Speedcrafting (noun): The practice of creating and publishing a game in less than 10 hours while streaming the insanity live on twitch.

Our 2-man studio was started with a simple word at its core. Speed. We've always been under the impression that if we make games fast enough we might just be able to succeed before we go broke. Towelfight 2, our critically acclaimed financial failure, took 48 work days to complete and got us noticed by some media and fans alike. Quadropus Rampage kept our heads above the water with its 2 million players, and it too took 48 work days to complete.

Our current project, Crashlands, is now on its 70th day, and will likely take another 30 to finish, if not more.

Somewhere along those 70 days we started feeling like we were going too slow (cancer issues aside). After returning from GDC, we wanted to challenge ourselves and see if we couldn't stand out using the one talent we have - speed. That's when we came up with Speedcrafting.

For the past month, Speedcrafting consumed our Mondays. We set out to extend our Butterscotchy brand, get more connected to our player base, and become better at everything game-related. Heck, maybe even make sandwich-money. And we'd do it in 10 hour publishing sprees that were recorded live, for heckling purposes.


Our original goal was to Speedcraft 8 games over 8 weeks, but after four games (5 if you count Roid Rage), we've found that the marginal benefit of each game has started to drop to the point where we should stop.

Still, we learned a lot through this process. Here are some of the big takeaways:

  1. The faster you make games, the more genre-restricted you become.
  2. The genres you land in when you make games REALLY FAST are also the most crowded genres (arcade, action).
  3. There's a fatigue effect regarding the consumption of any media (newsletter, game, tweets). Just because you can publish a game every week, doesn't mean that you should.

Without further ado, let's dive in to what went **FANTASTICALLY RIGHT **and HORRIBLY WRONG in our adventure in Speedcrafting!

Touchy Feely Goodness

One of our principle goals with Speedcrafting was to humanize ourselves to our player base and give them a glimpse of what we do on the daily. Further, it was our intent to grow the stream over time to rope in as many of our people as possible. To this end we promoted it in the newsletter and to our facebook and twitter followings, a mass of about 30,000 people.

We saw an interesting pattern emerge from our streaming. It experienced rapid growth for the first few streams and basically hit the floor by the fourth game.

stats over time | Note how each color follows the same pattern |

What was happening here? Let's make the assumption that we maintain a relatively stable level of entertainment value over time (DON'T QUESTION IT.) - that is, the quality of the stream wasn't deteriorating over time. With that assumption in mind the data suggests that we had reached a saturation point with our followers - those who were interested were already present by the third stream, and those who were not interested were not being pulled in. AND the novelty of watching game development wore off for most people after one or two visits. Indeed, looking back on the chats we had with everyone, it was routinely the same core group of 13 or so people who were, with one or two exceptions, present from day one. The additional people any given day faded in and out - they didn't stick to the stream.

patient zero | WHY WILL NO ONE TOUCH MEEEEE?! |

Streaming the Speedcraft series didn't allow us to reach our following on a large scale in the way we had hoped. But, it did give us the opportunity to chat with and scream at a variety of the people who play our games, which was exceptionally fun and cool. Development is often done in a cave, far from the sanitizing effects of light, and it was really enjoyable for us to make games as a sort of spectator sport. Though, as we saw, game dev doesn't have the sort of pull that genuine spectator sports do. It's mostly a lot of "two dudes staring intently at computer screens" with bursts of conversation and design sprinkled in.


One interesting note is that the "crossover" between minis (number of players they share between them) is absurdly high - between 25 % and 50%. What this means is that half players of any given mini have also played other minis and/or Towelfight and Quadropus. However, that group held steady at about 250 people throughout all of the minis and didn't seem to grow over time. In other words, we have a core contingent of fans who were following our Speedcrafting, and hardly anyone else noticed!

Sharpen those Skills

The second purpose of Speedcrafting was to further hone our craft. If we made and published a game in 10 hours each week we figured we'd find soft-spots that needed strengthening. This turned out to be COMPLETELY TRUE.



