Why I’m sceptical of Kinect

Motion controllers are the future, we are told, and none seem more advanced than Microsoft’s Kinect system. The launch lineup, as revealed at this year’s E3 conference, has a definite casual look to it (which is a nice way of saying it looks easy and shallow), but MS has been quick to assure us that in time the ‘core gamers’ will embrace it.

While this is definitely possible, there are reasons to be sceptical of this, and I’m not just talking about not wanting to look silly while jumping in front of my TV or the physical demands of such games.

During this piece I’ll attempt to describe game genres which, on the surface, sound like they might be enhanced by Kinect, and to show how such ideas actually come with serious obstacles.

Presented here are some of the issues that a system like Kinect has, in no particular order:

Lag issues, both real and imagined.

Actual controller lag probably isn’t any worse than a normal gamepad, and if it is I doubt the difference is that large. But there is also the issue of perceived lag. Large physical actions take longer to perform than pressing a button does, hence need to be started further in advance.

This is a learning issue rather than a technical issue, but early footage of Kinect’s rafting game (where it seems to be possible to jump well after the raft has gone off a cliff) suggests that some developers have already had to compensate for this.

Out of all the problems though, this one is most likely to be overcome simply through the user practicing the game, but it bears noting that motion controls also have a sort of ‘uncanny valley’. The closer a control system resembles a real life activity, the more people experienced with that activity will notice what differences still exist.

Precision lacking - not a full skeleton yet, still some occlusion issues.

There are still a few things preventing the current setup from being a full skeleton tracking system – namely that the resolution is still insufficient to capture small details, like fingers. When so many of our games are based around small, accurate movements like pulling a trigger, shifting gears and hitting frets, a precise controller isn’t just preferable, it’s essential.

Some hands-on reports suggest that in a ten-pin bowling game for Kinect, players can only put spin on the ball by moving their entire arm on release, which is not how it’s done in real life. In situations like this, the game ends up being less intuitive than the existing Wii bowling game we’ve probably all played by now, at least for anyone who can actually spin a bowling ball properly.

In addition, the camera still works from a single, fixed perspective, so occluded body parts still have to be approximated. It’s not for nothing that professional motion capture rigs surround the actors with cameras.

Feedback - not just rumble, how to handle losing sync?

Gamers like force-feedback – just look at the reaction to Sony’s SixAxis controller not having any. The Kinect system has even less. At least with an analog stick/trigger you know when you’ve reached the end of the detectable range. But the far bigger issue is that when your ingame character collides with something, there’s nothing stopping you from continuing.

Let’s look at an example from a hypothetical 1 on 1 fighting game: I try a high roundhouse kick, hoping to catch my opponent in the head, and then continue spinning around to return to my starting position. Ingame though, my opponent blocks my leg, preventing my avatar from continuing to spin. At this point the game has to disconnect (diskinect?) the avatar skeleton from tracking me 1:1 and, somehow, decide how and when to start tracking again.

There’s no obvious solution to this, and I haven’t seen any of the games shown thus far even attempt it. Sure there’s a light-saber game, but in the footage shown at E3, nothing was able to block a swing. In fact, just as two light-sabers were about to collide, they stopped the video. Curious.

Reduced action count - sometimes you need to pull a trigger.

There’s a reason gamepads have gotten more complex over the years, going from the NES’ d-pad and 2 buttons to the dual-analog, 4 face button, 4 shoulder button designs we’re now familiar with – more buttons means more potential actions, which in turn allows for deeper gameplay systems.

While Kinect opens up a lot of new physical actions, the fact that most menu navigation seems to be done by hovering over an option and waiting shows the potential problem here – the inability to do specific, abstract things in a hurry. Kevin Butler is right – I don’t want to point at a screen and yell “bang”.

Nor do I wish to navigate menus of any real complexity (see almost every RPG ever) by hovering over every option for a couple of seconds, and nor do I wish to pause a game by.... how are they planning to handle that again?

Sometimes it can be easy to forget just how many gaming conventions are defined by abstract actions and complex interactions between the player and ingame items. Where Kinect makes some things more intuitive, it requires a whole new way of thinking about many things that are now second-nature to many of us.

Play-space issues - big "rock band" space required.

Probably not the biggest issue of all, but it’s worth noting that not everyone has the kind of space required to play Kinect games, especially multiplayer ones. As I’m writing this, I’m sitting at a desk, which obscures the view of my legs from the TV, and I can only take two steps backwards before colliding with my bed.

Unless I can play everything standing on the bed, Kinect just isn’t an option with my current setup. I’m sure not everyone has such a play environment, but I can imagine quite a lot of people having to find places to store coffee tables every time they want to play a game.

Player inequalities - can all players perform actions as well as required or at all?

