Friday, November 11, 2005

Mini Game Design Reviews: Guitar Hero and Wierd Worlds

I wanted to briefly mention two very different games that I’ve been spending time with recently. The first is Guitar Hero by Harmonix and the second is Weird Worlds: Return to Infinite Space by Digital Eel. Neither is a traditional game, and both have some good lessons about game design to share.

Guitar Hero: The Importance of Setting
Guitar Hero is a rhythm game at its core. The abstracted game mechanics are little different than DDR or the dozens of other titles that have come before. Yet this is one helluva amazing title. I just watched my friend rip through a couple of sets and I’m completely pumped. Adrenaline, air guitar…my lady keeps looking at me with a bemused look when I break out into “More than a feeeeling. Baawah, bam, bam”

Deep down in the core of your being, you know Boston rules. And what about Black Sabbath playing Iron Man? I grew up listening to such music and I suspect a lot of folks did as well. These songs resonate.

And this is the genius of the title from a game design perspective. The risk / rewards schedule are driven by more than just abstract tokens. When you hit a wrong note, you are letting down your audience. When you hit a note and the crowd cheers, you are experiencing a microscopic moment of rock stardom. This is an experience you’ve been wanting for years. The psychological reward of hitting a correct key is multiplied by every time that you’ve air guitared in your bedroom (with the door shut) or sung the words to your favorite rock song.

The controller completes the psychological bridge. It is a guitar in the mind of the player. The immersiveness of the basic verbs – playing cords, wailing on the whammy bar – is accentuated by the player’s ability to carry a familiar object, and holding a familiar stance. This is a game controller loaded with cultural meaning. We’ve talked before about the ‘juiciness’ of low level risk / reward schedules. The controller turns a mechanical experience, “pressing a button” into a tactile experience with strong emotional connotations, “playing my favorite power chord.”

The star power activation method is ingenious. You tilt the guitar straight up, rock star style. Give that designer a raise.

When you tie into real world desires, you strengthen the power of your risk / reward schedules dramatically. Guitar Hero is an addictive game, far more than its Simon-style mechanics suggest. The title fills a real psychological need and that gives the design impressive power over its audience

Potential to tap into the non-gamer market
I love the audience profile on this one too:

  • If you’ve ever taken guitar lessons

  • If you’ve ever air guitared to a rock song

  • Doesn’t matter one lick if you are a gamer or have never touched a game console in your life.

This market has approximately zero competition in the US and is huge. A title such as this is the equivalent of inventing karaoke (Yes, I’m aware of Guitar Freaks…doesn’t count since the ever-so-critical setting isn’t American enough). Given the proper marketing, this is the sort of the title that could spark a crazy cultural fad.

Yet, I had to call around just to find a box. Dumping a title with such broad appeal in game stores is not the path to rock star level success. My thought? Harmonix should get Nintendo on the phone right away, sign up for the Revolution and get on their marketing schedule ASAP. Get a publisher behind this who knows how to sell to non-gamers.

Weird Worlds: The Benefits of Randomly Generated Worlds
Weird Worlds is the sequel to one of my favorite indie games of all time, the wonderful Strange Adventures in Infinite Space by Digital Eel. I have to wait until I’m in my new home before I can order one of the antiquated CDs from their new publisher, but I managed to snag a Weird Worlds demo.

The core game design is roughly the same with a coat of new graphics and new user interface and lots of new content. The interface could use a bit of polishing, but the game play here is the key.

First, this is a very small team. I count three main people with the sequel seeming to have a few extra folks helping out part time. They’ve managed to create a highly addictive single player experience that evokes a strong adventure feel without spending massive amount of money on content, cut scenes or elaborately scripted set pieces.

ROI of Randomly Generated Maps
Instead, the player travels about a randomly generated map. I’m sure balancing was a complete pain, but the end result is a nearly infinite number of short adventures. Others have tried this style of game play, but SAIS and Weird World have managed to produce something that is addictive and has surprisingly long term appeal. Even after playing SAIS for months, I’ve yet to burn out on the title.

There’s an ROI that comes from randomly generated worlds that is very impressive. Content costs for static levels increase in a linear fashion. Every level you add costs roughly the same as rest. Randomly generated levels have a higher up front cost. You need to create the map generation algorithm, develop the various classes of objects and create balancing metrics for those object classes. A simple static level might take a day or two to throw together, where a competent random map generator might take two or three months.