Seth and I have always insisted that we don't design things - we iterate. Our chops come from game jams where whipping up a design doc of 300 pages isn't a viable option. Things must be built, tested, and rebuilt as quickly as possible, and along the way the core, tiny idea is grown, much like a pearl in the mouth of a sandy clam.

The Speedcrafting sessions helped further hone our iteration skills. We worked on 4 different games of vastly different genres and managed to make each one fun, though some more than others. It was a great space for us to try things without much pressure -  3 of the 4 games were made in genres we've never had the opportunity to play with.

We also were slapped with the reality that some genres just take longer to cook up than others. For example, for Speedcraft #2 we made Extreme Burger Defense, a tower defense game involving bears and burgers. We'd never made a game with these mechanics before, so we set out at a frenzied pace to see if we could come up with anything fun in the short window of time we allotted ourselves. It wasn't until nearly the 7th hour that the game became playable, in the sense that it generated some fun. So many systems and so much content had to be in place before it started resonating that we didn't even have time to balance it before kicking it out the door!


And perhaps more importantly, the game's mechanic is so involved and content-dependent that we didn't have much time to try something totally crazy or new. We were flailing around too much just trying to get the thing playable.

Speedcrafting is genre-restricted. The games we're most proud of are Flop Rocket and Freeway Mutant. Both of these fall into the 'arcade' genre. They're small, their mechanics are instantly enjoyable (both involve movement of the player as the core fun-driver), and they allowed time for innovation. The others required a level of strategy to really sing, and strategy takes more thought and consideration than can be speedcrafted.


Seth managed to create a template for game-making after our first speedcraft that pushed the boundaries of what we could do. He did this by scriptifying and systematizing a lot of the things that we have to do over and over again with any given published game.

The last two hours of the Speedcrafting sessions are generally dedicated to UI elements and getting the game prepped for publication. This includes things like hooking up leaderboards, ads, IAP, menu screens, splash screens, credits, and all manner of non-game necessities. Perhaps the biggest boon on the programming side was the creation of some publishing checklists and the streamlining of how we implement all these publishing-related sub-systems. It will save us hours of time on literally every game we make from here on out. True, in a 3-4 month game, a couple hours isn't that big of a deal, but if we were to crank out a game in a couple weeks at some point in the near future, this stuff will really come in handy.

flop rocket | MORE GAS! |

But everything wasn't super peachy. Some of the systems within the games had to take hits in order for us to finish them. Again, with EBD as the example, the Artificial Intelligence of the bears is as simple as it can possibly be, and their pathfinding was relatively dumb. Doing AI right takes a lot of time and, again, careful consideration. When you're sprinting for 10 hours you don't have these luxuries, so all of the enemies have to be as basic as possible, which makes them feel a little bland.

Interface work is also exceptionally time consuming and game-specific. We ended up recycling the UI structure from Flop Rocket for Goopidemic, but we were both unhappy with the result as it made the game feel a bit "recycled" (even though it was a totally different game). These are corners we'd rather not cut but had to in an effort to get the game finished.


I managed to stumble on a tool that sped up my art production massively - COLOR PALETTES.

colors for flop rocket | THIS IS FLOP ROCKET. |

I've never used a color palette to make anything. My usual color-picking is decided by whim and ease of finding the color on the inkscape colorbar. I realized after the first two games that I was making TOO MANY decisions in too short a time. I was always exhausted by about noon. The color palette usage helped me reduce my decision-overhead so drastically that Flop Rocket, the first implementation of said palettes, was a breeze to make and is probably the most cohesive and best-looking game we've put together.

This was a huge win for me and is now being used to speed up the production of everything in Crashlands, from creatures to weapons to landscapes.


We managed to systematize a lot of the legwork that usually goes into publishing as well. A checklist of all the art assets, screenshots, etc. that are needed has been compiled and boosted our end-of-production efficiency by 30%-50% (in terms of how quickly we could complete that group of  tasks).

