bit-tech.net

Game Phone Home!

My name is Cliff Harris, and I want to record your thoughts. Don't worry, it's harmless. It might be a bit difficult though, so I'd really appreciate it if you would let your video games connect over the internet and upload data about how you play instead.

We need to learn that sharing usage data is good. Right now, a singleplayer game that connects to a developers website is automatically considered evil, regardless of intent or reality. This isn't without good reason. Intrusive DRM systems and suspicious data harvesting without asking permission have made gamers rightfully defensive of what programs get allowed through their firewalls.

This reaction is understandable, but it's also holding back game design.

Lets look at why a game developer wants his games to phone home, and why you shouldn't mind – or even be keen to enable it!

The first reasons are technical. Gathering data from your customers is the only real way to know what machines are running your games. On the PC, no two boxes are the same. And no two games have identical demographics. It's good to know so many people have bought widescreen monitors but how many of them play Democracy?

I need to know what screen resolution my customers machines run at if I'm to make sensible decisions on default (and supported) resolutions. Knowing how much RAM and CPU horsepower those machines have is another great piece of data that lets me as a coder tune the game for the perfect balance of performance against shininess. Note; nobody cares what the average new PC has inside it, or the Steam Hardware survey - what I and fellow devs care about is what my customers PC's are like.

Knowing the hardware is great, but knowing the software helps too. Should I use Windows XP and Vista-only code? The only way to know is to see how many people played the last game on windows 9x machines.

Knowing what revision of video card drivers people have would be good too, especially if you manage to auto-capture performance data at the same time. Imagine checking the support page on a site and seeing that 'driver revision X for this card improves the FPS by an average of 34 percent'.

Technical stats are great news for the coders, but designers also have reasons for auto collecting data. How many people got to level six? How many attempts did the average gamer need to complete that mission? How many found the hidden passage? How many people completed the game?

This data is all gold dust to designers. Without it, we basically throw a huge pile of features and content into the game and cross our fingers. We have no idea what bits people are spending time on, or where they get stuck, so no idea what to put in patches or expansions. Listening to the 1 percent of gamers on forums is great, but hardly truly representative of the majority.

I'm really keen on improving my game design skills, and want to go beyond just reading my customers opinions on forums. This brings me to the mind-reading dream...

Back to top

Cliff Harris

This is how I'd like to do playtesting for Kudos 2. I'd like to watch 100 people play the game, with video cameras capturing the game footage, synched up with footage of their facial expressions, and biometric data that recorded pulse rates, brain activity and whatever physiological signs of boredom, stress, or happiness I could get hold of.

If there was a way to capture all the neural activity of those gamers and have detailed charts of emotional states and thought processes, that would be just fantastic. .Some industries already attempt this. People have shown different adverts to people whilst inside MRI scanners. It's not science fiction.

Anything less than that, means I’m just guessing. I reckon that this part of the game is funny. I guess that this screen isn't too cluttered and confusing. But I don't really know if that meshes with people who are playing the game for the first (or thirtieth) time. Only by really having the data can I fine tune the gaming experience to be as good as it could be.

This might sound scarybut the good news is, that if you sell thousands of copies of your game (crosses fingers), and collect some basic activity data, then even without looking at peoples brainwaves, you can infer a lot of feedback statistically.

If people click the load button, then soon afterwards click the back button, then the new game button, and this happens with 22% of gamers the first time they play the game, then its pretty clear those button icons are poor, or the positioning of them is wrong. If only 3% of players ever use the "Zeal" ability, then maybe the cost of that ability needs tweaking. This is all great news for the games designers, and thus great news for players, because it means the designers can make better games.

None of my games so far have automatically tried to send usage data back to a server for collating. It's just too scary as a business idea because there is a passionate (and disproportionately vocal) minority of gamers who vehemently resist any attempt by a game to send data. Even if you pop up some big dialog explaining what all the data is, and what it's used for, it still scares people. The minute you have to tell people what you are doing is harmless, it sounds dodgy. Having it default to 'off' means you get a self selecting sample of people who enable it, which isn't as helpful.

I would LOVE to collect very, very detailed usage data on all my games. And I know it would enable me to make them better in patches and sequels. I just need to find a way to ask players of the game to trust the developer.

So, go on, trust me to watch you play my game, I won't be watching your thoughts, but I’d love to hear them…