Monday, November 21, 2011

Skyrim Analysis

This is an interesting analysis of the UI and play models in Skyrim. Its interesting to note how reviews are generally no longer focusing on graphics and performance. I feel like we are past that point (as long as you can throw enough CPU/GPU horse power at the game) My point is that playability and User Experience are now the weak points.

From the review above, the cons are:
  • Combat isn’t very visceral, and victories and losses feel unearned
  • Menus and interface are terrible
  • While the world is wide open, most quests and dungeons are very linear
  • Bugs abound, especially with physics
These issues are the next things to be addressed in games research.  The fact that everyone has been banging on about them for the past two decades shows just how hard they are to get right.   Especially in the context of producing a AAA title across a couple of platforms with different User experiences and interface models.

Combat isn't very visceral
Combat as a visceral experience... interesting idea from the reviewer.  While I agree that its the objective of this kind of game, to make any interaction really immersive is going to take something more than a mouse(insert UI device here) and a single dimension screen to make me feel immersed.  The monitor wall's and head mounted goggles are a big step to remove the need to manually track your head around in an environment, but there is still a long way to go before everyone has access to something like that.  Kinect is a step in the right direction but its too loose at the moment to really work well.  I think it would be much improved with some additional channels of resolution and maybe a few precision channels of data like a head axis positioning system,(crown with some position knobs) precision line of sight and precision location and orientation for hands (maybe some simple cuffs with 3D position knobs) and let the rest of the Kinect data stream fill in the gaps.

The second part of combat being visceral is feedback. Haptic controls have not really evolved much and are still trying hard to solve the most basic problems.  This is more about technology and materials than about intent.  The strategies are still the same.  Immerse the player in some system that can simulate feedback with reasonable resolution.  For a CRPG game this could be as diverse as holding an object, taking blows, falling or flying, jumping, landing, being swept down a river or just walking on a soft surface.  The ability to simulate these kinds of environments safely is not here.  Everyone will want one when they are availible from your local gadget seller... but they are just not here. So wishing combat was more visceral is important to keep the bar high... but its just not feasible to go beyond the limitations of the UI devices we currently have.

Victories and Losses feel unearned
This is an interesting problem that is much more complex than simply a limited UI.  There are real issues with perception and feedback at work here.  To complicate it, individual players will have quite different perceptions and desires in this particular space.

Lets break it down a little.

  • How hard should a player work for an objective? ( be it a victory or some other objective)
  • How balanced should this sub-section be of the overall section of narrative or mission? 
  • What are the highs and lows of the variable reward model currently in play? 
  • How do you keep something fresh, if the scenario itself is repetitive ( attacked by wolf?)
This is as much about player perceptions.  I have written about this particular group of problems and proposed some solutions before but honestly they are just guesses until I have the resources to get paid and solve these kinds of problems.  The strategy that I propose is that both the narrative engine and the perception engine are abstracted and deal with the narrative and player models explicitly as a construct that the game engine manipulates.  (Not that the game manipulates the player, just that the players state by explicitly measured and modeled continuously so the game engine has some concept of player state and can then make decisions based on that understanding.)
The key point is the abstraction of the narrative structure and its relationship with the player state at any point in time. The information in these two models can then be instantiated via the concrete game objects, characters and scenarios, rather than, as currently happens, each of these elements acts as either discreet and independant agents or as a chain of artificially scripted agents who hopefully fit all players equally (badly).
Again, this all comes back to the "commonality" of experience, which is based on a basic desire for repeatability and predictability.  This is essential for debugging... but only until the game engines evolve and abstract all this functionality out into a higher level construct.

So how does all this rambling address the "victories feel unearned" issue?  Well, if the player model was reasonably well developed and informed via some subtle evaluation of the players behavior, the game engine should be able to tune the "encounter" appropriately to provide the player with a challenge of sufficient uniqueness and magnitude to keep the player engaged.  This information is then combined with the current narrative model, to determine if this "encounter" should be big or small depending on the stage in the current narrative arc that the player is within.

This kind of model points the way forward for managing the engagement and satisfaction of players in the game.  Which then points the way forward for using these kinds of  immersive environments for specific training purposes.  

