Be Curious, Test your Ideas!

Run Experiments, you never know which of your ideas will work

Run Experiments, you never know which of your ideas will work

Small Experiments

When you're building a game or a world, you should start small. I made the mistake of not starting small with Mech Mice, and it hurt. I want to help you avoid some of the mistakes I've made.

Some people call early versions of games 'prototypes'. I prefer to think of them as 'experiments.' A prototype comes later, after learning from experiments.

Each experiment tests the growth, retention, and conversion of a game. For example, with the Box Critters experiments:

  • Experiment 1 - growth

  • Experiment 2 - retention (collecting items)

  • Experiment 3 - growth and retention

For example, only 1 in 1,000 players that started Mech Mice returned to play again. That's less than 1% of players. In contrast, 20 out of 100 players (20%) return to play Box Critters. And that's just with an early, one-room experiment.


In Box Critters, I launched an Easter egg hunt. And afterward, someone from the Hyper Hippo team asked me, “Why? What’s the importance of the Easter egg hunt?”

The truth is, I could care less about the Easter egg hunt. What mattered about the Easter egg hunt was that the Critter character could now activate a trigger on the server. A trigger feature was required before I could add a second room to Box Critters. Otherwise, the doors couldn't link.

Triggers link rooms and will make the game. Every game is really just a set of rules. If you play any of the LEGO games, it’s really just, “do you have the block, did you build the block to get the door open?”

When you're starting small, you have to take tiny steps. A trigger feature is a little step. With the Easter egg hunt in Box Critters, the trigger feature was being heavily tested — and there were so many bugs. I couldn't even believe how many bugs there were with the Easter egg hunt. I didn't expect as many bugs as we found.


The very first step in building a world is to have a vision. It can be grandiose. You have to have a vision, and you have to race for it.

With Club Penguin, I had this big Snow Blasters idea - a massive, Advance Wars-esque, multiplayer snowball war game with thousands of concurrent players on-screen. The idea was huge. I had the idea that you could take all these stacks of blocks and build your fort higher and higher and higher like in Minecraft.

But the idea was too big.

At the core of the idea was, "can you chat?" Because if you can chat, you can do anything. It’s a signal back-and-forth. Can I make the characters display on-screen, and can I make the characters move? And that sounds so simple — can you chat?

But there is one feature in there that is incredibly difficult, and that’s not the chat. It’s the persistence of joining the session. In a regular chat room, you just join. You enter and the next message is received, or you see a log of past messages.

In Club Penguin or any virtual world, the state of the world in its current form has to be passed to the users when they arrive. You have to maintain that state. Not only do you have to preserve the state, but you have to relay the state.

"I am currently standing here, facing this way. I don’t care what I said because that was the past. But my state is the present state."

That’s a lot of work. It's a big feature to update the state.

Learn as you go

Start with one room, one character moving around. Get that working. Then go from there. You’re going to learn as you go. That’s the learning. Eventually, you’ll get enough of your tools together that you can build a product. Your product might not look like your experiments. You need to learn each of the steps before you get to the product.

In my mind, a product connects your core game loops to the player's motivation. For example, a Collector player has goals, like, "I can display, I can earn, I can collect, I can organize." In my mind, once I have those loops done, then I have a full feature, and then a product.

After that, you can start to build a business around the product. Product tests go a little further - does something in that loop create value that I will generate income from? Then it’s a product.

Test the important stuff first

The next big thing I'm adding to Box Critters is the ability to collect items.

Based on lessons learned from Club Penguin and AdVenture Capitalist, items are essential. In the early experiments leading up to Club Penguin from 2000-2004, there were no items. Items were added in 2005 to Club Penguin, but they weren't in the experiments.

We learned that ITEMS ARE IT. The items are WHY YOU SHOWED UP. My DAUs (Daily Active Users) and retention are through the roof now because of items. The 30-day retention rate of Box Critters is near 10% right now. And it’s being driven by this drip of people showing up and collecting the items.

Remember, I'm still keeping it super simple. One room. One character. Now with items. And I have a flood of items coming. One room. One character. Items.

