Feedback and Difficulty in Video Games



I recently finished playing through Halo 2’s legendary co-op campaign, which, for those who are not masochistic gamers, is considered one of the most poorly balanced, excessively difficult games ever made. If you haven’t played, imagine being a delicate butterfly, made of crystal, trying to survive in a rock tumbler. With another butterfly. Who is tied to your wings, so when one of you dies, you both do. And the rocks are actually trying to hit you.

That is Halo 2’s legendary co-op campaign.

The sheer willpower it took to play through this game is sort of astounding. We’d replay the same section dozens of times. Sometimes, we’d replay so many times, the game would punish us for being terrible and would send us back in the level, erasing progress. It could take literally hours to clear a room.

So, why did we do it?

Just because something is difficult doesn’t deter people from mastering it.

Halo 2 (and the Halo series in general) is exceptionally well-designed from a gameplay perspective–ignoring the balance issues regarding difficulty. The enemies are tough, take advantage of your mistakes, try to flank you, use cover effectively, and even sacrifice themselves to hurt you if they know they are about to die anyway. You start with all the abilities you will ever have, so your progress is entirely dependent on your skill as a player.

At the same time, there is rarely just one correct way to accomplish a goal. Sometimes, a straight up firefight is the best option. Sometimes, sneaking around to the high ground is better. Sometimes, you just run and pray. One of our favorite strategies involved using weapons outside of their normal range by having one teammate spot and report where the other’s shots were landing.

Thus, the game encourages you to experiment and iterate. We might enter a room and try something that failed miserably. So we’d alter the strategy a little bit and get better results. Then we’d perfect the execution and get even better results, but still find that we had very little margin for error. A small adjustment in our timing might give us a bigger margin. And so it goes. Eventually, we’d have a perfectly timed and executed tactic involving jumping gaps, punching specific enemies at exact moments, swapping weapons in the split second it took them to respond, stealing a headshot on a hidden sniper, and then ducking for cover. All this had to be done in perfect sync. It was very much like Tom Cruise’s character in Edge of Tomorrow.

You are rewarded for exploring, iterating, and improving. You are punished for not taking cover, being sloppy, and failing to communicate.

It doesn’t matter that the game is brutally hard. You know why you succeeded, and you know why you failed, and it is within your control.

That’s not to say we didn’t have moments where we threw down our controllers in sheer frustration at the difficulty. But we knew we’d get through things. We knew we could get through things, if only we were willing to stay focused.

And the challenge was its own reward.

Good pedagogy

Halo 2 is “just a game,” but the fact that something so brutally hard and frustrating can be compelling led me to ask, “why is ‘education’ less compelling?”

This led me to ask when education was compelling: when have I taken a class that I loved and really connected with? There haven’t been many:

  1. Calculus III
  2. Discrete Mathematics
  3. Introduction to Number Theory
  4. Euclidean and Non-euclidean Geometry
  5. Probability
  6. Computer Architecture and Assembly Language

There are specific reasons these were compelling.

Calculus III

A professor committed to pedagogy. Feedback was quick because quizzes were given almost every class. There were group projects that took us out of our comfort zone, and we were allowed to approach them in whatever way we wanted.

Discrete Mathematics

Again, a great professor who focused on our learning, not just conveying the material. Quick feedback via weekly quizzes. HW that was not overwhelming busy-work, but gave us time to think. I could self-assess to some extent because I understood enough of the subject matter to know what made sense and what didn’t.

Introduction to Number Theory

Another professor who was passionate (wrote about) teaching and didn’t just teach the subject, but also how to learn the subject. Regular and fast feedback with numerous small exams (4 instead of the usual 2) and active office hours. Moderate HW. Project with lots of flexibility.