Let’s look at another issue with a 1 on 1 fighting game on Kinect. Apart from the collision issues already mentioned, ask yourself this: in such a game, could you kick at head-height? Depending on who is reading this, the answer could be ‘yes’, ‘no’ or ‘maybe on a good day, after warming up’.

In such a game, were it actually using 1:1 tracking, some players cannot do things that others can do, or can’t do them as fast or reliably. While all games have issues of differing player skill levels, it’s rare to come across any situation where some gamers literally cannot perform some actions that other players are able to do.

Is it fair to disadvantage players because of their real-world limitations? I’ll leave that as an open question, but it’s certainly one that will crop up over and over. Looking at the current game selection, the easiest solution seems to be not to present the user with anything that’s physically challenging, but surely that just reduces the potential uses of a system that can do 1:1 tracking?

No accounting for momentum - I can't throw everything at the same speed.

This one is the opposite of the player inequalities problem. Because of the lack of anything physical in my hand while playing Kinect games, I could theoretically throw a bowling ball and a tennis ball at the same speed.

The only way to get around this is to not base the speed directly on my actual motion, but then you’ve lost the 1:1 connection that underlies the whole experience. To give a boxing scenario (surely one of the most obvious uses for Kinect), because the software is unlikely to be able to determine how much force I’m putting into a punch, there’s little incentive for me to put in any.

Much like in Wii Sports Boxing, I might be better off just throwing weak punches out as fast as I can. What seems like it could be a really immersive experience can easily turn into button mashing without the buttons.

Tiring interactions - can we play for an hour?

I’m not going to complain about the idea of having to stand up for an hour, though there seem to have been plenty of people quick to voice such an opinion. Not everyone wants to play a game for an entire evening, but I don’t think I’m alone in wanting my gaming experiences to be long enough to actually feel some kind of progression.

Many of the games shown at E3 looked very short-burst focussed, which is fine for some segments of the market, but it also suggests a lack of depth. This in turn might cause some to question the value of full-priced software compared to the similarly short gameplay experiences you can get online or for smartphones that range from free to a few dollars.

Gamers already feel short-changed when their new game doesn’t last them more than 10 hours. How many of the games we’ve seen so far look like they’ll provide more than an hour or two of gameplay? And what if the mini-game collections for the Wii weren’t just trying to capitalise on the popularity of Wii Sports, but were actually a reflection of what motion controllers are best suited for?

Walking/running - what's more important in gaming?

I had to put this one in its own category. Can you think of a single action in games more common than walking or running? Looking at the games on my shelf, almost all except the driving games involve moving a character around. While the inability to pull a trigger makes developing a shooter a lot more challenging, this one surely makes multiple genres impossible in their current form.

In a boxing game, for example, the inability to move around would mean that I can’t control the distance between myself and the opponent, nor can I dodge, unless I follow every dodge by moving back to the starting position. Attempts to circle around my opponent would ultimately end up with me standing in my bathroom, having lost sight of the TV long ago.

 
Abstraction leads to gesture-based gameplay, which brings its own set of problems.

Some of the issues here can be potentially alleviated by changing from 1:1 tracking to a gesture-based system, one where the system doesn’t map your skeleton ingame, but instead looks for certain movements that the code recognises and then reproduces ingame based on pre-defined animations.

I could (and for that matter, did) write an entire article on the problems of gesture-based gameplay, but I’ll try and sum it up quickly here: Using gestures to control a game results in increased lag, decreased complexity, decreased precision, occasional missed or incorrect actions, etc. If you want to make a simpler, less precise, less reliable and less responsive game, then feel free to go down this route. Perhaps you’ve played a Wii game or two like that?

Conclusion

Before we go any further, it should be noted that there is already one game for the Kinect that could reasonably be called a serious game – Dance Central. Anyone who writes off a dancing game as for the casual market only has presumably never seen a DDR master at work.

For the purposes of this discussion, what matters is that the game seems to have a reasonably large amount of content, relies on players being coordinated instead of frenetic, and provides the potential for mastery through practice. So we should be careful not to say that hardcore games are not possible using such a system, and such games can present gameplay opportunities not possible elsewhere.

But let’s not kid ourselves around. The games you love now, if you’ve been playing games for decades like I have, will almost certainly not work, in almost any form, on the Kinect. All the issues covered above combine to make traditional games near impossible on Kinect.

This is a challenge for the game designers of the world. It may require more creativity than any other change that has happened in this market, including the change from 2D sprites to 3D polygons. But unless you resort to using a standard game controller in collaboration with Kinect, it will be very difficult to create truly deep experiences.

After all, how many activities can you think of that can be done within a living room, that don’t require touching anything in the real or virtual world? If you can answer that, perhaps you could have the beginnings of a great Kinect game design.