Google Games Chat #5

Uploaded by GoogleDevelopers on 27.09.2012

MALE SPEAKER 1: Shanghai GDG the is a very interesting
developer community.
FEMALE SPEAKER 1: I'm glad somebody
has asked this question.
MALE SPEAKER 2: This is where the magic happens.
FEMALE SPEAKER 2: This is primarily a question and
answer show, so if any of you out there
would like to ask questions.

TODD KERPELMAN: And we are live.
All right.
Welcome to Google Games Chat number
five, still not canceled.
COLT MCANLIS: Alto cinco.
TODD KERPELMAN: So I guess we should start by apologizing
for last week's episode not making it onto YouTube.
WOLFF DOBSON: Our awesomeness cannot be compressed.
COLT MCANLIS: It was probably for the better that that one
didn't make it on.
TODD KERPELMAN: In retrospect, I think we probably had a
little too much gratuitous nudity
that probably violated--
TODD KERPELMAN: I thought it was artistic.
I thought I was artistic.
But apparently, those--
WOLFF DOBSON: We don't have a release for those
parts of our bodies.
COLT MCANLIS: The Supreme Court will tell you what's
artistic and not.
TODD KERPELMAN: I know it when I see it.
I'm just saying, I thought it was tastefully done, and it
was a beautiful thing.
But apparently those people at YouTube just did not agree.
And so those of you who missed the last episode, episode
number four, it's been lost in the ether--
WOLFF DOBSON: You missed out.
That's what I'm saying.
I don't think it'll ever happen again.
JOHN MCCUTCHAN: A whole lot of Todd.

TODD KERPELMAN: So we are going to move off of this
uncomfortable topic and go into some introductions.
To my left is Daniel Wolff Dobson, Dr.
Daniel Wolff Dobson.
WOLFF DOBSON: I'm a doctor, and I play one on Google
Developers Live.
TODD KERPELMAN: To my immediate
right is Colt McAnlis.
Say hi, Colt.
TODD KERPELMAN: And to his right is John McCutchan.
JOHN MCCUTCHAN: Hey, everyone.
So actually I want to start by asking, you're a doctor?
TODD KERPELMAN: I did not know this until--
WOLFF DOBSON: I'm not the kind of doctor that does surgery,
although maybe the occasional unnecessary surgery.
But no, I have a PhD in computer science in artificial
I worked on intelligent user interfaces, so interfaces that
had an idea of what it was you were trying to do.
And then because it had a model of what you're doing, it
could help you do it better, more or less.
Old AI PhDs are not awesome reading, because a lot of it
is like, well, yeah, that would be a good idea.
And now, here, 10 years later, it was a good idea.
We should have done that.
TODD KERPELMAN: Is there an example of this put into
practice that you'd be able to point to and say--
WOLFF DOBSON: You don't want to ask a PhD to talk about
their thesis.
TODD KERPELMAN: That is probably true.
WOLFF DOBSON: That gets ugly quick.
But I will say that after I did some post-doc work doing a
lot of interesting things, rendering Mars Rover data and
things like that, I was like, I think I'm done with this,
and joined Sega Sports and started working on basketball.
TODD KERPELMAN: Sounds way more fun.
WOLFF DOBSON: Which was very much AI, although it's
interesting, because so much of the AI in console games,
especially of that era, is very much reactive planning,
which is a classic theory that came out of the University of
Chicago about control.
And what it really was, was this reactive planning from
the early '90s, but elaborated to the n-th degree, where we
just had tons of code for every possible situation,
every frame, deciding what should the
basketball player do next?
Well, is he doing this?
Hes' doing this.
We have hundreds of heuristics that are all watching what--
should he pass, should he not pass?
If I had the ball, would I have a better shot than if I
passed it to Colt and he had the ball?
If I pass the ball to Colt, would it get intercepted?
All these things, and you're calculating that every frame.
COLT MCANLIS: If you hand me the ball, I
generally take the shot.
That's how it works.
WOLFF DOBSON: Well, and then, you know--
TODD KERPELMAN: Doesn't matter where he is on the court.
WOLFF DOBSON: I mean, you actually put in emotion things
like that, where it's like, a guy like Kobe, he
holds onto the ball.
And it has some funny side effects, which is if you have
one character who's really out of balance, he'll never pass.
Because no matter what, he is the best guy with the ball.
So yeah, I'm rarely a practicing scientist now,
although I still review papers for the AAAI Conference on
computer games.
There's a whole lot about you I did not know.
Are neural networks-- like are those passe now?
People still do neural networks.
JOHN MCCUTCHAN: Wasn't the cat detector a neural network?
WOLFF DOBSON: No, neural networks are still very
There's lots of--
I mean, I had sort of good old-fashioned AI kind of
stuff, where we are doing symbolic
representations of things.
But there is also--
machine learning is really big.
Machine learning in games is very interesting because
machine learning is about optimization.
A lot of times, it doesn't actually sell very well when
you're doing it in a game.
So if you're playing a football game and you're doing
machine learning on what kind of plays the guy is doing or
something like that, it will gradually start getting better
than you or being able to predict every move you're
going to make.
And that's not actually that much fun.
TODD KERPELMAN: It's not fun to continually lose?
WOLFF DOBSON: Well, you--
COLT MCANLIS: Ask the programmer.
It's super fun.
WOLFF DOBSON: You don't see it, and you don't--
if you're doing this gradual optimization process, you
don't have any place where you can call attention to and say,
this is the moment in which I learned something, and
therefore, I will tell you this.
Which is why machine learning--
so if you have a machine learning thing, a neural
network that's driving your car, you can't really tell
it's being driven by a neural network unless you annotate it
with a whole bunch of stuff, like, oh, he's missed the
curve, or something like this.
So machine learning is really interesting.
It's really interesting in games.
