Like many of you, I bought Mario Maker on the day it came out, and it immediately devoured most of my weekend. The playful interface recalls Nintendo’s premier software toy, Mario Paint, both through direct homage (that save screen music? the rocket and dog?) and shared patterns of interface design. I let Nintendo’s choke-collar restrict my access to the game’s components for my first few stages, before I grew too impatient at being asked to wait more than a week for new content, and looked up on the internet how to spam my way being to 100% unlocked. After grinding my way through the Nintendo World Championship levels (augh) and ordering a Pixel Mario amiibo online at ludicrous scalper prices (augh) I have everything (or will, once my package arrives) and have had a wonderful time building levels.
Playing, though, has often been a different story.
In my 2014 paper “Changing Level Creation to Identify and Filter Low Quality User-Generated Levels” (which you can read at http://www.fdg2014.org/papers/fdg2014_paper_13.pdf) I looked at user authored levels in a somewhat different environment: my programming puzzle game BOTS (bots.game2learn.com). We observed a number of different anti-patterns created by players with specific metagame objectives, and all of those anti-patterns appear with similar frequency in Mario Maker, even after whatever filtering Nintendo is applying behind the scenes.The game offers players three ways to discover and explore user-authored puzzles. There’s a “leaderboard” offering access to the most played / most “starred” levels. Once you have found a level you like, you can “friend/follow” its creator, and be notified whenever they release something new. Finally, there’s the “100-Mario Challenge” which presents a progression of levels automatically selected, which seem to measure difficulty by death counts in those levels, and quality by stars (per level) and medals (per creator) which act as “ratings/badges” and enable some collaborative filtering to group similar level types.
All of these approaches take a traditional “social gamification” approach to filtering and presenting levels. They rely on players to perform most of the filtering work by “starring” and playing, and put only a tiny bit of the evaluation burden on the creators (who are acting as players, as well) by giving them tools to iterate and requiring them to clear their own stages. The problem is, players tend to react according to their playstyle when given social reward systems, and few of those ways result in GOOD LEVEL DESIGN. Mario Maker is already focusing on the playful aspects of game design, so it’s surprising that it falls so short here; by providing creators with the right kinds of playful objectives, the overall quality of levels could be dramatically improved. What’s more, we could collect “shortcut” data to help us filter out levels that we might not want to include in level progressions.
So what can we include to make the level creator our ally in our quest for good level designs? Here are my ideas, based on my experiences with level editors in other games and my research into building a better one for my game, BOTS.
How about this?
The tracks in the game are one of the glitchiest, most frustrating features, but they enable a lot of really cool gameplay. Trails like this enable even more interesting puzzles and challenges. They’re invisible, and can cross each other. They can pop out of ? blocks. Coin trails can lead to cool places (guiding the player) or create platforms with P Switches. Trails encourage players to provide visual cues to their users, but in a fun, indirect way, compared to the static, passive signposts (which really ought to be available from the start anyway.)
Players WANT to make super-difficult levels in Mario Maker, and they WANT to make the largest level possible. This pattern of maximizing the usage of available space was present in BOTS, where users would often build a full 10×10 level only to use a small portion of it. After stretching the canvas to its full width (and using the entire subzone, and occasionally doubling back a couple times) levels can end up taking several minutes to complete… and without checkpoints, even something of moderate difficulty can quickly become infuriating. I’d predict that the optional ability to place a single checkpoint in the level would have several consequences. First, authors would build a safe “resting place” for the check point. Right now, level designers have nothing forcing them to consider pacing other than their own playthroughs, so difficult levels tend to be high-tension throughout the entire stage. Second, authors would augment the difficulty of the second half. Adding a checkpoint would necessitate two completions before uploading, to ensure that the checkpoint doesn’t trap users in a no-win state. This means authors would experience the second half of their level twice, and could increase (or decrease) the relative difficulty.
3. Secret Exits and 2. Gold Coins
The best and most memorable levels from past Mario games feature well-hidden secrets and encourage exploration. However, in Mario Maker, there’s not only no incentive to provide hidden areas for the player, there’s also no meaningful way to REWARD the player when they do find them. You could use 1-ups, but when levels are played in isolation they’re meaningless, and in the 100-Marios mode, you… well, you have 100 Marios. Using power-ups as rewards changes the pacing of the entire level, using costume-shrooms as rewards only works in SMB mode… what’s a designer to do?
Even if they don’t go anywhere, secret exits feel subversive, which players often find to be entertaining. Placing the key or keyhole somewhere in plain sight while hiding the other one offers a visual cue for the secret objective. It also makes the player curious and encourages them to explore. That’s something I’ve managed to achieve in a couple of levels by placing a P-Switch in plain sight with no coins or blocks nearby… but there’s only so much that can be done with that, especially since the brick blocks can be broken in numerous ways.
Star Coins have a similar function, rewarding small explorations, without throwing off the entire pace of the level. What’s more, they provide us with a means of both controlling and measuring the difficulty of created stages. Let’s say that levels with Star Coins would be more popular, and more replayable. Next, let’s that in order to be considered for the 100-Marios challenge (or maybe a new, similar mode), levels MUST include Star Coins. Imagine that authors must collect ALL of the Star Coins in their run-through, before publishing the level, and that if they place 1 Star Coin, they must place 3 (or 4, if we’re using Yoshi Coins instead.) This would have two important outcomes.
First, adding Star Coins would prompt level designers to place small secrets or mini-objectives (at least) in each level. Players would secret them away in tough-to-reach places, behind trickier challenges than the rest of the stage. This would allow authors to include tougher-than-usual challenges in parts of their levels that aren’t required, letting adversarial designers have their proverbial cake.
Even in the worst case, where the designer was uninterested in using the coins, we’d have a lot of interesting data to use to filter levels. All 3 coins placed close together? We can tell the author is trying to cheat their way into the progression. A coin that only the author found, after a number of plays? Maybe the stage contains the kind of secret only the author would know how to find. Right now Nintendo is seemingly using mostly player deaths to evaluate level difficulty and quality. Using Star Coins we might be better able to estimate this, leading to more satisfying level progressions and less skips.
For a tool about game design, Mario Maker gives the designer precious little information about what players are actually DOING. As a designer, I want to see where players are dying, where players struggle, where they quit. What parts do players “solve” and then not have trouble with in the future? What parts are consistently challenging, even if the player has solved them before?
Mario Maker gives me Xs.
Seeing full player traces would be great, but not necessary. Without adding any additional data logging to the game, the ability to click on those Xs and edit that part of the level would be great. The ability to see those Xs when editing would be great. Pop-ups alerting me when players are consistently dying on one part of my level an not continuing would be great. The game collects this data but the player has to DIG for it… surfacing it would be great. Presenting it to players, teaching them how to address feedback and release new, improved versions of their levels?
That’d be amazing.