Help!

1:14:45 Help!

1:14:46 Yeah, it seems like at some point there was an E3

1:14:48 demo that we got together that we didn't end up showing,

1:14:51 but at least at that point it just felt like, "Oh, wow,

1:14:54 we actually have made something here."- We went through and made a bunch

1:14:58 of demos for 2002 and everybody looked at 'em and was like, "Eh,

1:15:01 we're still not confident about this and we don't think we've hit the bar,

1:15:05 so we're not gonna go at all." They had totally planned to go.

1:15:08 And then when 2003 rolled around, I just remember everyone was still terrified.

1:15:12 We were like, "Okay, how are we gonna craft everything we've had

1:15:15 into a number of messages about what this product

1:15:18 is?" And we came up with like the core theme of each one of these movies.

1:15:22 Oh, do be careful.

1:15:24 We didn't really understand at the time,

1:15:25 I think, the embarrassment of riches that we had.

1:15:29 We looked at it from two angles.

1:15:30 One was a set of technical features,

1:15:34 a sort of conceptual listing of things that we thought were important,

1:15:38 like physics and things like that, and then the other side was content based.

1:15:44 What areas of the game were innovative,

1:15:46 what areas of the story we thought were compelling,

1:15:49 which characters we thought were worth people paying attention to.

1:15:53 And so it was an interesting sort of marriage of which

1:15:56 features and which content are the best showcase for all those things.

1:16:01 A lot of what was in the demos was rendering technology,

1:16:03 but I think the stuff that felt the strongest

1:16:07 to most of us at the time was really physics based.

1:16:12 (steady gunshots) (wood splintering)- [Developer]

1:16:17 So if something looks like wood, then it sounds like wood,

1:16:20 scrapes like wood, floats like it,

1:16:22 and if you shoot it, it'll fragment like wood.

1:16:24 (steady gunshots)- To people at that time,

1:16:30 it was pretty mind blowing just to be able to pick things up,

1:16:34 knock other things over,

1:16:35 use things in a way that felt true to a three-dimensional reality.

1:16:39 Maybe it's on those videos, I don't remember,

1:16:41 but people were gasping and hitting the people next to them

1:16:45 because they couldn't really believe what they were seeing on the screen.

1:16:49 [Developer] There's no limitations to the complexity of those interactions.

1:16:56 So it's this level of believable and consistent interactivity that opens

1:17:00 the door to a wide variety of new gameplay mechanics.

1:17:03 [Audience] Think this will run in my 486?

1:17:05 (crowd laughing)- That person who was talking about a 486,

1:17:10 obviously it's just about,

1:17:12 "You must have a super computer behind..." Like, "How are you even doing this?

1:17:18 What is going on?" (muffled gunfire) (testers laughing) The lines

1:17:33 outside the booth became enormous and it was the thing

1:17:37 that it really distorted the rest of E3 because

1:17:40 so many people were anxious to come see our demo.

1:17:43 And I remember seeing this absolutely enormous line

1:17:46 with some pretty well known people in there, right?

1:17:49 Miyamoto was in line for this stuff.

1:17:50 And Will Wright, I think.

1:17:52 Yeah, Will Wright, there was a bunch.

1:17:53 And so then, yeah, after that was all done people on the team were just like,

1:17:57 "Okay, we are ready.

1:17:58 We can ship this game.

1:17:59 We know we've got something good." It was really easy at that point.

1:18:04 It felt like we were rolling the ball downhill after that.

1:18:07 Until it leaked.

1:18:08 Until it leaked.

1:18:10 (eerie music)- When something like this happens,

1:18:21 there's not like a big sign saying,

1:18:23 "You've been hacked and somebody has access." You start off

1:18:26 with something that just doesn't make a whole lot of sense,

1:18:28 and you start pulling on that sweater thread.

1:18:30 Gabe's brother had a company called

1:18:32 Tangis and they made wearable computer stuff.

1:18:34 So he was kind of winding it down and he had

1:18:37 one or two Unix servers that hosted the tangis.com domain name.

1:18:41 Unfortunately they were put inside the firewall and didn't

1:18:44 have the latest Linux or whatever security updates on them,

1:18:47 and so this kid from Germany was able to sort of exploit his way into that.

1:18:51 He sat there for months, I think,

1:18:52 he was able to sync the build, all the source code.

1:18:55 I was riffing ideas for, I think in particular "Counter-Strike."

1:19:00 that information ended up on one of the internet chat rooms,

1:19:04 and I only two people knew the contents of that email.

1:19:07 Of course, Gabe and myself.

1:19:09 And I went to ITG and said, "This could only have come from a leaked email.

1:19:15 We gotta figure out how that got out there."-

1:19:17 But I remember at some point Gabe walking around,

1:19:20 like one of those moments, like, "Turn your computer off.

1:19:22 Something just happened."- That's right.

1:19:23 I remember being on my desk and he's coming like, "Turn your computer off.

1:19:25 I think we're being hacked," and I was, "What?" I didn't know what to do.

1:19:29 I actually did not know what to do.

1:19:31 I remember coming in that morning and figuring it out and I

1:19:34 just felt like someone had just punched me in the gut.

1:19:36 It's like the bit in movies where after an explosion they'll

1:19:39 have a scene where someone's staggering around the street and everyone's like...

1:19:43 What's this mean?

1:19:44 Can we still keep working on the game?

1:19:47 Is someone just gonna release the whole game tomorrow?

1:19:49 And I don't know if the guy who hacked it intended to.

1:19:53 Yeah.

1:19:54 He showed it to friends.

1:19:54 I don't think he intended the wide distribution.

1:19:56 He showed it to friends, and then one of his friends posted it somewhere.

1:19:59 Yeah.

1:20:01 Unfortunate.

1:20:02 Yeah.

1:20:02 (eerie music) (eerie music fading)- And one of the things I

1:20:17 think that's pretty interesting is how much the leak of the game,

1:20:21 the hack of our network and the leak of the game,

1:20:23 how much that kind of shattered that confidence.

1:20:26 It was like having a movie that you're working on be spoiled for the world.

1:20:30 That's how it felt because we knew it would be hard for people

1:20:33 to know that that existed and not go look at it all.