But again, it tends to show up in quiet ways that you can't
quite see it.
You have showy stuff instead, to like, look at this AI.
It's being so smart.
That tends to be more of good old-fashioned AI.
COLT MCANLIS: What do you feel about modern games today?
Do you think that too much of the AI isn't really AI?
Because I know a lot of it is pretty much just pre-canned.
WOLFF DOBSON: Oh, there's a whole discussion of what--
every time something gets popular, a lot of times, the
AI establishment says, well, that was never
really AI all along.
That was an optimization problem or
something like this.
For a long time, it was believable agents.
You know, how could you make sidekicks who were believable,
who seemed like they were alive and things?
And then at some point, we started making games then had
believable agents, dogs that acted like
dogs, things like that.
There's interesting research--

No, but like CatZ, C-A-T-Z, the little pet cats that you
could have from Broderbund.
I mean, those were moments, like, believable agents.
Yeah, we can do that.
Eh, that's not really AI.
And it goes all the way back to computer games learning, AI
learning how to play chess.
At some point, oh, that's not chess.
That's just a search problem.
Or that's not AI, that's search.
COLT MCANLIS: Well, that's the goal, right?
I mean, chemistry is really just physics,
which is really just--
WOLFF DOBSON: Yeah, exactly.
Well, what I would say about games not really having real
AI, what I would actually say about that is that you only
need to add as much AI to your game to make it fun.
So there are situations like if you play Yakuza or
something like that, the AI for the guys getting out of
your way as you're running through the crowd is sort of
silly looking.
They sort of stagger and they pop to these weird animations,
and all this other stuff.
But you know what?
It's good enough.
I can get through the crowd.
They get of my way.
And so I think a lot of people, especially when I was
in grad school, I would play these games and think, oh my
gosh, if they just used our awesome super-advanced
algorithms, it wouldn't do this.
And then when you think about it, it's like, it's not really
worth the money to get it that beautiful or to
make it that seamless.
And if it isn't perfect and seamless, you don't even
notice how good it is.
JOHN MCCUTCHAN: Not just the money, but also the
computation, right?
Like, how much computational effort is it worth--
TODD KERPELMAN: How much can you spend towards--
JOHN MCCUTCHAN: Yeah, towards crowd algorithms when you
don't really notice.
WOLFF DOBSON: Well, exactly.
I mean, if it's happening--
or a classic example of this is playing
first-person shooters.
If your sidekicks are not--
if you're playing a single player game and your sidekicks
are not visible, they could be doing anything.
And in fact, they usually are.
They're teleporting around.
They're doing crazy stuff.
Because it does not matter what they're doing, because
you can't see them.
TODD KERPELMAN: I have noticed
occasionally if some follower--
I think this happens in Skyrim-- some follower isn't
able to path-find to where I am, just keep going and
eventually they turn the corner.
WOLFF DOBSON: Yep, and they're right there.
And if you think about that, it's good enough.
If you're involved in playing your game, and the guy is
there at the right moment, you don't really think about, did
he successfully path-plan his way from point A to point B.
COLT MCANLIS: Now there's been interesting cases of that.
Wasn't it Fable or Fable II?
One of the two had a mission you could send your horse on
or something like you could move a zone, and move another
zone, and then the horse would still try to follow you
through these zones and in the process would of course take
damage and gain experience points along the way.
And there was this interesting video on the net where someone
was like, OK, horse, stay.
And then he went like two zones away, and he whistled
for his horse.
And the horse came back two levels higher or something
like that, because it had killed mice along the way, or
some weird stuff.
WOLFF DOBSON: That's beautiful, and as an AI
developer, I think that's so cool.
As somebody who would be budgeting that game, I would
say to myself, why did we spend all this money on
off-screen horses?
If we could get three more good horses on screen, I would
totally skip the horse playing the game by itself.
TODD KERPELMAN: All right, we're going to move along to
our next topic.
TODD KERPELMAN: No, this is totally--
well, this great, right?
WOLFF DOBSON: Clearly, we there's an AI episode.
TODD KERPELMAN: We could an entire episode
just on game AI.
There's a lot more to talk about.
But we're going to move into our five-for-five section--
WOLFF DOBSON: Five for five.
TODD KERPELMAN: --where we--
yeah, we pick five topics that we're going to talk about for
five minutes.
I actually have my stopwatch set for 5 minutes, 30 seconds,
because I want us to have a little extra time.
And we always go over anyway.
COLT MCANLIS: He's feeling generous today.
So I'm going to start with a comment that Colt had made
right before this episode, which is basically declaring
that 3D was dead.
Why don't you back up this claim, and then we'll all tell
you why you're wrong.
WOLFF DOBSON: Unlike our completely uncontroversial
statements about game AI, now we're going to say
whether 3D is dead.
COLT MCANLIS: So let me preface this.
So I'm a graphics programmer, traditionally.
Before I came to Google, I spent time in the industry as
a system and graphics programmer, like hardcore.
The question that I had today was, where does 3D fit in our
modern gaming ecosystem?
We saw over the trend of a couple console cycles that the
amount of cost to create these movie production games
skyrocketed, mostly due to the advances in 3D rendering.
But what we've seen now is that the move to casual mobile
social doesn't require the same sort of Hollywood
backdrop, right?
We've seen better experiences coming out of 2D environments
and more creativity coming from low-cost development.
In addition to that, with the move to mobile--
your little stopwatch right there-- we actually find that
these things aren't really set up to do 3D well.
It burns through the battery.
It burns through the chips.
It takes a lot more away from the user.
So the question is, besides the console, is 3D really the
next step for tablet mobile?
Or is that really an ecosystem that should be
dominated by 2D?
And the same thing for the web, is there a spot
for 3D on the web?
JOHN MCCUTCHAN: So production values for 3D content are
extremely high.