Traffic is Valuable

I held myself back years ago. In 2004, I used a robot.txt tag to say that I’m not a Flash game website. At the time, RocketSnail was one of the top 3 Google searches for Flash games, up there with Miniclip. I was getting tons of traffic from people just visiting my own website. I didn’t realize how valuable that traffic truly was back then.

I had a million users per month visiting this website and playing games. If you look back at the site via Alexa, I was trending much higher than a lot of other websites.

So when I put the Penguin experiments up, they immediately had traffic. Thousands of people every day just browsing around the internet looking for something fun.

Experimental Penguins was here on the site. It was just an experiment to see if I could build a multiplayer game. It was the first experiment of many, and I learned a ton. It just exploded. It took over most of my traffic, got featured on Cool Site of the Day, and got tons of links.

But back in 2000, the bandwidth was just too expensive. I couldn’t afford to run the server. It was actually being run at my friend’s office. We didn’t have proper hosting for it or anything. I still had dial-up at the time.

I took Experimental Penguins down, and it was gone. But people never stopped asking for it to return. The problem was that the hosting was so expensive.

Freelance Work Can Help You Grow

At the time, I was spending more time making games. How it worked at the time is that sites would pay you a small one-time fee, take your game, and put that game on their website. Then I would upsell. If they wanted their logo on it, they'd pay a little more. If they wanted their characters in it, they'd pay more again. If they wanted the source code, then they'd pay a whole bunch more...

I started getting all these sites wanting the Penguin Chats.

Contact Music wanted a music version of Penguin Chat. What was different about those plans is that I said, “I can’t just license it to you, I need you to rent the servers, too.” So they started paying for all the servers. Over all those years, I had more and more people paying the server costs. And it was where I was making the most of my money.

Later ones, like the Crab Chat and Goat Chat... they were built on the Penguin Chat engine. So I decided to update the engine. That’s where we broke all the code apart, where you had separate characters and background and artwork just like Club Penguin. So I could maintain and build more than one chat. That’s also why Chris Hendricks was hired. I hired Chris to help me make custom versions of my games for clients. He did Crab Chat and a couple of other ones like Goat Chat, which was just for a short event.

So I built up a more robust engine to keep selling these Chats. And of course, the Contact Music guys were paying me money, so I kept upgrading the engine every Wednesday night, Fridays, and Saturday mornings. It wasn’t a bad gig. I was making over $60,000 a year from my side job, and RocketSnail just kept growing.

What people don’t get is that there’s actually only a Penguin Chat 2. There is no Penguin Chat 3. There’s Penguin Chat 2.0, and then there’s 2.1. The only difference between those two is that one of them added a database where you could save your character (2.0 couldn’t.)

Penguin Chat 3 is Club Penguin. They are the same. It’s what I’m doing with Box Critters. Done live.

I stopped working on my old server system. The gauntlet was thrown down when Rob from Miniclip wanted to launch Penguin Chat on his servers, but I couldn’t handle the capacity. I couldn’t scale Penguin Chat past 100 users on the entire server. That’s when I started researching SmartFox and Electrotank and two other ones at the time. I was trying all of them.

A new server was built. Then I started building a new client to use the new server with all the new features in it. That’s where the Ninja came in.

The Ninja was entirely there just to test - “Can I clothe or swap items on a character?” (plus I just like ninjas...) That was the only item.

Penguin Chat 3 (aka Club Penguin), it had multiple rooms, items, different colors, storage, etc. It had all the feature set needed for my vision of a big world. My vision was really just a safe place for kids to play, with a chat filter, multiple rooms, mini-games, and getting items. That was it — that was the feature list. The rest of it was just content, like "here’s my idea for a room, here’s my idea for an item." Rooms and items are not really features, they’re just more content.

Remember, my original pitch document for Club Penguin was only 2 pages long. Your vision can be big without a big document.


Here's my challenge to you: create a vision document. Make it less than 2 pages long. For inspiration, check out my original pitch document for Club Penguin.

Then figure out what your first experiment will be for your vision.

I look forward to seeing what you come up with!