1:20:36 It looked all broken to us and it just felt really demoralizing.

1:20:40 Oh, man, I was seriously affected by that.

1:20:43 Yeah, I was pretty angry.

1:20:45 So I don't know, it kind of affected me for a while.

1:20:49 Having it be having it called the beta when

1:20:51 it was not even close to being a beta.

1:20:53 Work in progress, you know?

1:20:55 Wasn't really ready to be seen yet.

1:20:57 Normally the office is pretty vibrant and loud and people

1:20:59 riding scooters in the hallway and doing all sorts of stuff,

1:21:02 Gabe's Tuvan throat singing; and it got very still and very quiet.

1:21:05 I remember that.

1:21:06 That was very eerie.

1:21:07 That was the game that was gonna make or break the company.

1:21:10 We didn't know if we were gonna have to go find jobs.

1:21:13 Not only did it show everybody that we were

1:21:15 not anywhere near where we said we would be,

1:21:18 but all the work that we had in progress,

1:21:21 all the plans that we were making were just laid out bare.

1:21:24 And so a lot of us just thought, "Oh, we're toast.

1:21:27 We've worked so hard.

1:21:29 Are we gonna go bankrupt?" And then all that while,

1:21:31 Scott Lynch and Gabe were being sued by Sierra,

1:21:35 their wives being delivered subpoenas.

1:21:37 And so it was this super dark time.

1:21:42 So there were two interesting threads going on.

1:21:43 One was traditional law enforcement intelligent agencies,

1:21:47 and they were doing their thing and issuing subpoenas and seizing devices.

1:21:51 And then we asked our community to help us out.

1:21:53 And I think Axel Gembe started to realize

1:21:56 that he was being tracked down by the community, right?

1:21:59 Everything was being essentially, it was an open-sourced investigation,

1:22:03 so he could see that they were getting closer,

1:22:05 and that's when he reached out to me.

1:22:06 Of course I was insanely furious with him,

1:22:09 but we ended up running this sort of scheme with him,

1:22:12 where I was like, "Oh, wow, you must be really good at security issues.

1:22:15 We should have you come out and interview." And the FBI

1:22:18 was gonna arrest him when he got off the plane.

1:22:21 And then apparently the German police sweep in right

1:22:23 as he's about to leave because it turns out he's also guilty of a bunch of bank

1:22:29 fraud and bank hacking and all that stuff.

1:22:30 He had hacked in a bunch of huge, huge companies over there as well,

1:22:34 and so they were already well aware of him.

1:22:36 How much it harmed us financially or how much

1:22:39 the success of the game was harmed by that, very debatable.

1:22:42 So it was hard for us to go through internally,

1:22:44 but in the end I wouldn't have wanted to have

1:22:47 it be a whole law enforcement debacle after that.

1:22:52 Yeah, it wouldn't have really helped anyone.

1:22:54 And I'm not saying we should ever be lax about security,

1:22:56 but if we could survive that, maybe we'll survive anything.

1:23:00 Because at the end of the day,

1:23:01 the special sauce was everything that happened between

1:23:04 the moment of that leak and when we actually shipped.

1:23:06 And if you look at that version of the game,

1:23:09 it's just so not what the final game ended up being.

1:23:12 And so even if our competition had all the source code and that executable,

1:23:17 they're still not able to produce "Half-Life 2" from it because

1:23:19 the company kept moving forward

1:23:21 and the company was the interesting special sauce.

1:23:25 (gentle instrumental music) (suspenseful, intense music) (suspenseful,

1:23:43 intense music continues)- Gabe in particular,

1:23:47 he had kind of a pencil sketch of an idea

1:23:50 in his head of what would become Steam,

1:23:53 but it was clear that "Team Fortress Classic"

1:23:55 and then later "Counter-Strike," it was fundamentally the thing

1:23:59 that we were really attracted to was

1:24:01 our ability to ship content directly to our customers.

1:24:05 I mean, there was a set of business goals that ended up being part of Steam,

1:24:09 but fundamentally it was a bunch of game development goals

1:24:14 that it was servicing that were so attractive to us.

1:24:17 It was a very weird time.

1:24:18 I don't think people understand how many times we would go to people and say,

1:24:22 "No, you will be able to distribute software

1:24:25 over the internet." And have people just say,

1:24:28 "No, it will never happen." I'm not talking about one or two people,

1:24:31 I mean like 99% of the companies we talked to said, "It will never happen.

1:24:36 Your retail sales force will never let it happen." But also people would say,

1:24:39 "Users aren't gonna want this," right?

1:24:41 "People want physical copy." There were so

1:24:44 many bad faith arguments that were being made.

1:24:46 Retail sales is not the goal, right?

1:24:49 It's actually an impediment,

1:24:52 it's somebody who sits between you and the customer.

1:24:55 And we ended up going out, finding this company called Applied Microsystems.

1:24:59 So we ended up hiring most of the original Steam team from that other

1:25:03 company to build initially this sort of in game advertising streaming model,

1:25:07 but then there was the epiphany that, "Hey, it's just bits.

1:25:09 Why don't we just download whole games this way?

1:25:12 You guys go off and do it."- The decision to not

1:25:15 only use Steam to ship "Half-Life 2" but actually to require Steam,

1:25:19 even in the versions that were purchased at retail in a box,

1:25:24 was the most interesting decision of all those because it turned out

1:25:28 to be an incredibly important decision for the future of the company.

1:25:32 And a lot of us were nervous,

1:25:34 and a lot of the people who had been at Valve for a long

1:25:37 time since the very beginning were the most nervous about that decision.

1:25:41 And so it was one of the rare exceptions to our decision making process usually,

1:25:46 and Gabe had to really step in and say,

1:25:49 "No, actually we're doing it this way." (music fades

1:25:54 out)- I remember coming to work in the morning,

1:25:58 working on "Half-Life 2," which meant by the time you go home at night,

1:26:03 me being a 24-year-old idiot, all I wanted to do was keep working on the game.

1:26:08 And I started just making the game work on multiplayer, right?

1:26:12 Mostly just working on prediction.

1:26:14 Like, what would it feel like to have

1:26:15 a gravity gun on a multiplayer setting, right?

1:26:17 Getting physics objects to work.