And mobile titles sell for very little.
You can't really sustain a multimillion dollar game
budget when you're selling your unit for $0.99.
WOLFF DOBSON: Well, unless you're making Infinity Blade,
and selling it for however much it is now.
JOHN MCCUTCHAN: $9.99 or something like that.
WOLFF DOBSON: It could be that 3D is the thing that persuades
people that this game is worth, yeah, $4.99, instead of
$0.99, or instead of free to play.
But I mean, I still think that modern, really high-quality
AAA games sell for $70 a pop.
And they are in development for four years and have huge
armies of programmers and artists.
And even selling a game for $10, it's hard to recoup those
costs, I think.
WOLFF DOBSON: Well, but some of that--
some of that is tools.
If you have good tools, you don't need that army of
What you need is an army of artists.
COLT MCANLIS: Unless you're writing a
really good AI algorithm.
JOHN MCCUTCHAN: Then it does it for you.
TODD KERPELMAN: How much of this--
WOLFF DOBSON: Actually, in all seriousness, I read a paper
about that a while ago where they were actually generating
buildings complete-- filled with everything down to
pencils on the desk from basically AI algorithms.
JOHN MCCUTCHAN: That's great if you want to have kind of--
if you have a whole bunch of geometry, like a city, that
can very easily be procedurally generated.
We all kind of know what a city looks like.
But gamers of 3D games and gamers expect very cinematic
experiences, which is essentially like you are
scripting a movie scene that the player
gets to play through.
And in that case, it's difficult to procedurally
generate that.
Because the helicopter has to come in now and blow this city
block up at that moment.
TODD KERPELMAN: See, I don't know all this has necessarily
anything to do with it being in 3D versus people just sort
of wanting big set piece kinds of games.
I mean, there's obviously--
if you take something like--
why am I blanking on it--
Obviously, a 3D game, obviously a still simplistic
stylized art style, was done fairly inexpensively.
It's not like there's anything inherent in 3D that I think
sort of makes it more expensive, so much as people
kind of want the big games.
They want the big experiences.
And if somehow 3D had never been invented and we were
still doing 2D, we would be complaining about the cost of
hiring all these animators to hand draw all
these cells and how--
I think we'd be talking about resolutions are getting
bigger, so sprites are getting bigger, and my god, everyone
wants to customize everything, and we
want 60 frames a second--
WOLFF DOBSON: I think there's an alternate universe, like a
TV show, if 3D was never invented.
TODD KERPELMAN: Those guys would be saying, if only
things were in 3D then we would have to span--
WOLFF DOBSON: If only I had a Z buffer.
TODD KERPELMAN: --90 frames drawing write this walk cycle.
COLT MCANLIS: Let's do it this way.
Google fires you today.
You start a game company.
Is your game 3D or 2D?
WOLFF DOBSON: Depending on how this show goes.
COLT MCANLIS: Is your game 3D or 2D?
You start a game studio.
Google fires you.
Is your game 3D or 2D?
WOLFF DOBSON: Oh gosh, it absolutely comes down to what
I'm building.
Right now, is it 3D or 2D?
TODD KERPELMAN: Eh, 2 and 1/2 D.
WOLFF DOBSON: I would probably do 3D because it would be a
differentiating factor.

TODD KERPELMAN: Daniel Cook had a really interesting
point, which was like, if you're a small budget
developer, you can't compete on massive art budgets, so you
go for the stylized stuff.
And when you kind of stop and think about it, every indie
game, whether it's 3D or 2D, something like Fez, is still
kind of done in sort of a stylized art style.
So that you don't actually care whether--
JOHN MCCUTCHAN: It's funny that you bring that up,
because this week, Mike Acton just asked his Twitter
followers, like hey, do you think that the push towards
realistic graphics is improving or harming the
quality of games?
And the very nonscientific poll came back saying people
don't like it as much.
They prefer the stylized approach.
I would have to say I often like games that are stylized.
I think one of the prettiest games I've ever played was
Wind Waker.
That game is just beautiful, and it looks nothing like real
life, but it looks like an awesome cartoon.
WOLFF DOBSON: Going back a little bit, about the tools, I
mean, this is--
like part of the reason why, if I were starting my game
studio, I would think about 3D because everybody can do 2D.
And if I have the special skills, and find the right
people, and invest in the pipeline at the beginning, I
mean this is what makes--
why 2D is cheap and 3D is expensive.
3D, you've got to set up a pipeline.
You've got to get a rig, animation, all this stuff.
And you have to figure out how to get those into your game in
some sane way.
And there are tools like Unity where they do
some of that for you.
But you still have to build the pipeline from your
modelers to the motion models and to the game finally.
And that's very expensive.
That's a big upfront cost.
With a 2D game, dude, I've just got to get some sprites
up there, and that's all I need.
But for every level I add onto the 3D game, I have to do a
lot less work than for every level I have to add on--
TODD KERPELMAN: That is true.
WOLFF DOBSON: Each new animation costs exactly the
same for a 2D game, but each new animation in my 3D game
costs almost nothing, because I'm just
moving the rig around.
TODD KERPELMAN: All right, well, we're going to move on.
We're out of time Next topic, Dart, Dart as a language.
JOHN MCCUTCHAN: Yeah, it's pretty sweet.
TODD KERPELMAN: -we've been chatting about it quite a bit
here at Google.
COLT MCANLIS: It's exciting.
It's a really cool language.
TODD KERPELMAN: I mean, how do we think it's going to affect
the game industry?
How do you think game developers
could make use of Dart?
WOLFF DOBSON: You know who really knows about this is JJ.
Don't say JJ's name aloud.
If you do, if you say it three times, he appears.
JJ BURNS: Did somebody say my name?
COLT MCANLIS: How long have you been down there?
TODD KERPELMAN: Oh my god, he lives there.