However, after the initial investment, the random map generator is insanely efficient. New objects can be added in an iterative fashion. Major balance changes to the entire game can be made nearly instantly. What you lose in your initial investment, you gain back a hundred fold in flexibility and the ability to provide your players with lots of content.

Also, static levels have their own hidden costs. As you add new tokens to your game, the cost of building a level increases. Later levels will often take substantially longer to polish than initial test levels. Randomly generated maps don’t have this issue. You only have to pay the cost of creating the token. There is little additional level creation cost since it is an automated process.

Benefits of automated builds
The phrase, “automated build process” excites me more than you can possibly imagine. Games initially were about software development. Over time, they have turned into a production heavy activity with the vast majority of effort being spent on waterfall-style Disney-esque content creation. Production heavy endeavors tend to have a lot of momentum and be difficult to change. Project managers have a tendency to minimize changes to high risk items such as core game mechanics in order to preserve the heavy investment in existing content.

When you start making your game development more like software development and less like production work, you can more easily take advantage of agile development processes. Core game mechanics can shift more fluidly if you aren’t weighed down by the thought of breaking a hundred handcrafted levels. There is nothing worse than changing the jump distance on your character and ruining multiple man years of labor because your character can no longer successfully navigate the lovingly handcrafted map files. With content being ‘built on the fly’, such a change is trivial.

Fluid core game mechanics means more low risk opportunities to polish those game mechanics. You can iterate on original mechanics more freely and ideally create games that are more psychological addictive. If the game play is what matters, build your game using a process that gives you the freedom to make changes to your game play at a low cost.

You’ll see the term algorithmically generated content tossed around. The money men focus on the cost savings over static content. That is a good benefit. The ultimate benefit however is the unique ability to inexpensively polish the game play of your new designs. Think of it as a critical building block in the practile of agile game design.

Turn-based: Fits into your life
The other aspect of Weird World that is exciting is this is a turn-based title. A busy player can stop it at any point with no worries about dying or losing their path. Life for many is a series of constant interruptions and games that fit into a multitasking environment occupy a useful niche.
I’d love to see this concept expand out into a setting with a bit more appeal. Space is great, but it is somewhat abstract and can alienate the more casual, female gamers (though there are more women sci-fi fans than men these days). Would a treasure hunting, exploration game be more popular with the casual audience? What about an urban shopping title? It would be interesting to take this great game design in additional directions.

These are two very different games with two powerful lessons. If you haven’t checked them out, you need to.

I can’t help thinking about combining the two. What if you had algorithmically generated game with a setting the resonated with a strong customer need? I’m not sure what that might be, but it certainly is a fun design exercise to contemplate. :-)

Take care



  1. First, let me say that I've been playing Guitar Freaks off and on for several years. Most of the time it seemed like a pretty good thing.

    But today, when I played Guitar Hero for the first time, I realized how far Harmonix has gone beyond the Konami experience.

    1. The songs in Guitar Hero are full-length. To maximize arcade revenue, Guitar Freaks uses short mixes of each song.

    2. Note matching is slightly relaxed. This makes strumming up-and-down in one motion much less frustrating than in GF.

    3. Holding notes and five fret buttons are a *huge* deal. The authenticity of the Guitar Hero experience goes up several notches when graduating to Normal and Hard modes.

    As you mentioned, the American-ness of Guitar Hero is crucial. I've tried showing Guitar Freaks to a number of people, but no one wants to play J-Pop and weird in-house Konami songs. On the other hand, give someone Black Sabbatth and Bad Religion and heck, Franz Ferdinand and they will want to ROCK.

    If RedOctane can sort out the distribution problems this title should go far. They need to get it on some morning shows, send review copies to the non-gaming press, *anything* to get this in the hands of new players.

  2. I loved Strange Adventures, and am looking forward to trying Weird Worlds. I tried to get them to give me a copy at the IGF this year, but apparently it wasn't finished. :)

    Digital Eel are simply wonderful - I just hope that they are earning enough to feed themselves.