1:26:18 And it was just curious curiosity.

1:26:20 Scott Dalton saw me working on it and he's like,

1:26:23 "I'll make a map and then we play." Put the map up,

1:26:26 we started playing in the office in between

1:26:28 builds of "Half-Life 2," like play test.

1:26:30 And I remember Gabe walking by my desk and I remember him going like,

1:26:35 "What's that?" And I said, "Oh, it's 'Half-Life' multiplayer." And he's like,

1:26:39 "Show it to me tomorrow." And he just walked away.

1:26:42 [Robin] I think that is one of the advantages of the general

1:26:45 cost we always pay to keep our code in a single code base.

1:26:51 So when we're working on "TF 2" and "Half-Life

1:26:53 2" at the same time at that point,

1:26:55 we're paying this continuous cost in "Half-Life 2" for the fact

1:26:59 that it's built upon an engine that's designed to do multiplayer.

1:27:04 But it means that when you get

1:27:05 to this later point where you start asking, "Well,

1:27:07 can we build multiplayer death match outta this?" The code's

1:27:10 all sort of built upon a fundamentally client-server sort of foundation.

1:27:14 (loud gunshots)- Definitely the most entertaining thing

1:27:18 that you could throw at somebody that happened

1:27:20 to be scattered around the world was picking

1:27:22 up a toilet and hurling it at someone's head,

1:27:25 and then having that show up in the upper right corner of your screen with name,

1:27:30 toilet, name was very satisfying.

1:27:33 (automatic gunfire)- [Adrian] I was also selfishly trying to get

1:27:41 the game ported to multiplayer so then we can release the tools,

1:27:45 the SDK ready to go.

1:27:46 Like it meant mod makers could just

1:27:48 immediately start making multiplayer game mods rather

1:27:51 than waiting for us to do it or someone else to pick it up.

1:27:53 (loud automatic gunfire) (character yells in pain)-

1:27:57 We were so heads down working on what

1:27:58 we were working on we didn't really get a chance to play the train station.

1:28:02 I remember that that was one of the lessons

1:28:03 that came away from "Half-Life 1," is like,

1:28:05 "We need to be building these games backwards

1:28:07 in a sense." The first thing we're gonna put

1:28:09 in front of people is the thing we built when we're as good as we can be at it.

1:28:14 We know our tools and things like that, and a lot of the anxiety over,

1:28:17 "Are we building a good game?" was behind us and I felt really confident.

1:28:21 "Yeah, we're building a really good game.

1:28:26 (tense music) (aircraft whirring) (tense music

1:28:36 continues)- We're entering the Gare d'Austerlitz,

1:28:41 and this train station is very important because it's one of the main

1:28:46 inspirations for the train station of City 17 where the player starts.

1:28:53 The reason why I'm using a Parisian station produced by Eiffel,

1:28:57 who did the Eiffel Tower, is that there's an identical station in Budapest.

1:29:03 Train station's the last bit we worked on, right?

1:29:05 And we always want to build the start of the game

1:29:08 as late as we can 'cause we understand our game at that point.

1:29:13 In general, the train station was a really

1:29:15 interesting problem because it was basically one big choreo

1:29:18 sequence intended to deliver atmosphere and story and background

1:29:21 and sort of set your expectations for the world.

1:29:23 We let the level designers who were really good

1:29:25 at this stuff figure out what was a good experience,

1:29:29 and then we were kind of always there to provide dialogue,

1:29:34 acting, stuff like this guy.

1:29:37 [Bill V] I think it was pretty clear too that we wanted to provide

1:29:40 incentive to the player at this point about how people were being mistreated,

1:29:44 about how bad the situation was here,

1:29:48 and kind of start to trigger that righteous indignation early.

1:29:52 All this stuff is fun to look at, but you

1:29:54 just assume it doesn't move and you can't interact with it,

1:29:57 you can't pick it up.

1:29:58 Because all the stuff we're gonna do later with the gravity gun and everything,

1:30:01 first we gotta start just mundane stuff, like that little empty bit of food.

1:30:06 [Adrian] I mean, it happens the moment you leave the train,

1:30:09 wind comes and blows all the trash away it.

1:30:11 Doing it everywhere.

1:30:11 [Adrian] Can see the Vortigaunt sweeping, the guy pushing on the cases.

1:30:15 [Robin] Yeah, and all those things fall off and you can pick those up.

1:30:19 We were like, "If it doesn't have to be bolted down,

1:30:21 just make sure that it gets knocked over."- It can move.

1:30:25 You, citizen!

1:30:26 Come with me.

1:30:27 [Bill V] This guy looks vaguely familiar.

1:30:29 Something about the way he walks.

1:30:31 [Fletcher] The way he walks.

1:30:32 It's like a beer drinker.

1:30:34 (Marc laughs)- Who would've thought Barney

1:30:37 has the capacity to go undercover, right?

1:30:40 That seemed to be above his pay grade,

1:30:43 above his intelligence pay grade from the get go.

1:30:47 [Marc] Look at that beautiful animation.

1:30:49 But where'd his hat go?

1:30:51 It's gone!

1:30:52 Something is not right here.

1:30:54 That's magical Combine tech.

1:30:56 (all laughing)- We were trying to figure out how to get

1:30:59 people to figure out how to pick up things and do things,

1:31:02 and we were all just sitting around 'cause we'd

1:31:04 been having this problem for weeks, racking our brains.

1:31:06 And I think Robin had a Coke can on his desk,

1:31:09 and at some point I got frustrated and I smacked it off his desk.

1:31:12 I'm like, "Well, you pick that up then."

1:31:13 And everyone kind of sat around and was like...

1:31:16 (radio beeping)- Pick up that can.

1:31:20 I just found it just an immensely

1:31:21 satisfying piece of a storytelling and world setting.

1:31:24 We got to use our physics tools and the choreo tools

1:31:26 and all of those things blended together into this very simple, clean moment.

1:31:31 The can cup is perfect also because

1:31:33 while it's teaching you how to manipulate objects,

1:31:37 it's also telling you something about the Combine

1:31:39 themselves and the Metro Cups- And your relationship.

1:31:41 Yeah.

1:31:42 So it was a twofer, right?

1:31:43 We got those two at the same time.