JJ BURNS: You really don't want to know.
TODD KERPELMAN: Dude, don't look in that coffee can.
We're going to start.
JJ BURNS: Howdy.
TODD KERPELMAN: How's it going?
WOLFF DOBSON: Did not expect that.
JJ BURNS: Nice to be here.
COLT MCANLIS: JJ, by the way everyone, is--
introduce yourself to the crowd.
JJ BURNS: I'm JJ Burns I'm a developer advocate for Dart.
So why don't you tell us why should game developers care
about Dart?
What do they need to know about Dart?
COLT MCANLIS: Does it increase your cloaking ability,
JJ BURNS: Well, the thing about building web games these
days is that it's hard.
And the larger the game is, the harder it is.
Because JavaScript wasn't really built for
coding in the large.
When it first came out, web apps didn't really exist.
And so Dart tries to retrofit and fix a lot of those
problems with optional static typing.
I like to say that it's kind of a scalable language in that
it scales with the size of your project.
If I wanted to build a game with 100 developers and a
million lines of code, I'd want to do it in Dart, not
and I guess we sort of talked about this before in previous
discussions, but Dart does compile essentially into
JavaScript if you are on a browser that does not support
native client.
JJ BURNS: That's right.
There's a VM that's been integrated into a fork of
Chromium, and that's called Dartium.
But since almost no one runs Dartium, we have it compiling
down to JavaScript.
And one of the interesting things about is the Dart to JS
compiler can actually compile JavaScript--
TODD KERPELMAN: Hold on, Jimmy's telling us something.
What's that?
Oh, you can pick it up, hey.
That's why we had this mic sitting here this whole time.
In case some random person living in our
studio popped up.
JOHN MCCUTCHAN: In case somebody has cut a hole in the
back of the studio and crawled through.
COLT MCANLIS: What is this thing?
Why is this over here?
Like I'm leaning away.
I thought it was a stage mic.
TODD KERPELMAN: Anyway, let's--

JJ BURNS: The Dart to JS guys, they're pretty amazing in that
they're not trying to do just like straightforward
one-to-one translation between Dart and JavaScript, but
rather, they're applying all kinds of interesting
So a lot of times, the JavaScript that they can
compile down to is actually faster than the JavaScript
that you could write by hand.
Because they can do all kinds of tricks like
in-lining and so forth.
And so you'll see this once in a while.
If you end up looking at the generated JavaScript, it's
like whoa, look at what they did to me my code.
That's not what I wrote.
But it's kind of brilliant.
COLT MCANLIS: So for game developers out there who don't
have exposure to Dart, but come from the web, what are
the big three things they should know?
If they're an action script developer or a Flash
developer, what are the three things they should know that
make Dart awesome?
JJ BURNS: Let me see.
I would think of it in terms of the three types of
So if you're an ActionScript 3 developer, I think that you're
going to find Dart very familiar.
Because syntactically, it's somewhere between
JavaScript and Java.
It's got the optional static types things.
So I think that the ActionScript 3 developer will
find it very comfortable.
Now if you're a JavaScript guy, you might be used to
tools not helping you out enough, not catching errors
early enough, not being able to refactor in ways, not
understanding your code well enough.
And I think that the Dart editor is going to really
surprise you in what it can do.
Now, the third way, if you're from like a C# Java
background, you're probably used to really, really solid
development tools like Visual Studio and IntelliJ.
And you might look at JavaScript and say I just
can't do it.
I know that John said this before that--
JOHN MCCUTCHAN: There are no rules in JavaScript.
There's no rules.
WOLFF DOBSON: That's what I love about JavaScript.
There are no rules.
JJ BURNS: If you can think like a Lisp guy, or Python,
Perl, Ruby hacker, then JavaScript makes a lot of
sense, but if you're coming from a C# Java background, it
might be a little bit frustrating, and I think that
Dart alleviates a lot of that frustration.
WOLFF DOBSON: Speaking as somebody who has a PhD in AI,
I frickin' love Lisp.
Sorry, I'm distracting from you.
COLT MCANLIS: Quick question here.
So game development is all the performance.
COLT MCANLIS: And When it comes down to the metal, we
care about how you can iterate over the arrays, how we can
optimize for that.
So what does Dart bring to the table for games that care
about performance?
JJ BURNS: So actually, I think John could talk about this a
little bit.
I've been working with the Dart VM team to make this
fast, to make things like vector math faster under the
Dart VM than is currently possible
in any other language.
So by extending access to floating point arrays so that
you could do things like convert a four-by-four matrix
that's stored inside an array using the SIMD extensions that
your CPU has, these sorts of algorithms can be greatly
COLT MCANLIS: Fantastic.
So we can do fast linear algebra.
We can do transforms.
We can do processing.
WOLFF DOBSON: Tell me about debugging tools, like when I'm
trying to live debug Dart.
JOHN MCCUTCHAN: This is an awesome thing.
Dart has an IDEA.
It has tools.
It has a type system.
And you set breakpoints in your editor.
It's all the way there.
That's awesome.
You set breakpoints in the editor.
You launch your application.
It's running in Dartium.
The breakpoint hits.
It goes back to the editor.
There you go.
You can fan out your variables, look
at your call stack.
JJ BURNS: Look at your stack, yeah.
Look at your variables, mouse over a variable in the source
code, and it will show you the value.
COLT MCANLIS: And is the plan to migrate that from Dartium
to Chromium?
JOHN MCCUTCHAN: I think that in the future that's a
TODD KERPELMAN: So I'll end with one final question.
So obviously, in JavaScript right now, there's a lot of
libraries, Box2D for physics, stuff to kind of help manage
graphics, all of that.
Like Dart, my impression is it's just
starting as a language.
Are those libraries available, and can I use them just as
How does that work?
JJ BURNS: I would break the libraries into two categories.
