Sunday 10 February 2013

User Intent & Skill

I am not a great player of games.  My loop from eye to brain to body (specifically fingers for gaming) is particularly slow and susceptible to stalls (rabbit in headlights).  When I play a shooter online then chances are I will be in the 50% of the population who don't quite hit a 1:1 KD ratio, rather than the 50% who are somewhere above that magic line (which 90% of people claim to be above - don't you love statistics).  My ability to accurately point a crosshair at your upper torso and click, then react to the potential movements of bits of the system to maintain that position and throttle my clicking rate to best provide an accurate stream of virtual bullets to pierce your virtual head is not good and is only partially compensated for by my ability to pick sensible places to engage.  But give me a support class or an objective and I'll make a good addition to your team.  Hell, even in something like Counter-Strike standard maps I'll work out non-traditional disruptive play to prevent the other team from playing in their comfort zone, at least in a casual skill environment (no doubt high skill players can shut down my shenanigans).  This is something I am completely happy with, by definition half of the group have to be below the median and if we're looking at a skewed distribution (assume a normal population distribution of reaction times, most very low players give up, everyone else sticks around) then most of us are below the mean.  I have no proof to support the idea that most game populations are skewed that way so feel free to reject the second idea, that most of you play at a sub-average reaction level.

So what has brought about this claim that not only am I not very good at the see, analyse, react chain but most of you are (comparatively) down in the mud with me?  Does it matter?  Should game designers care?


When I first saw Super Hexagon I thought it was a game about analysing the scene, deciding on a rotation needed for the pointer, and hoping that you'd made a decision about that rotation with enough time for the rotation to complete and get through the gap.  The skill coming from that reaction speed and correctly choosing left or right rotation and chaining those decisions together into the sequence that matches the ever contracting and moving world around your pointer.  The rejection of the music game memorisation by adding randomisation and constant speed of the pointer's movement (baring stationary) seemed to clinch my reading of the game.

And then I purchased a copy and played it.  Or I should say I repeatedly played the first second to 9 seconds of the game.  I was not expecting to have to express my intent for rotation as a press on the left or right side of the screen for a certain number of ticks (with error margin for the width of the hole narrowed by future positional needs / rotation change lag in your input reactions), I was expecting the touch to provide rotational intent despite the lack of an analogue stick.  This leads to my reading of the game as both a call against and demonstration of players' willingness to work around needlessly bad reading of user intent.  I do not think this interpretation was the author's goal but I don't think that should be a relevant factor.

When someone complains about the pointless tank controls in earlier Resident Evil games when played on a device with analogue sticks, that it is harder than it should be to express their intent via their avatar, then this is exactly what Super Hexagon is demonstrating.  A game where the avatar cannot react in zero time to rotate perfectly to a new angle should express that limitation in the animation system, not in the controls.  When people express the difference between their failure and a failure of the game it is usually described as "I made the wrong decision" vs "the game didn't do what I wanted" or unfairly/arbitrarily killed them as they had no way of knowing what the right decision was.  Dark Souls is lauded for the canned animation system (you cannot break from an animation once you initiate it so decisions cannot be aborted) and interactions that are challenging but feel fair and push players to be very methodical and make the right choices.  Hard does not mean lightning reflexes, it means making the right choices and fair means the game gave you the information you needed to make the right choice.

At 5:40 seconds into the game the cursor snagged the end of a wall, the wall I had time to get past but had mistakenly released the right side of the screen early to avoid overshooting the hole.  The hole I had the intent to get through and the reaction times to initiate movement to complete in the right direction.  But I didn't hold down my finger for exactly the right length of time and so the pointer hit the wall.  Stupid game, let me move my avatar to where I want it to go!


While discussing that game with Paul from Mode 7 Games, I was being my usual contrarian self and making no headway expressing how I see the game as tank control analogous, narrative against bad controls by using bad controls and this eventually pushed him to posit, "I don't know how you categorise a disconnect in user-intent vs. making something skill based - don't you need a disconnect for micro".  Now there is an interesting question, is skill (specifically as expressed in micro) just the player having to overcome problematic controls to express their intent efficiently?