1:31:45 Physics is gonna be a big part of our game.

1:31:47 By the time you left train station as a player,

1:31:50 you understood that and you'd come to appreciate it 'cause we couldn't wait

1:31:54 until you got to Ravenholm before you found out how awesome physics is.

1:31:59 If you'd had any kind of weapon, I think they would've had to kill you.

1:32:02 And we found the longer we sort of held that away from ourselves,

1:32:07 the more we were forced to come up

1:32:08 with stuff that was still entertaining that you could do.

1:32:11 Like the raid in the tenements is a whole

1:32:14 thing that could work because the pilot could only run.

1:32:17 We don't very often force the player into a very specific set of actions.

1:32:22 That's one of the few times that we had to learn to really corral the player,

1:32:25 and I think we ended up being pretty successful.

1:32:28 I think the consequences for sort of not

1:32:31 following on the clear path were quick and efficient.

1:32:34 (Sawyer chuckles) (chatter over radio)- Get in here quick!

1:32:41 Keep moving!

1:32:42 Head for the roof!

1:32:44 So when you build the world, I start with the infrastructure.

1:32:46 What's really interesting for me is what is not visible to the tourist,

1:32:51 which is the courtyard, the backyard,

1:32:54 the shafts and the smoke stacks and the wires.

1:32:57 And every city has this whole existence,

1:33:02 and people are usually blind to it and we tend to look at storefronts, right?

1:33:08 Yeah, I mean a lot of it came down to art direction.

1:33:11 I mean, Viktor was very, very familiar with all that, the look and feel.

1:33:15 There was certain patterns and certain iconic shapes

1:33:19 that anybody who lived there could pick up on.

1:33:23 They're like, "Oh, yeah, my grandmother had that dresser." Or whatever.

1:33:26 And my parents are from Argentina and so they have that kind of sensibility,

1:33:29 but a lot of it came down to the reference gathering,

1:33:33 which was from Bulgaria, from Paris.

1:33:36 We have these buildings that, again, have back courtyards.

1:33:41 They have big staircases, they have cellars, attic spaces,

1:33:45 cellar spaces, and the apartments have big windows and thick walls.

1:33:49 So this was providing already a lot of context

1:33:52 for people who can hide in these apartments.

1:33:56 Help me!

1:33:57 Stop!

1:33:58 We didn't do anything!

1:34:01 (loud thudding) (flatline beeping) They'll be looking for you now.

1:34:07 You'd better run.

1:34:08 I remember that we wanted to have the player

1:34:11 see proof that the Combine were evil people,

1:34:14 not that they were gray in any way, that they were objectively the bad guys.

1:34:19 And then we had it to where you go upstairs,

1:34:21 and then we really carefully and intentionally framed the reveal

1:34:24 of the world to where you see the Citadel front and center,

1:34:27 and it's clearly in an alert state.

1:34:30 The four of us, the City 17 cabal, we'd done City 17,

1:34:33 which is a bunch of hard problems, we'd done Ravenholm,

1:34:37 which is a different set of hard problems.

1:34:39 We were like, I don't know,

1:34:40 four months from trying to lock down the game or something.

1:34:43 I mean, it was really late.

1:34:43 I mean, we had the dates wrong, but it was really late.

1:34:46 And there was a bunch of gameplay that that was

1:34:48 in there that hadn't really received a lot of development.

1:34:53 We were coming from working with pretty much every tool in the game

1:34:57 in City 17 to going back to basically you have a pistol and Metro Cops.

1:35:04 Not much really.

1:35:05 So John and I, John Guthrie and I, absolutely love exploding barrels.

1:35:09 So we just used them absolutely everywhere we could.

1:35:13 We set them up.

1:35:14 We loved ragdoll magnets and physics.

1:35:16 It's just one of my favorite tools.

1:35:17 It's basically you have a point in space that when a character dies,

1:35:21 they turn into a ragdoll and then you have a ragdoll magnet that says,

1:35:25 "Okay, if there is any ragdoll in this radius,

1:35:29 pull it towards me." And then you can give it a strength.

1:35:32 And so if somebody dies in an especially dramatic location,

1:35:35 you can have an especially dramatic death.

1:35:38 And we just use that absolutely everywhere.

1:35:40 I mean, you can use it to just pull someone directly down.

1:35:42 So in that bridge, the train bridge, I put magnets directly beneath their feet,

1:35:47 so it looks like they just,

1:35:49 like the planks beneath their feet broke and they just fell directly down.

1:35:52 That's right.

1:35:53 I mean, they died from a broken plank,

1:35:55 which doesn't make any sense, but they looked pretty cool.

1:35:57 It was so orchestrated, I guess,

1:36:00 because we didn't want it to, we didn't really have a lot

1:36:03 of time to have a lot of bugs come out of it.

1:36:07 (loud gunshots) (metal clanging)- Our team was the one that did the Citadel.

1:36:17 You know how I said that there was multiple

1:36:19 passes on the product over the course of the thing?

1:36:20 Well, Citadel had almost no pass.

1:36:23 And we were thinking about street war and we were like,

1:36:25 "Well, we just had this incredibly intense Strider fight.

1:36:29 How are we gonna outdo that, especially when we only got a month

1:36:33 to do this thing?" We then came up with the idea of like,

1:36:36 "Well, what if we made this just be 'you're god',

1:36:38 and this is the reward." "You've won the game.

1:36:41 Enjoy.

1:36:41 We're gonna like break all the rules and let you let you be God for a little

1:36:45 bit."- Picking up the enemies as ragdolls and throwing

1:36:49 them around really came outta that development tool,

1:36:51 because you could pick them up and move them around.

1:36:54 And people were like, "I don't know, that's pretty fun.

1:36:56 Why can't we do that in the game?"- It was either a week or a day.

1:37:00 It was incredibly quick.

1:37:01 I had every single mechanic of the Citadel mocked up.

1:37:06 The Combine ball was actually a very,

1:37:08 very heavy watermelon that would bounce around and eventually break up.

1:37:11 I had spikes in the wall that you could skewer them on, we had electric fields

1:37:14 that you can throw them in, we had

1:37:17 the zero-G fields that they would get stuck in.

1:37:19 And we had ripping off the machines off the walls.