There's things like jQuery.
And I always like to say that Dart is its own jQuery,
because Dart provides a nicer API for the DOM.
As far as the Box2D, there's a port of--
JOHN MCCUTCHAN: Yeah, actually there's a port being developed
by Dominic Hamon, who works at Google.
And I've actually been working with him on optimizing my
vector math library for the Box2D port.
TODD KERPELMAN: What about graphics?
JOHN MCCUTCHAN: I have a graphics engine, which
presents something closer to console graphics APIs and
DirectX 11 graphics APIs, rather than WebGL.
Then a demo, like higher-level library that sits on top of
that allows you to have a dynamically hot reloadable
scene graph and all kinds of animations
and things like that.
COLT MCANLIS: And this is the public, right?
JOHN MCCUTCHAN: Yeah, it's all on my GitHub.
JJ BURNS: Yeah that's the really nice thing about Dart
is that it is fully open source and
developed in the open.
And so you could look at John and I yelling at each other
over code reviews and so forth.
JOHN MCCUTCHAN: Over breakfast every day.
WOLFF DOBSON: I actually can frequently hear that.
JJ BURNS: Yeah, well, you know.
JOHN MCCUTCHAN: All the way from 19--
WOLFF DOBSON: Yeah, exactly.
I'm just sitting in 1900, and I can hear you guys shouting
at each other.
JJ BURNS: It's true.
TODD KERPELMAN: So thank you very much for dropping--
what you going to do now?
Are you just going to hang out?
JJ BURNS: I've got to go back to work, man.
I can't pop up during every GDL, man.
I've got work to do.
TODD KERPELMAN: We've got to remember not to
say his name again.
And he's out.
Don't say his name three times.
COLT MCANLIS: Yeah, he'll have to come back.
I'm going to let you finish about your Dart talk, but that
VM was the best.
I did not know he was back there.
WOLFF DOBSON: I'm frightened now.
What's the next subject?
Do we have more people back there?
COLT MCANLIS: Next time, we need to get Sergey back there.
Have him pop up with some Google Glass.
TODD KERPELMAN: That would be fantastic.
I'll talk to him about that.
COLT MCANLIS: Jimmy, can you send an email on that?
JIMMY: Saying his name three times does
not make him appear.
TODD KERPELMAN: All right, next topic.
We didn't really talk-- we sort of
missed this last time--
Big picture mode on scene.
WOLFF DOBSON: The big picture.
And presumably at some point, some type of Valve-specific
hardware coming out.
What do we think about all this?
JOHN MCCUTCHAN: I think it's pretty exciting.
Valve has been investing heavily in OpenGL and, in
particular, Linux performance.
So if I were to take a guess, I would say that whatever
Valve is producing is going to be running Linux because of
how much they've been publicly putting into R&D on the Linux
OpenGL stack.
It's really exciting.
I, at some point in time, kind of stopped gaming on my PC
because I like the idea of sitting around on the couch.
I missed that experience of hanging out with friends in
the living room, playing the games, and that seems to be
what the Steam Box Big Picture mode is all about.
WOLFF DOBSON: One of the reasons why--
I mean, I'm a huge Steam fan, and one of the things that's
great about it is it just works.
Like when you download a game and you play it, it just works
as opposed to--
COLT MCANLIS: Have you tried the Big Picture mode?
WOLFF DOBSON: No, I haven't tried Big Picture mode.
But I'm saying just in general, they have this
experience that to me, as a console developer, it felt
like console games.
You just go, you click on the game you want to play, you hit
play, and you're playing the game, often.
But when you're installing the game off disks, I remember
there was this point where I-- it took me--
I think I started Doom 3 seven times, trying to get it-- like
the settings in some sane way in which people's faces
weren't peeling off.
Just like I'm never--
TODD KERPELMAN: That might have been part of the game.
JOHN MCCUTCHAN: That was intended.
WOLFF DOBSON: No, but it was during the opening cut scenes,
which they were clearly not.
You were always seeing the back of people and the back of
people's faces and stuff.
And I was like, I'm never--
was it that?
Oh no, sorry, it just getting- incredibly tarry, and I was
like, OK, I'm never playing on a PC again.
JOHN MCCUTCHAN: I would love to have all of the games that
are on Steam in my living room on my TV, holding some
kind of game pad.
I would jump at that.
WOLFF DOBSON: If they can agree on the game pad layout,
then it's a wonderful thing, because you can then code to
that, and it's a--
JOHN MCCUTCHAN: And you actually get to target PC
games and this new meta console.
COLT MCANLIS: Is that sort of like-- so you
mentioned that, right?
If everyone can agree on a controller layout, is that the
logical flow?
WOLFF DOBSON: Oh, well, that's pretty important.
We have all agreed that there's a keyboard
and there's a mouse.
And that the keys are in this order unless you're like a
Dvorak person.
Even those people when they're sitting down
to play Team Fortress--
WOLFF DOBSON: Yeah, exactly.
It's like--
TODD KERPELMAN: You've got to play like this.
COLT MCANLIS: No, it's Dvorak.
Just type Dvorak.
Every game, you just type Dvorak over and over and over.
WOLFF DOBSON: But in all seriousness, like controls are
what makes games.
And you need to make sure that the thing you're
touching is the same.
So if one guy is sitting there like, yeah, I just press this
button, and it puts on the handbrake, and I can spin
around turns.
The other guy's like, I have to the left trigger and press
the left joystick at the same time, and I can't--
JOHN MCCUTCHAN: Actually, that's a great point.
I mean, if they were to present some kind of like
Steam Box home console, the quality of
the game pad is key.
I mean, if it's just take any game pad off of the Best Buy
shelves, I don't think it'll be that good of an experience.
But if there is a, this is the Steam Box game pad that--
COLT MCANLIS: We need to go back to power glove.