Further, I got 10+ more hours of editing experience in and we created our own brand of Let's Play videos, which generally consist of me screaming like a banshee while Seth giggles. If you missed them be sure to at least watch Flop Rocket's.


Our goal with the Speedcrafting series, and the minis in general, was never to make bank. We reserve that right for our larger, more splendid titles. But it was our hope that the extremely low cost of production (~$150/game) would be recaptured and eventually built on. We were aiming for what we call 'sandwich money', which is literally enough extra money per day to buy a sandwich, or about $6.


You'll notice that EBD and Goopidemic aren't on this list - that's because both are of a low enough quality (by our personal standards) that we don't feel proper in monetizing them. (Actually, we haven't even released Goopidemic yet!) What we're looking at is just the revenue of Freeway Mutant and Flop Rocket, our 1st and 3rd Speedcrafted games. It's likely that they'll even out to about $4 a day, combined.

It's not quite sandwich money, but it does mean these games will clear the profitability hurdle within two months. Plus, we've seen firsthand the effects of launching a large game on our smaller titles, so it's likely each one of these games will experience a large boost in downloads following Crashlands' publication.

Note: The data above is from the writing of this post, which was Thursday, April 17th. On Saturday, April 20th, Flop Rocket got featured in a few countries, so chances are it'll bring in a bit more cash over time.



Streaming turns out to be rather stressful. Further, we always have more work we'd like to do on the games we make, and the minis turned out to be no exception. What was originally intended as a 1-day only event ended up spilling into the other days of the week. Not in a work sense, but in an exhaustion sense. Tuesdays usually found us nearly brain dead until noon, and even then the amount of work we were able to pump out was worthy of scorn by even the laziest of individuals. This rapidly began impinging on Crashlands' progress, which we found to be... UNACCEPTABLE.

The Fatigue Effect

There's one pattern that underlies all of the data we've examined from the Speedcrafting series. We call it the Fatigue Effect. That's fatigue on our players, not on us. It boils down to the fact that people value things more if they are scarce, and less if they are plentiful. The sentiment can quickly go from, "Finally, another Butterscotch game! WOO!" to, "Oh... yet another Butterscotch game."

By making games every single Monday, and sending out Newsletter follow-ups on Fridays, we began to eat into the attention our fans were able to pay to us. The more frequently we made a game or sent a newsletter, the less it was interacted with, and presumably the less people were inclined to care.

As an example, Roid Rage was the first game we published in 7 months. Much to our bewilderment it managed a feature spot in a handful of countries in Google Play and has amassed a player base the size of Towelfight 2's, ~ 40,000 people.

Freeway Mutant and Extreme Burger Defense have yet to clear the 1,000 player hurdle, though they have been out for weeks (they both have about 700 players as of this writing).

Flop Rocket, on the other hand, was the 3rd and arguably best Speedcrafting title. It experienced little organic growth after its launch so we experimented with an advertising platform, one which we hope to use for Crashlands. With the end of that advertising campaign, which itself was equal to the cost of creating Flop Rocket, we managed to push it above both Freeway Mutant and Extreme Burger Defense.

Global Cooldowns

Too many games in too short a time means less attention, fewer (if any) features, fewer available genres, fewer new players, and lower possibility of turning a profit (because of the aforementioned). We've now established that our previous production cycle of 8-12 weeks is likely a sweet spot for releasing games: It's not so long that people forget we exist, but it's long enough for them to be excited for the next Butterscotch game.

And that's a relief. Speedcrafting has been exceptionally valuable to us internally as a studio, but has proven to be quite ineffective at helping us achieve the external goals of higher fan involvement and business sustainability.

Still, we may dig into the ESSENCE of speedcrafting in the not-too-distant future. If we are involved in a project that's planned to take about 6 months to make, we'll make sure to take a week or so break from its production and release a smaller game in the middle.

TL;DR Version -- Making games incredibly fast is useful. Publishing games incredibly fast is much less so.