1:37:21 All that stuff all came together within an incredibly short period of time.

1:37:25 Just a day or a couple days.

1:37:26 Very shortly after that people were like, "Eh,

1:37:29 we're having a real hard time with the AR2's alt fire.

1:37:32 So maybe we'll just steal this 'cause it's kind of cool

1:37:35 and we'll bring it earlier in the game." Game development never does

1:37:38 this, it never just materializes out of nothing in that short

1:37:42 of time and is just something that everybody's excited and happy about,

1:37:45 and it just felt really good to be a part of that.

1:37:51 (curious music) (curious music continues)- I started working on trying

1:38:04 to find another source of funding to keep Valve going.

1:38:09 I started working on a deal with a big

1:38:12 publisher that was for "Counter-Strike 2." So we're right there,

1:38:16 ready to sign, we're really kind of running on fumes at that point.

1:38:20 And then get a phone call and they're like, "Yeah, we changed our mind.

1:38:23 We're not doing the deal." I remember one conversation with Gabe,

1:38:27 which is he was like,

1:38:29 "So how fucked are we?" And I was like, "Yeah, we're kind of screwed.

1:38:34 We could probably reformat everything that we're doing,

1:38:37 but it means we lay off people." And we still had

1:38:40 the huge uncertainty of what is gonna happen in the lawsuit.

1:38:43 Immediately started chasing another publisher and were able to get a deal

1:38:49 done with another publisher for "Counter-Strike 2." When we got the deal done,

1:38:54 Gabe at that point was super into knives and as the deal gift for the parties,

1:38:59 he built a knife that was inscribed

1:39:02 "Counter-Strike 2." And so the deal structure was,

1:39:06 "After we ship 'Half-Life 2,' if you decide

1:39:09 that you don't want to keep moving forward,

1:39:11 then you can just decide to terminate the deal

1:39:14 and we'll pay the money back." That ended up happening.

1:39:17 So maybe a good choice because it took us a long time

1:39:20 to ship "Counter-Strike 2." So that we got done in probably May of 2003.

1:39:27 And then through some more depositions,

1:39:30 Vivendi figured out that we had gotten this new infusion of capital

1:39:35 and kind of lost their mind trying to figure out how did that happen?

1:39:39 Because clearly part of their strategy was running us outta money.

1:39:44 But what we decided was, "Hey, let's focus on the real nub of this and the thing

1:39:50 that's going to lead to a resolution.

1:39:53 Getting the judge to decide that we were right about the cyber cafe business,

1:40:00 and also to decide that we were right

1:40:02 that if you were distributing these properties without a license,

1:40:07 that that's copyright infringement." We had asked for the information

1:40:10 about what they were doing in Asia, including Korea,

1:40:13 and they, I think to try to make our lives more difficult,

1:40:18 produced a big chunk of documents in Korean.

1:40:22 They were in Korean.

1:40:23 I mean, everything from email to contracts to you name it.

1:40:27 And one of the things they were doing is just dumping huge amounts of discovery

1:40:30 on it to force us to go and just have somebody go and look through it.

1:40:35 It's like, "Oh, look, they ordered sandwiches again," right?

1:40:38 It has nothing to do with the lawsuit,

1:40:40 it's just an attempt to overwhelm us with tens of millions of pages of documents

1:40:44 on the assumption that there's nothing in it

1:40:47 that will be particularly problematic for their case,

1:40:50 that we'll eventually just either spend a bunch of money,

1:40:53 which plays into their hands,

1:40:55 or we'll just eventually give up because it's finding a needle in the haystack.

1:41:00 And then we found the needle in the haystack,

1:41:01 and that changed things quite dramatically.

1:41:06 [Karl] It just so happened that there was an intern, summer associate,

1:41:10 who was a native Korean speaker, and I said, "Andrew, hey,

1:41:15 can you look at this pile here and can

1:41:17 you just tell me what's in there?"- I would

1:41:21 guess that that particular intern probably went through a couple

1:41:24 thousand documents that day and only one was consequential.

1:41:27 And then they said, "Well,

1:41:29 there's these other emails between these vice presidents,

1:41:33 where one says something like, 'Hey,

1:41:35 we destroyed those Valve documents like you asked.'" And I said,

1:41:41 "Are you sure it says that?" And he said,

1:41:46 "Yeah, that's what it says." And so of course that led

1:41:51 to a conversation between me and the lawyer for the other side about,

1:41:56 "Hey, it looks like your guys are destroying evidence.

1:42:00 What's up with that?" They wrote back and said, "You're outta your mind.

1:42:03 Whoever translated doesn't know what they're talking

1:42:06 about." And I talked to Andrew again,

1:42:07 I said, "Well, it looks like we're gonna have kind of a fight about this.

1:42:12 We're gonna need an affidavit from you about your translation.

1:42:15 Can you tell me a little bit about your Korean skills?" He goes,

1:42:18 "Well, I'm a native speaker.

1:42:19 At UCLA, I majored in Korean language

1:42:22 studies." (producer laughing) And I said, "That's good.

1:42:25 That sounds like we're in good shape."- The Assistant General Manager

1:42:29 in Korea sending the letter to the General Manager of Korea saying,

1:42:34 "I have destroyed those documents related to the Valve case,

1:42:37 as directed," which I think is one of Karl's

1:42:40 happiest moments as a lawyer in his entire career.

1:42:43 Just stuff like this never, ever happens.

1:42:46 I did a lot of lawsuits for a long time, I never saw anything like that.

1:42:50 Never.

1:42:51 Maybe there was document destruction that couldn't be proved, but that was like,

1:42:57 "We're talking about it in writing." I'd never seen that before.

1:43:02 At that point, Judge Zilly said,

1:43:04 "All matters of fact are now according to Valve.

1:43:08 You don't get to contest any of those.

1:43:10 Now we're just discussing the settle, how much you're gonna have to pay and what

1:43:15 the damages are."- And Gabe was still pretty mad.

1:43:19 He was mad.

1:43:20 He really didn't want to settle.

1:43:22 He was mad enough that he wanted to keep going.

1:43:26 But I remember sitting down with Karl and having a conversation with Karl,

1:43:32 and I was like, "So Karl, what do you think?