Make it happen, Nintendo.
JOHN MCCUTCHAN: Bring it back.
WOLFF DOBSON: We gave you the Wii.
What do you want?
COLT MCANLIS: It's not in glove form.
So here's my question for everybody.
What's driving this?
In reality, I think we all agree that Steam is a
fantastic product.
It's been probably, in my personal opinion, keeping
gaming on desktops alive.
TODD KERPELMAN: I think they've done a huge service to
the PC gaming industry.
COLT MCANLIS: Absolutely.
When we saw consoles taking over, we
saw a big trend there.
Even the switch to mobile and web now, I think it's safe to
say that Steam is still standing alone.
Why the move to console?
What's driving that?
JOHN MCCUTCHAN: Gabe has come out and said that he's very
unhappy with Windows 8.
He doesn't think that's going to deliver as good of
experience for gamers.
TODD KERPELMAN: I think he used the word "catastrophe."
JOHN MCCUTCHAN: Yeah, it was very strong.
TODD KERPELMAN: It was very strongly worded.
I mean, clearly some of this is from--
it appears as though Microsoft is going to put its own
version of Steam into the operating system.
I think it's going to make it--
well, I don't exactly know what they're doing, but I
think they want to make their own app store that is probably
going to be a lot more visible than whatever you would have
to do to get to Steam.
I haven't used Windows 8.
I don't know.
This is all just speculation.
COLT MCANLIS: Also, there's kittens everywhere.
TODD KERPELMAN: But clearly, they're
saying, gee, we're awfully--
WOLFF DOBSON: That's YouTube.
TODD KERPELMAN: --we're awfully tied to--
they are kind of at the mercy of one, possibly two companies
that are not them, running these platforms.
WOLFF DOBSON: I just think the other thing is that it's the
continuous upgrade.
I find this with car electronics.
My wife has a car that's got a built-in Maps thing, or a
built in GPS in the radio or whatever.
And my incredibly old car, I have an aftermarket thing that
I just put my Android phone in.
And every time I buy a better Android
phone, my car gets better.
I think the same thing goes with the Steam Box experience
or this Big Picture experience, which is it's been
how long since the Xbox 360 came out?
Waiting for these hardware cycles for the console
industry, especially where we're at this point where 3D
games with 3,000 polygons per character look about the same
as games with 7,000 polygons per character.
At that point, eh, what's the hurry to upgrade?
But if you're a hardcore gamer, and I think Steam
definitely leans into that hardcore market, you're going
to keep upgrading your PC as you go along, which means your
console going to get-- your console, your Big Picture, is
going to better and better as you go along.
TODD KERPELMAN: Well, that kind of goes against the whole
dedicated Steam Box idea.
JOHN MCCUTCHAN: Yeah, I would really--
WOLFF DOBSON: But the trick with that is that they can
keep releasing Steam Box 2, Steam Box 3, Steam Box 4, and
since they're all PC games at their heart, they're just
going to keep getting better.
JOHN MCCUTCHAN: Sure, but I mean I think one of the
biggest reasons why I moved away from PC gaming was
because of that upgrade treadmill, which was every
year a new game came that required a new GPU and a new
faster CPU.
TODD KERPELMAN: It made you feel awesome when you went
down to the store and you're like, put down your $600.
JOHN MCCUTCHAN: It was lot of fun, but console games also
continue to look better and get better as the life cycle
goes on because they figure out a way to extract more
performance out of--

TODD KERPELMAN: It is nice knowing that a console game is
coming out and I just know it will work on my Xbox.
TODD KERPELMAN: I have to look at the specs.
COLT MCANLIS: We've got to--
TODD KERPELMAN: We've got to move on.
We've got to move on to--
this is a question from Dr. Dobson.
Who's your favorite sidekick in a game?
WOLFF DOBSON: That's a real scientific
question right there.
TODD KERPELMAN: Well, it's a good one because we've
talking about AI.
We talked about the Fable horse.
WOLFF DOBSON: I will give you two favorite sidekicks.
The first one is Alex from Half-Life, Episode 2.
Because there, they really spent the time lingering on
her emotional state as part of yours, like how she feels
about this world.
I went through and played the beginning of it again just the
other day, and I was like, wow, this is really neat.
I know what she's feeling because I'm looking at her
face and sort of seeing what's going on.
JOHN MCCUTCHAN: And Freeman is blank, right?
He's everyman.
JOHN MCCUTCHAN: So she's reflecting emotions that you
should probably be feeling back onto you.
WOLFF DOBSON: I wanted to go on an
adventure with her forever.
I felt the same--
it was just like, this is great.
I mean, except that world is sort of a horrible one to live
in, but that's a different problem.
COLT MCANLIS: Besides all the death and the
robots and the aliens.
JOHN MCCUTCHAN: But hey, at least she's here.
But at least I have somebody--
COLT MCANLIS: For some reason, that tells me that should be a
Hallmark card.
Besides the aliens and the robots, thanks for being here.
WOLFF DOBSON: You are the girl for me.
JOHN MCCUTCHAN: Fascist government system.
WOLFF DOBSON: I also think that the fact that she doesn't
go and get herself killed all the time is super important.
You don't get frustrated with her.
WOLFF DOBSON: She takes logical cover.
She falls back which she's supposed to.
And she isn't clearly invincible, but in practice,
it turns out she is.
The other thing--
the other companion is the D that you get in NetHack that
follows you around at the beginning, who, if you play
like I do, when you finally run out of food on level four,
you kill him and eat him.
He's like, not only a best friend, but also a meal.
And I just want to tell you all that--
COLT MCANLIS: Why do I think that Wolff is going to quit
Google one day and make a Half-Life 2 meets NetHack
clone, where in level four, he's really conflicted.

COLT MCANLIS: For like a week, he's just like, what did I do?