  3. Glad you guys are enjoying it; everyone at Harmonix is very pleased with the way things turned out with the game. Hopefully enough people will be caught by it's lure to pull us over the critical mass boundry. I know within my circle of non-gamer friends it's spreading at an amazing pace.

    We currently have demo stations at Best Buy, which are attracting quite a bit of attention. It's one thing to see the game in action, it's another to experience it. I'm convinced that if we can get people experiencing the game in any way, we'll hook them all.

  4. About random (and pseudo-random) generated content, I really don't understand why it's not used more... Remember how Elite and specially Elite2:Frontier managed to get a very complex and detailed galaxy with real planetary systems and motion all into a single 640kb disk? It still is amazing by today's standards (And back then they even did the graphics with real 3D curves... )

  5. It'd be very easy to create an algorithmically generated version of Guitar Hero. It's just that you'd be restricted to the Bad Religion catalog. :)

  6. Oh, thats a zinger.

  7. Good point that procedural content allows you to tune gameplay much later in the project.

  8. I like procedural content a lot, but there are limits to how far it can go. Imagine an adventure game with procedural content (ick). Some genres, like the adventure game or the wonderful old shmup, depend on the level being tuned and crafted with absolute precision.

  9. Jason,

    Just another indie dev here reading DanC's meanderings. :)

    I have to say I frickin LOVE Guitar Hero, and I'd buy vol 2 and 3 in a heartbeat.

    I'm actually thinking seriously about purchasing another slimline ps2 and GH for my bro who's a musician (and mostly a non-gamer.) He'd totally dig it.

    Rock on.

  10. Hey Jason,

    Any thoughts on how we could buy a copy of GH here in the UK?

    Ive got this big buzz right now of looking into different controllers for an exercise in thinking outside my usual design process.

    So I blagged a couple of the "GameTrak" hand-motion-tracking devices from and I'll be experimenting with those a fair bit.

    I'd love to play the guitar hero controller though. Weirdly enough I'm also an ex-semi-pro muso too, so I've done some gigging in my time and I must admit, the rush of playing to a packed tight audience is something Ive missed for far too long :)

    Do you have a UK distributor?

    So, on to Dan's post..

    Dan, Rich Carlsson who does the programming on the SAIS and new games has a definitely unique way of looking at games. He's an industry guy who basically decided to do something else. I dunno if business is brisk, but they definitely seem to be making a go of it as I've known him for a few years now and he's still at it :)

    I've not played the games myself, as theyre not really my bag baby :) but you have to admire the intent of them.

    I guess I should pick them up, just so I have a complete understanding of the design of them, which I really like (they target ultra-short gameplay sessions, with high repetition). Obviously thats pretty unique in games these days.

    As for your assertion about generated content, I can say that one of the biggest features of the worms franchise, other than the customisation of the worms themselves has been the generation on the fly of the levels to play on. We obviously spent a fair amount of the development time on this feature, but honestly it brought so many benefits its amazing. We had in the end, a bunch of hand-build content and a generation feature that automated generation of random landscapes too. This meant that we could ship with some pre-built content with mission based concepts, but we also had unlimited combinations (I cant remember if it was in the order of millions or billions of combinations).

    The best thing from all of this of course, was that fact that we boiled the whole random generation thing down to a simple 16 character string. Essentially you could type in a string into an edit box and the same level would be created EVERY TIME.

    This was an amazingly powerful tool, becuase someone could generate a great landscape and then share it with thier friends in plain text format. Essentially an algorithm turned the text into a bitfield which then turned on various generation features etc.

    I've got a game design in 2D I want to make that has very similar concepts because its SO very powerful at creating content for very little effort and with such an intuitive user interface.

    Maybe someday when I get time :)


  11. Procedual content has many advantages to game designs which rely on systematic exploration. As an example, take Diablo. When you consider the procedural content in Diablo, you probrably think of the procedurally generated dungeons. But what I think of is the treasure system.

    The randomly generated dungeon layouts did little for the replayability of Diablo; and I'd contest that you'll get similiar results for any system which produces content of this style. Instead, the real exploration of Diablo is through the treasure system; not the world space.

    The same is true of the space games I've played that procedurally generate the universe. The content there has more in common with Diablo's treasure system than it does with it's world layout system. If your game system provides a large content space within it's system, you can map that over area and get a useful result. In other words, planets are just treasure, and treasure is the munging of several axis's of statistical information to produce a unique result.