1:43:34 Should we settle at this point?

1:43:36 Is (static) the right number?" And he was like, "Well, there's this old saying,

1:43:41 'There's a difference between being a pig

1:43:44 and a hog.'" (Scott Lynch laughs)- Karl was done

1:43:50 with the deposition and that was the only question

1:43:52 I asked him to ask that particular witness was,

1:43:55 "Is 'Half-Life 2' being replicated and shipped to retailers?" He said yes,

1:43:59 and the relief was, "Hey, no matter what, at least the game's gonna

1:44:05 get out to people," which is kind of a...

1:44:08 That's a pretty unusual way to find out

1:44:10 from your publisher that they're shipping your game to retail.

1:44:13 (gentle music) (gentle music continues)- I've kind of been

1:44:26 thinking about this for the last few days, and if there was one thing that I can

1:44:29 point to that made us successful over that development period,

1:44:34 and I think that it was really this culture that Gabe created

1:44:37 where we were hiring people that were senior enough to be managers,

1:44:41 but also passionate enough to still be individual contributors.

1:44:44 They were willing to stake their credibility and their name

1:44:48 on these big features and sometimes spend years working on those features.

1:44:53 Yeah and it was six, seven days a week, 10 to 14 hours a day.

1:44:58 It wasn't always, I would say a good 18 months at least.

1:45:04 But a lot of good memories though.

1:45:06 I mean, everyone in that room had such great sense of humor.

1:45:09 We all, we knew each other's sense of humor, we trusted each other,

1:45:13 we respected each other enough that we could rib

1:45:15 each other without stepping on each other's toes and stuff,

1:45:19 and it just worked pretty well.

1:45:20 It was just bigger, more pressure.

1:45:24 And I think that seriousness in some ways, there was still humor,

1:45:29 but that seriousness kind of infuses into the game.

1:45:32 You feel it's a heavier game.

1:45:34 It's the most cohesive product feeling I can think

1:45:38 of at scale that we've done as a company.

1:45:41 The whole company was just aligned and got it done.

1:45:44 I think back on this experience, and it was a really hard time,

1:45:48 but I feel super grateful because we were supported

1:45:51 to make the very best game that we possibly could.

1:45:55 And given the time to do it, we're gonna ship it when it's done,

1:45:58 even when we made unreasonable promises about when that might be in advance.

1:46:03 And it was also at a time before the game

1:46:05 industry was quite as specialized as it is now,

1:46:08 where people got to wear multiple hats

1:46:10 and contribute in all these different ways.

1:46:13 So it was a rare life experience and I'm

1:46:16 really grateful to have been part of it.

1:46:18 (gentle music fades) (eerie music) (eerie music continues)- Yeah,

1:46:37 I mean part of the fallout of "Half-Life 2" shipping was the realization

1:46:41 that we can't do six-year long bet the whole company bets over and over again.

1:46:45 And if we get it wrong one time,

1:46:47 we're a big crater in the ground, as Gabe would say.

1:46:50 And so we started to think about episodic releases.

1:46:53 The first idea when they pitched it to us, like,

1:46:56 "We're gonna do this episodic content, we'll do one every," how much, 12 months?

1:47:01 Yeah, we were gonna have alternating teams come out every year and a half,

1:47:04 something like that.

1:47:05 [Douglas] So I was like, "Yeah, this is neat," but then it just got outta hand.

1:47:10 [David S] So when they decided to start playing with episodic content,

1:47:14 the promise for me that I felt was, "We can build smaller pieces,

1:47:19 denser pieces faster, and we know what we're doing,

1:47:21 we're gonna be really good at this."- So

1:47:23 a lot of it was like, "Do more with less,

1:47:26 keep a bunch of the existing stuff and retweak it to make some

1:47:31 new interesting scenario." Like probably the biggest

1:47:33 real new thing for episode one

1:47:35 as far as technology was just having Alyx with you throughout the entire

1:47:38 thing and having her behave in a way that was believable and fun.

1:47:44 She's with you for a few distinct parts in "Half-Life 2,"

1:47:47 but they're always pretty heavily scripted and not really in combat.

1:47:49 Having her there and able to actually be a useful companion,

1:47:54 so there's a lot of little tricks that we did

1:47:56 to try to make that as fun as possible for players.

1:47:59 She doesn't have supernatural knowledge,

1:48:00 she's not just finding things without you lighting them up,

1:48:04 but if you throw a flare out or you look at 'em with a flashlight,

1:48:07 it's like, "Okay, she's gonna go for that target because she can

1:48:09 see it."- [Douglas] We were cranking on that, and it took forever.

1:48:13 We were not just partying, hanging out.

1:48:16 No, I was pretty much crunched that whole project, I think.

1:48:19 Yeah, we were crunch mode.

1:48:21 And then same thing with episode two.

1:48:24 We started at the same time as episode one,

1:48:27 and so we ran for an extra year that way,

1:48:30 which allowed us to stretch and push more.

1:48:35 So the Hunters were initially something that we wanted to have

1:48:38 a more sort of frenetic pursuit built into the design of the character.

1:48:43 So it was really focused on, at the time,

1:48:45 creating a creature that could follow the player into more types of spaces.

1:48:49 One of the questions was, how do we keep Alyx outta your way during

1:48:53 this really intense high action scene where the player has agency

1:48:57 to make whatever decision they want about where they want

1:48:59 to be inside the house and that sort of thing,

1:49:01 and that's when we got the idea to have her sort of camp out by a window and she

1:49:05 would fire out the window and let you know

1:49:07 if she saw any Hunters or soldiers out that window.

1:49:10 And if the player started to spend time at that window,

1:49:13 she would find another one to go be at.

1:49:16 I mean, so Dario Casali was trying to create this final fight in episode two.

1:49:22 The Striders always reminded me of AT-ATs

1:49:24 in "Empire Strikes Back." And of course we wanted

1:49:28 something fresh because just rocketing a ton of Striders

1:49:31 was gonna be a repetitive after a while.

1:49:33 So that's when we started collaborating,

1:49:35 talking about the Magnusson device and the Hunters and how would

1:49:40 those all come together to make this sort of an overwhelming 25,

1:49:45 30 minute battle set piece.