WOLFF DOBSON: It's alive.
JOHN MCCUTCHAN: It's The Road, in game form.
Wow, I'm really running out of food here.
Wow, that is like a whole different
version of Oregon Trail.
JOHN MCCUTCHAN: Well, one of us made it.
COLT MCANLIS: And we all sort of made it, because you're a
part of me now.
TODD KERPELMAN: The Donner Party edition of Oregon Trail.
All right.
I've got to think about this.
COLT MCANLIS: So I'm going to go back to a Valve favorite
and actually say GLaDOS.
WOLFF DOBSON: Oh, I thought you were going to go with the
companion cube.
COLT MCANLIS: I was not going to go with the companion cube.
I think that's a little overdone.
A lot of people underestimate the drive that GLaDOS brings
to the table.
She motivates you.
TODD KERPELMAN: Are you talking about number one or
two, when she's in the potato?
TODD KERPELMAN: Yeah, I'm sorry.
COLT MCANLIS: What's up with that?
JOHN MCCUTCHAN: You haven't played a
two-year old game yet.
No, in Portal 1.
I think what they did to bring--
WOLFF DOBSON: She is a wonderful companion there.
COLT MCANLIS: She was an amazing
companion and sidekick.
She got you where you needed to go.
She motivated you even if it meant giving you the pointy
end of the stick.
Even though she tried to kill you, which I think that all
good sidekicks should try to kill their--
WOLFF DOBSON: If you make the dog mad in NetHack, he will
come after you.
COLT MCANLIS: Well, that's because he saw what you did to
the D.
JOHN MCCUTCHAN: Kind of aloe the same lines as Alex, I've
really enjoyed the sidekicks that you get
in Uncharted games.
Oh, Elena is another example of a--
there was a wonderful moment in Uncharted 3 where--
or 2, I guess it is, where you're walking through Nepal
and your two ex-girlfriends are fighting with--
are arguing and talking about you, and you're just--
the character is doing--
JOHN MCCUTCHAN: It feels so natural.
COLT MCANLIS: When I'm walking through the streets of Nepal,
it brings back so many memories.
WOLFF DOBSON: But in all seriousness, the fact that
they were having this wonderful conversation that
was completely grounded in the story that you're having, and
it was bringing their characters
forward, it was amazing.
TODD KERPELMAN: I'm going to go with--
I've got two picks.
One would probably be the dog or cat in Torchlight.
Because they're useful.
WOLFF DOBSON: You know that dog is getting like
a 50% cut of every--
TODD KERPELMAN: I don't care.
He's going back to town for me, doing all the work that I
don't want to do.
WOLFF DOBSON: This is what younger siblings were for when
you were growing up.
COLT MCANLIS: Back on the farm.
TODD KERPELMAN: I'm sure he's embezzling money from me, but
I don't care.
And number two--
WOLFF DOBSON: You have to audit the dog.
It's a really boring video game.
COLT MCANLIS: He's got a great lawyer.
Weird thing.
Number two, I'll go with--
what is it, HK-47?
WOLFF DOBSON: Ah, kill all the meatbags.
Ah, I love that.
This was the best robot ever.
He was the anti C-3PO.
For those of you who haven't played the game, in Knights of
the Old Republic, you got a Droid sidekick.
And his thesis was kill all the meatbags.
He barely tolerated you, and any human, he was just like--
he was the polar opposite of every robot that had ever been
in Star Wars.
It was wonderful.
An assassin robot, and he was hilarious.
JOHN MCCUTCHAN: I like that term "meatbag." I've never
really thought of us as bags of meat.
TODD KERPELMAN: You have this whole conversation with him
about why do you call me meatbag?
And he's like, well, it's kind of what you are.
You're bags, and you're full of meat.
COLT MCANLIS: It's sort of Fifth Element style, like, are
you humanoid?
Identify yourself.
Negative, I'm a meat popsicle.
TODD KERPELMAN: So let's move on to--
we've got time.
We're kind of running low.
Maybe we just go into what we're playing--
WOLFF DOBSON: I think we should do what we're playing.
TODD KERPELMAN: --and save the fifth topic for next time.
So all right, we'll go--
we'll start with Colt.
COLT MCANLIS: No, he deferred last time.
TODD KERPELMAN: We'll start with John.
You're right John, what are you playing?
JOHN MCCUTCHAN: All right, so I am playing--
WOLFF DOBSON: We've heard your trials with 100 Floors.
JOHN MCCUTCHAN: Yes, I've been playing 100
Floors a lot lately.
Got ya!
JOHN MCCUTCHAN: It is a frustrating experience, but it
just keeps pulling me back.
I keep going on my phone.
I'm like all right, I'm going to make it through this floor.
WOLFF DOBSON: It's such a slot machine of I got it, I got it!
JOHN MCCUTCHAN: Some of the stages, I just--
as soon as I see the layout of the stage, I'm like, oh, doo,
doo, doo, and I'm onto the next floor.
And other ones, I'm stuck for a day.
And actually, in the end when I succeed, it makes sense.
I'm kind of like, yeah, I see that.
And there's just something about--
the game has these rough edges that drives you away from the
obvious answer, but the answer is always obvious.
WOLFF DOBSON: It's like a programmer question.
Like what are you not assuming?
TODD KERPELMAN: All right, Colt.
What are you playing?
COLT MCANLIS: I'm actually playing a handful of games
that are uninteresting.
I'm actually-- if you've been following my Twitter feed or
my G+ feed, then you're probably aware that I'm
working on trying to analyze streaming algorithms and how
to get large game worlds streamed off of disk and
through the internet and whatnot.
And so I'm actually going back to the table.
I'm actually trying to play a lot of different games that
have done streaming very well and very bad and sort of take
a technical look at them.
So I've been looking at the elevator scenes
in Mass Effect 2.