Euclidean and Non-euclidean Geometry

Another professor passionate about teaching. The entire class was projects and it was a little too new-age for my tastes, but what I really loved about this class was that you could re-do any of the assignments as many times as you wanted until you got the grade you wanted. So I didn’t panic on HW. If I didn’t understand something, I did my best, turned it in, then revisited it over and over and over until I did understand it. I spent a lot of time with stuff I struggled with, and less time with stuff I didn’t. Magic.

Probability

Maybe not so passionate about teaching per se, but the professor recognized that probability is hard and met us halfway. Frequent quizzes. Accessible office hours.

Computer Architecture and Assembly Language

It’s a programming class, so feedback is pretty much instantaneous: your program runs or it doesn’t. The class was really, really hard though, but we were allowed to experiment and explore. As long as the program compiled and ran, you could submit it and get credit.

None of these classes were easy. I had weeks where I spent 15+ hours on homework for a single one of these classes. So what do all these have in common, and what does this have to do with Halo 2?

Now, compare this to traditional education

The feedback loops are slow. In most of my college courses, HW feedback took a week, at which point we’d be on to the next topic which depended on an understanding of the previous material. Two, maybe three, quizzes/exams a semester was the norm.

The consequences for failure are steep and permanent. Despite that, you are often allowed to progress, telling you that gaps are not important or don’t exist.

Trying a lot will still generally guarantee success, but trying is costly (and without fast feedback, you don’t know if you’re even trying the right thing). If you fail high school, there’s stigma to overcome. College is expensive and very, very time-consuming. A second, or third, run is not trivial.

Imagine if a game worked that way. It would be horrible. Nobody would play such a game. If you died enough times, the game would just lock you out and you’d have to actually buy it again.

Obviously, this is a worst-case example. There are schools that provide well-structured pedagogy, but even those implement systems that allow students to move forward despite having failed to master previous material.

Consider Khan Academy, which I used to prepare for returning to college and loved. Feedback is instant: you know immediately when you get a problem right or wrong, and you are tested constantly, because every learning module is basically a mini-quiz. You can’t move forward until you demonstrate mastery.

Successful students accelerate the feedback loop

Many students who succeed at school spend a lot of time figuring out the game being played by the educational system, and a good school provides lots of quality feedback to all its students.

But if a school isn’t going to provide that information, the students have to figure it out. And that’s where I think colleges and universities often fail. Students show up to university having succeeded (or not) in high school. Often not even knowing why.

They are told that university will be harder, but not exactly how it will be harder. Students think the material is harder and the professors are more demanding. But I think the reality is very different: college is more like real life, in that feedback loops aren’t built into the learning automatically.

Thus, a student who goes to all office hours and help sessions is considered dedicated and hard-working, and that’s the reason they succeed. This is a vague, subjective explanation, never mind the fact that it is literally not possible for every student to attend office hours. A better explanation highlights the benefit of office hours:

These students are accelerating the feedback loop.

Study groups are another way to accelerate the feedback loop, by comparing your answers to your classmates'. These don’t provide “the answers” but the idea that a student should come to office hours looking to get a sense of the correctness of their approach is often frowned upon by professors. But the professors I learned the most from were those who simply gave me feedback, rather than giving cagey responses or challenging me for asking about the homework instead of engaging them in speculative discussion about the edges of their field.

This isn’t taught.

As someone who returned to college when he was 29 to pursue a degree in a completely different field, I have been interested in the process of learning for some time. I didn’t find anything besides “study skills” in school. Outside of school, all the advice I found was of the “study skills” variety without context: use spaced repetition, use memory techniques, use diffuse vs. active learning, record your mistakes, self quiz.

None of these (very useful) actions provide context or a target. What should I use spaced repetition on? The material my professors provide is not always what matters for understanding.

The question I wanted (and still want) to answer is, “how do I take a body of (difficult) knowledge and structure it so that I can learn it?” Tallying the number of hours I dedicated to Halo 2, clearly there is a way to structure hard stuff that makes it compelling.

Maybe it is, “just a game,” but I think there is something valuable to be learned from the way games approach the problem of convincing their players to persist in attempting hard things.