    In a more tradition content driven game, lets say a FPS, then procedural methods need to be thought of as a paintbrush in the hands of a painter. The human touch is simply what makes these games; and no procedural technique can replace that. Instead, in those types of systems, procedural techniques should be used to speed up the workflow by removing the tedium of certain tasks.

    In the Turbine engine (AC1, AC2, DDO, MEO), we mixed procedural object placement with hand painted maps to get great results. For instance, I could specify what a section of forest would look like and define a range of possibility for each area. Then I could use a brush to simply paint down forest over an area. If I changed the reference files to point to snow covered trees and terrain textures, then the world turned to winter.

    However, it was far to easy for people to forget the small details. They could paint huge areas of land and fill them with monsters without thinking much about the flow of the area. As such, we often had huge open areas of content which was exactly the same. When we started on Dungeon's and Dragon's Online, I banned the use of 'scene painter' for a while, requiring that the level designers work on much smaller areas and much tighter content. Once they had identified what would make the content for a small area good, we opened up use of the other tools again. I sum up this lesson as "What you make easy to do you will get a lot of, what you make hard to do you will get very little of". So now, when building tools, I aim to make the quality of content better and faster to produce, not the quantity.

    This type of semi-procedural method is far more applicable to many games than a purely procedural system. To make a parallel to music, lets consider sythisis of music.

    We've heard algorithmically generated music. It sounds algorithmic. Even though it can write perfect counterpoint, it lacks the creative edge that Bach would apply to break the rules of counterpoint in a creative way. But procedural methods exist all over music production these days.

    Take synthisis; we're using procedural methods to create tones and tambers, while a musician chooses the notes. Simple enough. But over the years, we've taken things much further. Sythisis has evolved to create entire grooves from simple parameters being mixed and animated on the fly. And while any amature can dial up a pretty cool groove with one of these devices, it still requires a human's touch to give the song it's identity through lyrics, melody, and arangement.

    So, my point is this: You can turn over certain types of content to a purely procedural method; but in doing so you will be limiting your design to certain styles of play; and your design must fundimentally be shaped around this.

    The other approach is to use procedural methods to move the content experience to a higher level of abstraction; but you'll need to make sure that level of abstraction is useful to the user creating the content, so they can put in the human touch. You need to give them a paintbrush.


    Another clear example: The hand animation system I wrote for Guitar Hero. It combines a small amount of authored data with a large amount of procedural data to produce an extremely convincing result. Essentially, we author in where my left hand is on the fret board (as midi notes), and what shape (chromatic, pentatonic) my fingers are moving in. The actual fingerings, vibratos, and strums are all derived from the gameplay data (gems). It takes moments to author, and looks much better than any purely algorithmic result. And it's much faster than animating everything by hand.

    Anyway, I think the largest mistake you can make with procedural systems is expect that you can apply them to any type of game space as pure algorithm. While some dynamics will allow this, the majority require some human touch to be great. In those cases, use procedural content as another paint brush in your toolset.

  12. Question for Danc:

    Weird Worlds does teach us a lot about the benefits of procedural content and short gameplay sessions, but we already learned these lessons from its predecessor, no? After playing the demo, I like the new coat of paint but I don't see a whole lot of new design lessons that weren't learned from SAIS years ago.

    Bonus question for zoombapup:

    If procedural generation of terrain was so key in the Worms franchise (I agree it was), why was that dropped in Worms Forts?

  13. In almost all cases, randomly generated content has to be combined with the human touch or else it will get really boring really fast. Anyone else stop playing Dark Cloud 2 because of the randomly generated dungeons? I think Michael Mateas said it best at the Future Play Conference last month (I’ll have to paraphrase), when he said that trying to delegate everything to random generation and AI algorithms is as absurd a goal as saying “If we just get the AI right, we’ll solve this whole Painting thing once and for all.” Good art is always going to be tied directly to the ever-changing human experience.

    Another thing that can help randomly generated content is designing it in such a way as to increase the likelihood of emergent behavior – when your content “gems” (I like that term, Jason, so I’m stealing it) are designed to have a lot of interactions with other gems – and the end result can be much more fun than the sum of its parts.