1:49:48 (loud shot) (sci-fi laser whirring) (loud laser

1:49:52 shots) (loud explosion)- The citizens and the Hunters,

1:49:57 and to a lesser extent the Strider,

1:50:00 are all secretly in cahoots to create this tension.

1:50:05 Because what we always wanted was, "Gotten the Magnusson device on the last guy,

1:50:10 are you gonna be able to shoot it?

1:50:11 Oh my God, the rocket's gonna be destroyed." And then like, boom,

1:50:14 you hit it so that when everyone comes out running, you really feel heroic.

1:50:18 [Dario] You have the car and you have a big space and you're not guided anyway.

1:50:22 You have to go and take care of this threat in a way that you think is best.

1:50:27 I think it pays off, it makes the player feel like, "I did this myself.

1:50:31 I wasn't guided by level designer at all,"

1:50:33 because we give them so many options, right?

1:50:35 You can kill the Hunters with driving the car into them,

1:50:39 you can pick up the log and throw at them,

1:50:41 you can pick up a Magnusson device and throw

1:50:43 it too soon and they'll shoot it down.

1:50:45 You have to learn, "Oh, yeah,

1:50:46 I have to kill the Hunters first." You can use rockets to kill the Hunters.

1:50:49 You have options and you have time and you have space,

1:50:53 but we dial the tension and the pacing up pretty slowly over time.

1:51:00 (loud shot) (intense crashing)- [Jay] Yeah, it is pretty cool,

1:51:05 all of of cinematic destruction stuff that happens in there.

1:51:09 It was so hideously faked because basically we would send

1:51:12 a Strider who would do that sort of low scuttling run,

1:51:15 and he would just sit there and wait,

1:51:17 just staring at this building until the player

1:51:19 could see the Strider plus the building.

1:51:21 Yeah, till it was framed up correctly.

1:51:22 He would literally sit there forever.

1:51:24 (loud explosion) But we dial the tension

1:51:29 and the pacing up pretty slowly over time.

1:51:33 The play tests we like the most is when the Strider's,

1:51:36 he's getting ready to shoot the rocket,

1:51:38 and then the last Magnusson device goes off, and then he's down.

1:51:41 (sci-fi whirring) (loud explosion) (people celebrating)- Nice one, Freeman!

1:51:49 We did it!

1:51:50 We held them off!

1:51:52 [Tom] People were failing left and right,

1:51:53 and the team was calling for the map to be cut.

1:51:57 [Dario] Well, I took a lot of heat for that.

1:51:59 Both of us did.

1:52:00 Yeah, we just had to see it through.

1:52:02 We were gonna do an arc of three, of course,

1:52:06 because that's just how you think of stories in those terms.

1:52:09 And then you want the high point to come at the end of the second one,

1:52:12 and Gabe at one point saying,

1:52:13 "Who's the important character you're gonna kill?" I'm like,

1:52:16 "Lamar is going down.

1:52:17 This is gonna be it.

1:52:19 I want episode two to end with Lamar

1:52:22 floating into space and everyone cries." And Gabe's like,

1:52:26 "That's not good enough."- No!

1:52:27 No!

1:52:30 The episodes were, a lot of that was

1:52:32 an experiment on what's the right amount of episode.

1:52:36 Episode two was longer than episode one and bigger,

1:52:40 and people like that better, but-- It also took longer.

1:52:45 It took longer to make and people didn't like it as much

1:52:48 as they liked "Half-Life" or "Half-Life 2" kind of scope, right?

1:52:52 Yeah.

1:52:52 So by saying episode, are we solving a lot of problems here

1:52:55 or are we just making a sideways trade off?

1:53:00 I think that's maybe the way I felt about it.

1:53:03 Even into episode three,

1:53:04 I still don't know what that would've been if we'd built it,

1:53:08 because it hadn't been built.

1:53:10 That was the feeling of excitement of something I

1:53:14 can't even imagine is gonna happen with as a team.

1:53:17 I was not imposing a top-down,

1:53:20 "This is what we must do to tell our very important tale." It's like,

1:53:26 "Oh, we have new features?

1:53:29 What kind of story can we do with these now?"- Well,

1:53:32 I worked on episode three and I was working on this gun called

1:53:36 the Ice Gun that basically let you create amorphous shapes out of ice,

1:53:44 so you could raise a little ice wall in front of yourself.

1:53:48 It would be attached to the floor,

1:53:50 and then the Combine soldiers could shoot holes in it

1:53:53 and it would shatter and break kinda like glass, I guess.

1:53:56 So that was the primary mode,

1:53:57 of you could build ledges for yourself to get down cliff faces.

1:54:03 It another mode that was kind of like a Silver Surfer mode,

1:54:06 where you would extrude the ice in front of yourself

1:54:09 and then run along it and use it to cross gaps,

1:54:14 go over chasms and things like that.

1:54:16 What else do we have?

1:54:17 It was set in the Arctic 'cause it was set around Borealis.

1:54:21 [Jay] I remember the little blobby, squishy.

1:54:23 [David S] Oh, yeah, yeah.

1:54:25 Brian was working on the blobs.

1:54:26 So we had the isosurfaces that we ended up using in "Portal" for the paint.

1:54:31 So we had that as an enemy.

1:54:34 It was a blob enemy, and it could change its shape and you

1:54:39 could choreograph its shapes with some content tools.

1:54:44 And then it could also split into little

1:54:48 hoppy blobs that were almost like little Headcrab-y things,

1:54:52 but they were little splats that would hop around.

1:54:56 And we were doing all kinds of stuff with that.

1:54:58 It could pass through grates.

1:55:00 [Jay] You could get more gameplay out of something where

1:55:02 you had maybe a couple of creatures with simpler behaviors.

1:55:06 Oh, yeah, yeah.

1:55:07 And then you would spend time combining

1:55:09 them and combining them with other game mechanics.

1:55:13 Story was never the boss of anything.

1:55:15 It was always, we worked on a peer-to-peer thing.

1:55:19 We were, "What can we do?

1:55:21 What do you think would be cool?"- It was still a collection of playable

1:55:27 levels in no particular order and a collection

1:55:31 of story beats and story concepts.

1:55:33 And we were still like,

1:55:36 probably another six months we would've had a critical mass

1:55:40 of mechanics and then start putting them in a timeline.

