Greg Wohlwend recently posted a fantastic postmortem on TumbleSeed, now two months from launch. This postmortem really delves into on of they key elements of TumbleSeed, the difficulty, how it developed, and how it effected the game’s reception.
The postmortem originally appeared on the aeiowu blog, and has been copied here with Greg’s permission.
I remember a drawing I made in my first year of design school. It wasn’t anything special. A simple still-life of various fabrics pinned against the wall done in charcoal. At the time I was only comfortable drawing mechanical or symmetrical things from life. Crates, books, bottles and cans. That sort of thing. So I had to work really hard to capture the volume, patterns and chaos of the cloth. I worked on it constantly.
When it came time to show our work it wasn’t the best drawing up on the wall, but I felt good about what I’d accomplished. After class was over I filed it away in my portfolio and stuck it between the bed frame and wall. As the days went by I would occasionally imagine the still-life in my head. Each time I did I’d chip away at my satisfaction with it. After a few weeks of this I had convinced myself it was complete shit so I’d pull out my portfolio and have a look. It wasn’t.
To this day that drawing stands up to that test. I’m still proud of it. I’ve worked on only a handful of things I can say that about. TumbleSeed is certainly one of them. It’s not shit. It’s good, even.
Benedict and I took nearly two years to design TumbleSeed. When we struck on the final design it was hard to believe. We had been burned by so many “this is it” ideas that when the final one came around instead of celebrating, we nestled into our familiar habit of worry. That this would yet again be another disappointment.
We’d arrived at a design full of challenge, strategy and tension. And every time we checked back in to assure ourselves of that, it did. But that assurance was never more in danger than after our launch. And yet, even with the post-release depression, poor reviews and bad sales I fired up the game alone on my couch a week or two after launch and was reminded by the simple act of playing it. This is it. It’s good.
But yet, it was not enough. How come?
What went wrong?
We released TumbleSeed on May 2nd to the critical consensus that it was “too hard.” Large outlets gave us tepid scores and though others scored us higher, it was a too big hit to the collective opinion of our potential audience. Many considered TS unfair and unforgiving. That’s the wrong kind of hard and this stigma permeated the discussion of our game.
This impact of this can be seen pretty clearly in our poor Steam sales. A glance at our SteamSpy will give you an idea of those numbers, and I believe those are largely due to the very unified critical response that TumbleSeed is “too hard.”
Steam users are an educated bunch that make well-researched purchase decisions. They are also inundated with new games constantly. So if a weird new game doesn’t have at least one glowing recommendation, why try it?
Throughout development TumbleSeed’s stream-ability appeared to be one of our biggest assets. Runs amongst us were always engaging to watch and events brought in crowds that joined in for the tension of rolling. Even if a newcomer didn’t quite understand all the intricacies of TS, they could certainly understand “don’t fall in the holes.”
Yet that too was a let down as many streamers and press cooled off on potential series’ covering TumbleSeed. While many streamed for a little bit, it never quite stuck with any of the mainstream streamers. In fact, those audiences seem to particularly dislike watching TumbleSeed.
We also set up a contest called the $500 Daily Challenge that didn’t perform how we hoped. No outlets picked up the news when we announced it. There was some minor buzz on Twitter, but ultimately we decided to downplay the contest. A contest like this acts as a multiplier of sorts and given the tepid response we decided it wasn’t worth multiplying such a small fraction.
Sales did not live up to our expectations either. While things could have certainly gone much worse, our hopes were to recoup our costs within the first month or two. That didn’t happen, and it’s looking like it never will.
As it stands right now, calculating an average of 40hrs/week/person for the duration of the project, we’ve each made about $15/hr for the hours put in based on our revenue split. The trouble is that we worked many more hours than 40 per week. During the last 6 months of the project Benedict, David and I did little else but work day and night without weekends. Accounting for all that we likely earned something closer to $10/hr.
Our costs were fairly sizable for any project I’ve worked on. We spent about $25,000 on things like legal fees, our contest, expo booths, festivals, travel and anything else we might of needed along the way. Over years of development things start to add up.
These costs do not include our own living expenses, which of course, were the largest expense of all. While often overlooked in these kinds of calculations, I am including them in our concept of “recouping costs.” For us to get close to that, we’d need to sell a little over two times of what we have already sold.
It seems that most players agree with the critics. TumbleSeed is “too hard.” And they’re right. Or rather it’s a completely valid point. Though I hold in my heart that the foundation of TumbleSeed is not fundamentally flawed to the point of damnation. However the TumbleSeed we released on May 2nd certainly is flawed and worth damning.
But unlike press and streamers, we have solid data from our leaderboards about how many people are actually hitting their heads against our game and where.
Percentage of players that…
0.2% Beat the game
0.8% Reached tne Summit
1.8% Reached the Snow
8.3% Reached the Desert
41% Reached the Jungle
While I wouldn’t expect this distribution to be perfectly linear, the sharp drop off after the Jungle and even in the Forest lines up with what critics and players are saying. “I quit after playing the Jungle”.