    Lastly, regarding Guitar Hero (fantastic game, by the way). I’ve always thought that the best way to find the next big game is to just look around at what fun, creative, fantasy-based things people are already doing and see if there’s a way that technology can be used to make the fantasy more real and responsive, or as “creativity amplification” (ok, so I stole that term from Will Wright). Kids were already playing “house” and “dolls” for a long time before The Sims. Kids were already playing “guns” and “cops & robbers” long before the first shooters came along. People have been playing air guitar for decades, and it just took some smart company to figure out how to make the fantasy more real. Now, here’s to hoping it becomes the next big game.

    PS – I would also love to buy my brother a PS2 Slim / Guitar Hero bundle this Christmas. I really hope they come out with one (say, $200? Eh? Eh?).

    - Jeb

  14. The holy grail of gaming is automatically generated scenarios that don't suck. Of course it's also a franchise killer...


  15. When you start thinking of games as software and game sequels as release 2.0, you stop worrying so much about franchise killing. Add in subscription or micropayment services to the model and there's even less worry.

    Good points about the 'artist' enhanced generated content.

  16. micropayments only work if you've got some means to collect them, usually this means the internet, though I would imagine cell phones could put some kind of outfit together.

    I agree this is probably an underexploited market, but game enthusiasts don't necessarily want or need the internet to play games, nor should they. I suppose this is where boardgames have a lot of advantage.

    As a boring dad, I almost dread the thought of spending money for more internet than I already have.


  17. Are you ever going to do a game design review of World of Warcraft, or explain its incredible success?

  18. Ian:

    Worms forts was a political decision rather than an artistic one. The development of that game went outside of the bounds normal for a worms game, in that it was another team entirely doing it (none of us guys who had worked on multiple worms games).

    They had art leads and such who decided they wanted full control of the environments in maya etc.

    That kind of thing is ultimately why I left Team17, because the franchise has been turned into a parody of itself by poor choices in terms of development of features and games.

    Not being in a position to affect the choices, I felt like there was nothing left to give.

    A ray of hope there, was that there was a PSP project started a year or two after I left that if completed would actually be the Worms 3D that I personally was pushing for all along. But I dont know if they got behind the project properly.

    Ah well, its more fun being indie sometimes :)

  19. Red Octane sells a multiplayer guitar hero bundle online (2 guitars and the game) allready, but no slim-line ps2 included version.

    I have a viola case I've modified to hold 2 guitars and a slimline ps2.

  20. This comment has been removed by a blog administrator.

  21. I actually just interviewed Greg LoPiccolo (one of the makers of Guitar Hero) for my blog. He had some interesting things to say about the difficulties associated with designing a game with a custom peripheral, and about innovation in games in general.

  22. Jason: Is there a red octane distrubutor in the UK? I really wouldnt mind this game for Xmas (and of course I wouldnt mind getting the guitar controllers for our game design degree course).

  23. Surely one of the strongest examples of algorithmically generated content is the Civ series? 'Levels' such as they are (referring to each randomly generated world) don't seem to contain content gems at all, and yet the random generation of resources, terrain and starting points sets the scene for some excellent set pieces. Are the content gems in this instance the tech tree, the AI, etc.?

  24. [obligatory OFF TOPIC post by Ray]

    Hey Danc,

    You should review Harry Potter... give us something to talk about... ;) (Thanksgiving was awesome and it was great to meet Aya in person!! She's definitely one of those rare personalities that you have to search the world to find. You're one lucky man.) Hopefully you made it home safely.

    Hey Phil, Leinad, Lennart, I wish you could've been here. Seattle's the place to be, we have a guest room if you ever wish to visit!

    Huggers and All my best!


  25. Ray!

    I'm always in that neck of the woods, at least once a year around October time I come over for a conference in Eugene, Oregon. Just down the road!

    I'll make sure to get some extra time next time I'm out there and come visit!

    My friend is likely moving to vancouver too, so it might be a nice lil round-trip to do :)

  26. hey that would be so sweet. Seriously, if you come out, you've gotta come for a visit... you'll freak, I've got like four kids... Hehehe... gone are the heady days when we were designing demos with the title "BTB: Bazooka Toting Bimbos..." Sigh...

    [shameless plug mode on]
    Support these guys... and my novels will someday be published too! ;)
    [shameless plug mode off]