It's certainly a sane viewpoint, my counter would be that perfect reading of user intent is hard to impossible depending on the range of things the user is able to express in the game but being as good as possible at reading them is a requirement for an honest game.  For something like Starcraft then there is clearly an issue with users expressing the exact movements and actions of every unit in the game as soon as they think of what they should be doing.  We simply can't work out how to use our current input methods to achieve that and so we build the tools as best we can and consequently the ability to better express intent using fast reactions is part of micro.  But knowing what you want your units to do is also massively important.  The APM to pull back units and prevent them being destroyed is key to being most effective in an engagement, knowing that you want to fight with an army of half-health units by pulling some out of range of attack rather than half an army at full health and half corpses is the skill.  Starcraft even uses that time being locked up at the controls for micro as a balance mechanism, with players who are weaker at micro knowing this and so devoting their APM budget to other things.  Micro is much more than reactions and good play is informed by understanding of your own reactions and application of it to the limitations of the controls; giving the players the best chance to express their intent is critical, while accepting the limitations of input will provide some with ability to do more, faster, up to the constraints of the input system.

I'm sure most people who have played Super Hexagon and read my interpretation of the game didn't agree, I've expressed the view to responses of o_O enough times to expect it.  But what if my Super Hexagon and your game aren't the same game?  Maybe on my Nexus the lag on the digitiser or on the renderer's output / screen means I have a couple less frames to make my reaction, that the feedback loop of when to lift off my finger has to be done by precise mental calculation rather than via screen feedback.  My laggy brain may make for some disadvantage in twitchy multiplayer shooters but what about laggy hardware, especially in the wide ecosystems of PC gaming, Android/iOS (when grouped collectively), or consoles (mainly what TV you hooked them up to).  Is this skill or just a randomised impediment that the game design should accept and try to minimise?

Consider a shooter with a handicap system for health.  When you buy the game you get a health value between 50 and 150 and every time you spawn that's your health, your given handicap was randomly chosen.  Only you play on a nice large IPS TV as your screen, when you spawn your health is now half of what it was.  You own that brand of GPU and those drivers with default settings?  Take 30 health off.  Didn't configure your $60 mouse correctly for 1ms updates?  Drop off another 10.  This is what lag to the input and output are doing, on top of the random lag of the user's own processing abilities we have massive, uncertain lag from the variable hardware and settings.  By not trying to optimise reading of user intent we exacerbate the ability for luck rather than skill to rule the systems because we can't know from user to user if what they can actually see and react to on their hardware is 100ms behind a different user.


This line of thought can be taken to extremes, the simplest straw-man is to declare that I am saying no timing checks are allowed and so all games much become turn-based.  Reactions are part of micro and that's a key element of gaming, especially competitive gaming, and even our experience of reality in general; time moves forward and we go along with it.  But we can't equate skill with enforced disconnects in reading user intent.  Obviously the user doesn't intend to generate a game over screen or lose so our game, to have challenge or provide player rankings, must offer paths to failure that the user should not take but has the option to.  I think games are most worthwhile when that is offering a choice that the user takes and then leads to failure, even if they only took that option due to being rushed and needing to make some reaction (one of which can be no input at all).  We need to be aware of the variable reaction times of the devices making a mockery of any intent to be aggressive with dividing players by speed of reaction as a skill or a skill check for expressing precise intent that can be recorded by other means.

An input system that is harder to interact with than is necessary, making correctly expressing intent the skill, is missing the mark; a throw back to before we had the processing power, input bandwidth, and know-how to do better.  We will always have a problem with correctly reading user intent until people have brain interfaces but minimising those disconnects does not prevent games of reaction, even if the fair balancing of such a game seems impossible due to the range of hardware people use to play games.

That is the hole I see when I look at using hardening expression of user intent as a skill mechanism, with real-time games already at the mercy of input and output lag those games that walk that path are just doubling down on giving luck as skill, luck of the hardware and luck of the nervous system.  Games are about offering choices but in Super Hexagon my best choice it to stop playing.