Weak AI
The reviewer took time to critique the same-ness of the AI used in the game, which I feel ties into this point.  For a game of this scope to have weak AI is pretty embarrassing.  I would suggest that after the epic re-write of their engine, the AI subsection is just not as polished yet.  This is a fairly solved problem, so it does raise the question of priorities in the development process.  My guess would be that "good enough" AI was acceptable by the majority of the playtester and they stopped putting resources into that section.  After all, development is a resource constrained activity.

Hopefully the mod community will be able to patch this particular problem or expose the API enough to let others have a run at building better AI for the game.

Menus and Interface are Terrible
There are some novel interface metaphors in Skyrim ( such as the star constellations for the progression tree) which add a very nice immersive touch.  These are more information visualization elements, rather than functional information navigation tools.

The fact that the interfaces have been tuned for navigation by a controller with limited buttons is both a blessing and a curse.  Having a slew of hotkeys availible on a PC keyboard is useful but can present a steep learning curve for novice players, this affects the pace that the game can be played between novice and expert players and has detrimental effects for players who want to play in a "pickup" style.  ( It could also be argued that these kinds of games are probably not intended for a "pickup" audience... they are almost study Sims that need a fair bit of experience before you can get into them.) 

Anyway, by keeping the Interface approachable, it forces the designers to reign in the control explosion that can happen and keeps the cognitive load manageable.  On the other hand, it puts a ceiling on the way expert players can drive the interface.  This is the sort of thing the mod community is probably better at solving than the initial development team; as long as the API's are open enough to allow the modders to rebuild the UI.

Linear Quests and Dungeon
This is where my interest lies.  Again, as I stated above, I believe this is strongly tied to the problem of "commonality" of experience.  (Again, both for debugging and for comparison between players)
The reason that the quests and dungeons are linear is that its simply to expensive to hand tune quests and dungeons with multiple permutations of possible paths.  This is simply a manpower issue that is not solvable by current development teams.  THIS PROBLEM CANNOT BE FIXED WITH CURRENT TOOLS.

The only solution is to make the computer do the heavy lifting (which is what we keep them around for), but for this to happen, we need an abstract way to describe the narrative which the computer can then manage and for this to work, we need an abstract model of the player as a component in the narrative.  See above rant for more details. 

Once we have a functional model for narratives ( .. oh wait there's a bunch already...) that are easy to implement and a control engine that can handle it under resource competition with the graphics and animation engines ( resource budgets again) it should be fairly straight forward to script.  The problem is that "commonality" of experience may evaporate.  This has been happening already so it's probably not going to be the radical shock that I suggest, but it will still be a revolutionary change that will overturn a couple of the conventions that exist in the game world.  How do we know we are playing the same game if it evolves every time we play it?  My experience will be totally different to yours... theoretically down to really fundamental levels that currently we take for granted as completely immutable truths in the environment. Well immutable until the modders take a hack at it anyway. But even mods are fairly static. I am talking about a system that can dynamically mix and match the narrative and experience to the player at many levels.  Unless the player is exactly the same each time, every replay, even by the same player should be distinct.

I still expect some mechanism to emerge that will allow side by side comparisons of player experiences, perhaps the idea of "set" quests that do not dynamically adapt or some sort of set "level" to play at that fixes the player model so the game can be tested and debugged as well as allow "walkthroughs" and other advice services to still have a frame of reference. But beyond these I expect a fully dynamic play experience that focuses on the player and adapts the world to provide the maximum play experience possible using all the narrative trickery possible.

Bugs in the physics 
Ok, the bugs in the physics engine is just embarrassing, but this is kind of accepted for a game of this magnitude and will probably get tuned out in the first round of patches.

The same problems are still around that have not been solved from the first CRPG's.  The thing to keep in mind is that unlike FPS games, CRPG games are usually solo experiences, online MMPRPG's have a different dynamic and will never be the same type of player centric experience.  

Here's a link to the skyrim site for more info and pretty pics

Here's a link to the skyrim modding community (I'm sure there are more sites growing as I type...)

No comments:

Post a Comment