Effect I've been taking a look at the streaming system inside
of the source engine for Valve.
I think they do a great job of it there.
A lot of open world engines with low load times, like GTA
and Jak and Daxter, and those sorts of games.
WOLFF DOBSON: Run backwards in Jak and Daxter.
COLT MCANLIS: Yeah, exactly, right?
And so just taking a look at a handful of these things and
see how people are solving the problem, where is the
technical hurdles?
How did they solve them?
See if I can't grok any information out of it.
And if you're watching the show and you've got a good
experience with streaming large worlds, please send me
an email, send me a Tweet.
I'd love to talk with you more about it.
TODD KERPELMAN: As for me, let's see,
I've got three games.
On I'm still playing Skyrim, and that will probably be a
standard answer for quite a while now.
COLT MCANLIS: Took an arrow to the knee.
TODD KERPELMAN: I am playing The Walking Dead.
WOLFF DOBSON: Is it working for you?
Are you saying you had technical problems?
WOLFF DOBSON: No, no, working for you like I wasn't sure how
that would translate into a point-and-click adventure.
TODD KERPELMAN: So it works really well as sort of a piece
of storytelling.
The mood is really good.
The setting is really good.
The characters are interesting.
The dialogue is well written.

You sort of play it, and it really kind of gets you into
the feeling of like, I am in a world where
zombies have taken over.
Sometimes I'll finish playing, I'm ready to go to bed, and
I'm like, I'm going to lock the door because I don't want
the zombie to come in.
And then I realize, no wait, that was just the game.
The actual game play itself is not particularly interesting.
It's a fair number of conversation trees that I've
realized now, looking at a couple of game FAQs, often
don't seem to have an effect.
You're entertaining yourself by answering these questions.
TODD KERPELMAN: And then the standard adventure game puzzle
type of things that--
WOLFF DOBSON: Use skunk with white paint.
Use white paint on fence.
Use fence to get splinter.
TODD KERPELMAN: But it is slight--
I will say their puzzles are more logical than your typical
adventure games.
COLT MCANLIS: Better puzzles than 100 Floors or worse?
JOHN MCCUTCHAN: 100 Floors is very--
TODD KERPELMAN: That's like hey, we're--
that's like Myst.
And this is like Monkey Island style.
Yeah, but anyway, I would say just as an interesting piece
of storytelling, it's totally worth it.
So, check it out if you haven't.
And then on mobile, I am playing--
I picked this up in the $0.25 sale yesterday--
Let's Create Pottery, which is a sort of pottery-making game
that after playing it, I'm like, this is so obvious.
Why didn't anyone put a pottery wheel on a mobile
touch device before?
You're just like, I'm going to push it in here and pull it
out here, and you get this vase.
JOHN MCCUTCHAN: There's actually a really
awesome tech demo for--
that comes with Move controllers for the
PlayStation 3 SDK where it's essentially that.
You hold one Move controller and you use the trigger and it
spins this long tube of something.
And then you use the other one, and you kind
of lathe in, like--

COLT MCANLIS: --for the Move controller which required to
like, dance around the base and sculpt it with your hips.
JOHN MCCUTCHAN: Some of the tech demos available to
developers with the Move controller are the most
amazing usages of one-to-one motion tracking
that you've ever seen.
Being able to spin this thing and pull and have it come out
to and make all these weird swords and clubs and things
like that, it's so creative and fun.
WOLFF DOBSON: So for my--
I'm also playing a zombie game, but I'm playing Lone
Survivor, which is a 2D game that was clearly--
it's very pixely, and you shouldn't be frightened of it,
and I can only play it for about half hour at a time.
And then I'm like, I don't want to go down that hallway.
I think I'm going to go work on something else.
So yeah, I've been playing like 31 minutes
a day of this game.
COLT MCANLIS: What platform is that on?
WOLFF DOBSON: It's on Steam.
I'm playing on a Mac, but it's on PC as well.
It's tight.
The sound design is really good.
You hear the zombies far away.
And you're like, oh god, zombies are coming.
And they don't groan.
They make this weird staticky noise.
TODD KERPELMAN: It's more Silent Hill, and it's--
WOLFF DOBSON: Yeah, I guess it is more Silent Hill.
I was getting kind of a Resident Evil vibe off it,
because they give you the gun, and it's totally ineffective.
COLT MCANLIS: It's been too long since we've had a good
suspense style game.
The first Bioshock was the last really good one that I
think came out.
We've been missing that for a couple years.
WOLFF DOBSON: Yeah, it's good.
I'm like I said, I haven't gotten all the way through it
because it's too scary.
And on mobile, I'm playing SteamBirds, which just--
I've been playing in my Nexus 7, and it's exactly the right
size for a Nexus 7.
And it's got this beautiful touch interface.
And it's just--
I'm sitting there, I'm at Go at our cafe, with my sandwich
in one hand and playing SteamBirds in the other.
It's totally fun.
This is what touch gaming is about.
TODD KERPELMAN: All right, I think we are out of time.
We've got no one else behind there.
WOLFF DOBSON: No, thank goodness.
TODD KERPELMAN: So remember, don't say
JJ's name three times.
No one.
And we'd like to thank you for joining us for
Google Games Chat.
We will see you in two weeks.
I'd like to thank my panelists--
TODD KERPELMAN: --posse for joining us.
COLT MCANLIS: Entourage.
TODD KERPELMAN: And we will talk to you--
WOLFF DOBSON: I walk around in front of Todd opening doors.
It's pretty much how we roll.
good-looking one in the entourage.
JOHN MCCUTCHAN: I don't watch the show, so--
TODD KERPELMAN: I'm probably E. I'm the shortest.
Anyway, we'll talk we'll talk to you guys later.
Billy, you can fade us out.
TODD KERPELMAN: Jimmy, sorry.