1:55:45 I think probably a year and a half easily, depending on how ambitious we got.

1:55:50 Could have been two and a half more years.

1:55:54 Yeah.

1:55:54 It's hard to be a lot less than two years.

1:55:57 Yeah, yeah.

1:55:58 Yeah.

1:55:58 Yeah, I think we were six months in on it

1:56:01 when we moved to "Left 4 Dead." I don't know,

1:56:03 we were moving at a pretty brisk pace too.

1:56:06 I thought that we needed to go much

1:56:08 bigger on episode three or do something else,

1:56:12 because I don't know if this is the right way to describe it,

1:56:15 but element fatigue?

1:56:16 I think we had really explored a lot

1:56:19 of what made sense in the "Half-Life" universe and setting.

1:56:23 Arkane was building the "Ravenholm" game,

1:56:25 and even they were having trouble doing cool new stuff with this tool set.

1:56:30 And if those guys can't figure out a bunch of cool

1:56:32 stuff to do with this, I think we're running out of fuel.

1:56:35 Episode three's coming along, it's like, "Well, what's the no big thing here.

1:56:38 It's gotta be bigger than all the stuff we just did,

1:56:40 even episode two." And it's like, "Oh my God," you know?

1:56:42 It just becomes this-- Are we allowed to cry in this documentary?

1:56:45 That's what I wanna know.

1:56:46 "Left 4 Dead" needed kind of an all hands on sort of an effort to ship.

1:56:54 And so we put down ep three to go help "Left 4 Dead," and it was a really

1:57:01 tight shipping schedule and it required a lot

1:57:04 of work from everybody to get out the door, and it was worth it.

1:57:08 Yeah, "Left 4 Dead" came out...

1:57:09 "Left 4 Dead" came out great, but it took long enough,

1:57:13 and this is the tragic and almost comical thing about it,

1:57:17 was it took long enough that then

1:57:18 by the time we considered going back to episode three,

1:57:22 the argument was made like, "Well, we missed it.

1:57:24 It's too late now." (both chuckling) And we really need to make

1:57:28 a new engine to continue the "Half-Life" series and all that.

1:57:31 And now that just seems, in hindsight, so wrong.

1:57:36 We could have definitely gone back and spent two years to make episode three.

1:57:40 And you can't get lazy and say, "Oh,

1:57:43 we're moving the story forward." That's copping

1:57:45 out of your obligation to gamers, right?

1:57:48 Yes, of course they love the story.

1:57:50 They love many, many aspects of it, but sort of saying that your reason to do

1:57:53 it is because people wanna know what happens next.

1:57:56 We could have shipped it.

1:57:58 It wouldn't have been that hard.

1:57:59 The failure was...

1:58:02 My personal failure was being stumped.

1:58:05 I couldn't figure out why doing episode three was pushing anything forward.

1:58:11 But yeah, I mean, some of those people made

1:58:13 "Left 4 Dead 2" and then "Portal 2."- Yeah, we did other things instead.

1:58:17 It's not like that choice would've been free.

1:58:21 Right.

1:58:21 In terms of the players,

1:58:23 they may have gotten that and not-- And given up something else.

1:58:25 Not something else, yeah.

1:58:26 So it's hard to say.

1:58:28 [Kelly] But a lot of us have been doing "Half-Life" for eight plus years.

1:58:33 We saw a bunch of really compelling reasons

1:58:35 to go explore more multiplayer based projects too,

1:58:39 because not only just what customers were doing, but internally,

1:58:43 we found ourselves wanting to play a lot of multiplayers.

1:58:47 It sounds silly to say it 'cause we're

1:58:49 a relatively big company that has a bunch of stuff,

1:58:52 but we're not that many people.

1:58:53 There's 300-something people here.

1:58:55 And so we're working on a bunch of different projects and it's like,

1:58:59 people are passionate about "Portal," people are passionate about "TF," people

1:59:02 are passionate about "Left 4 Dead" and "Half-Life" and "CS" and "Dota,

1:59:08 "Deadlock." I mean, there's a lot of things there.

1:59:10 And I also work on our hardware, so I've worked on Steam Deck and the original

1:59:14 Steam controller and a tiny bit on VR.

1:59:17 Unfortunately for any decision we make, somebody is gonna be unhappy,

1:59:21 but hopefully we're making a bunch of people happy with the choices we do make.

1:59:25 If it weren't for the episodes,

1:59:26 there'd be no such thing as "Dota," and I know that sounds really weird,

1:59:29 but it was the things that we learned in developing

1:59:31 the episodes that led to "Team Fortress" to more rapid updates.

1:59:35 I think everybody that worked on "Half-Life" misses the working on that thing,

1:59:39 but it's also hard not to be like,

1:59:41 "Man, I've kind of seen every way that you can fight an Antlion," or whatever.

1:59:47 And so you wanna get some space away from it

1:59:49 until you can come back to it with fresh eyes.

1:59:53 It was easy to think about VR being a vehicle for "Half-Life" because that was

1:59:57 a big technological innovation and kind

1:59:59 of a core reason for that product's existence.

2:00:01 And I think like one of the things we have

2:00:03 internally tended to attach to the "Half-Life" IP is innovation.

2:00:09 Gameplay innovation is oftentimes enabled by a technological innovation.

2:00:13 Clearly there was a ton in "Half-Life" 1 and 2, and so yeah,

2:00:16 it's an interesting challenge moving forward to think about

2:00:19 what that means for future "Half-Life" stuff, for sure.

2:00:23 The ending of "Half-Life: Alyx" is somewhat a self-critical realization,

2:00:26 so that was super satisfying and all credit to the people

2:00:30 who were specifically involved in that decision and those sets of designs.

2:00:33 I think that "Half-Life" represents a tool

2:00:36 we have and promises made to customers

2:00:39 to capitalize on innovation and opportunities to build

2:00:44 game experiences that haven't been involved previously,

2:00:47 and I think that there are no shortage

2:00:49 of those opportunities facing us as an industry right now.

2:00:53 (gentle music) (music fading out)- [Kleiner] Oh, fiddlesticks.

2:01:09 What now?

2:01:14 (glitchy audio SFX)