So that’s how it all went down. And in this case the verdict seems extremely clear. The game is “too hard.” Anyone who’s heard of TumbleSeed could tell you that. But it’s not so simple. It’s important to be precise, so I’d say the real core issue is that “Players find the game to be too hard.”
Why do players find the game too difficult?
The reason I make the distinction between the game’s difficulty and its perception is because after watching reviewers and players play our game over many days, it became pretty clear that we left a gaping hole in our game. We knew this because these are the strategies we would hear about from struggling (most) players:
“It’s best to just avoid all the enemies.”
“Go as fast as you can in the Jungle.”
“Only focus on the core powers, the rest hurt you.”
Employing these strategies is like playing the game on the hardest difficulty from the outset. In fact, these are the exact strategies employed by TumbleSeed speed-runners. They only work if you can beat the game in under 10 minutes and have full mastery of the controls. While they might net short term gains of a few meters higher up the mountain, they will dig you deeper in a hole and strengthen bad habits.
And that’s the real problem. Players become desperate because they aren’t even close to ready for the kind of challenge Adventure Mode poses. It’s too hard. And it’s too hard because there are too many new things going on at once. It’s undigestible.
New control scheme to master
New game system and rules to internalize
New mountain to understand (randomly generated every time)
New enemies to learn (sometimes multiple)
New powers to utilize (some that are dangerous to you)
With all of these, it’s no wonder players feel overwhelmed and end up regressing to a primitive “go up as fast as you can” strategy. It’s sort of like kicking your kid out of the house right after grade school. They won’t survive very long, and if they do, they’ll certainly be worse for wear.
So the issue is not necessarily that the game is too difficult, but that players feel overwhelmed with far too many things to learn at once. On top of this, they also feel overly punished for not learning them fast enough. It’s a pressure cooker filled with gunpowder that only a monk could endure. (And this might be why I love our TS community so much).
How did this happen?
When a game is accessible, it doesn’t just mean it has that quality, it’s also a kind of aesthetic. TumbleSeed is a game that looks accessible. Some may call it “casual.” Regardless, it presents itself clearly with bright colors and a distinct visual language in hopes its inner systems will be more easily understood.
This aesthetic and philosophy is core to TumbleSeed. The simple controls that end up providing an incredible amount of nuance (inspired by Ice Cold Beer) can be picked up by all ages and experience levels. Because of this, the game lends itself to look and feel “accessible.” And sometimes that’s a liability.
On Steam, for example, an accessible game can appear to be a childish game. The audience on Steam may reach for something that looks inscrutable and shadowy far before something clear and colorful.
But despite all of this, I don’t regret our aesthetic choices. What I do regret is not fully following through with our game’s accessibility. This goes back to the abyss that players fall into shortly after finishing the tutorial. The whole game is designed to be accessible with the exception of that huge gap in the middle we forgot about.
This disconnect between what our game appears to be and what it actually is must have contributed to the bitter taste in the mouths of so many TumbleSeed players. While an accessible game is often not so challenging, it’s my belief that it’s not the challenging nature of TS that felt like the anomaly, but its functional inaccessibility.
We didn’t have enough of it.
Nobody ever does, of course. As development continued on TumbleSeed it became more apparent that the Switch would likely be our biggest audience. And because of that, we pushed ourselves to release as an early Switch game ASAP. That pressure helped us make tough cuts that ultimately made the game better and more focused, but it also drained us of our ability to see our chasm-sized blindspot of accessibility.
While a month more would not of made a difference, giving ourselves another six could have. With the pressure off we would have stopped crunching and quite possibly might have seen it. Friends around us were saying “it’s too difficult” right to our faces, but amidst our nonstop-no-days-or-time-off working, we were too dull to let it inspire action.
Besides, we needed critical distance to parse what “it’s too hard” really meant. Under duress, we might of just tweaked the numbers a bit which I’m pretty confident wouldn’t of changed anything.
“Filler” is a bad word when it comes to game design. Or really anything you consume. Why wouldn’t you want pure parmesan cheese? Hold the sawdust, I say. But with games I’m not sure it’s the truism that purist designers would have you believe.
We have always wanted to funnel people to Adventure Mode as quickly as possible. Even as the tutorial stands now, we still feel like it’s a tad bit too long. After all, we want players to get to the meat! We introduce new and interesting concepts by the fistful with each new area players discover. But like I mentioned before, it’s too much all at once and none of it gets absorbed.
We needed to fill our inaccessible chasm with filler and we never considered it because we were too concerned with staying true to the roguelike-y design goals of having the first section of our game be interesting even if it was the 1,000th time playing it.
This is a real design snag when it comes to perma-death games and we held to it too closely.
Our thanks to Greg for letting us share this incredibly insightful and personal look at what went right and what went wrong with TumbleSeed.
TumbleSeed just had a MASSIVE update titled “The 4 Peaks” which fundamentally changes the game. Check that out here.