Category: Post-mortem

Recently I conducted a post-mortem of The Dungeon Under My House, my unfinished non-violent, first-person dungeon crawler that was meant to be the next Freshly Squeezed Entertainment project.

Now I’ve created a presentation based on that post-mortem, so if you prefer videos, you’re welcome:

—

Want to learn when I release updates to Toytles: Leaf Raking, Toy Factory Fixer, or about future Freshly Squeezed games I am creating? Sign up for the GBGames Curiosities newsletter, and get full color PDFs of the Toytles: Leaf Raking and Toy Factory Fixer Player’s Guides for free!

After 20 months of development, I’ve decided to put on hold further work on my non-violent, first-person role-playing game The Dungeon Under My House, what was supposed to be my second Freshly Squeezed Entertainment project.

It’s a decision that makes me both sad and glad. I really liked this project and I enjoyed working on it and the idea of what it could become when it is finished, but as a solo indie game developer who is working very, very part-time, it is taking me too long, and from my very rough estimates that are probably very optimistic, I’d have at least a year of work left.

That’s a long time, and I need to be more prolific if I want to succeed. Maybe in the future if my capacity increases, I’ll return to it. I hope so. I think it was going to be a neat game.

Still, just because the project isn’t going to be finished and released (for now), it doesn’t mean there aren’t lessons that could be learned from it, which is why I’ve decided to write this post-mortem for it!

I hope one day in the not-too-distant future a second post-mortem of the project links back to this one.

EDIT: I have also created a presentation version of the post-mortem.

Information about the Project

The Dungeon Under My House started out as just an idea that I had at the end of October 2022. Now, ideas are a dime a dozen and all, but this idea wouldn’t let me go. I got excited about making a non-violent Wizardry-like first-person dungeon crawler in which you are a child hanging out with friends in your modern day home, only to discover a secret door in your basement that leads to a multilevel dungeon. It turns out that your family comes from a long line of explorers and adventurers.

So each day after school, you pick a subset of your friends (with unique strengths/weaknesses) to go into the dungeon for adventures. Adventures might take multiple days (you are children with bedtimes, after all, so you can only explore for so long), and you can prepare for adventures by raiding the house for supplies.

Encounters require you to use your skills and tools to distract, befriend, scare away, bribe, help, etc the creatures, monsters, or other adventurers you find in the dungeons.

In my head, The Dungeon Under My House would be a cross between Wizardry and The Goonies, in which your friends and your friendships play a huge role. It would also be a Freshly Squeezed Entertainment project, which as a reminder are games meant to be quickly created and given away for free.

The idea behind giving away free games is that I want my games to have as little friction finding their audience as possible, and if enough demand exists for a particular game, perhaps I will create a “deluxe” version for sale. In other words, rather than guess at what random strangers might want based on trends and fads, I’m trying to find and get faster feedback from the people who would be interested in playing the kinds of games I am creating.

I started working on the project in earnest in November of 2022, with the hope that it would be a six month project, but anticipating that it might take me as long as a year.

Ha. Haha.

I don’t know why I was so optimistic. This was a big project, and I’ve never worked on and completed an RPG before, nor did I sit down to scope out the project to figure out how to fit it all within six to 12 months. But more on that later.

What Went Right

1. Creating a design document gave the project a North Star

   Obviously I’m not a believer in creating 1,000 page tomes as a game design document, but I found that not creating a game design document at all for Toy Factory Fixer resulted in a lot of wasted time.

   So I created a document that is mostly based off of Rosa Carbó-Mascarell’s Game Design template, which you can find here: https://rosacarbo.notion.site/Game-design-template-0132383574dd4c2dbff5d14e3a90761c

   I found it very helpful to identify the core design pillars, the theme and mood I was looking for, and general activities the player would participate in.

   I anticipated that the design might change a bit. For instance, I originally thought the idea of designing your own adventure plans and then executing them was going to be a key part of the game play. I envisioned being at the titular house marking up the map with objectives, then entering the dungeon trying to achieve those objectives. As the project progressed, I could drop this set of features easily as it wasn’t strictly core to what the game needed to be.

   But other things stayed the same. The game was always set in a sleepy modern-day suburb, so I never had to waste any time wondering about medieval or fantasy themes fitting in.

   And the core pillar of non-violence helped me steer away from having any sense of a threat to the physical safety of the characters in the game. In my document, I wrote:

   There should be no attacking or threat of attacks, no abuse, no physical or psychological danger, and no killing.
   – So can there be threats that aren’t related to physical/bodily harm?

   And then I listed out ideas of threats to property (theft, vandalism, etc), relationships (break-ups, loneliness, hurt feelings), and information (secrets, lies, incomplete knowledge).

   The document got updated periodically as I needed to jot down parts of the design I hadn’t anticipated, so it wasn’t like I wrote the design once and forgot about it or pretended that I needed to stick strictly to it. It was a living document that I kept open on my computer any time I was working on the game.

   And having all of my design notes and decisions in one place definitely helped me know what forward progress looked like.

2. Creating cheap prototypes and mock-ups sped up decision-making

   The beginning of a project is always exciting, when anything and everything can be possible, but at some point I need to make decisions that take the project in a certain direction and necessarily cut off other options.

   The design doc helped make some decisions quite easy as mentioned above, but other decisions involved choices that weren’t necessarily right or wrong for the project.

   For instance, knowing that I wanted the game to have a 1st-person view in the dungeon, the question came up: what angle should the camera be at?

   A simple solution is: centered! It would be simple, but it might look boring and uninteresting. Having the imaginary camera higher up and looking down more would show more of the current dungeon tile while also giving the impression that the ceiling is lower or that the player is taller. And having the camera lower would maybe be a more appropriate vantage point for the young children who would be the main characters in the game, but it might prevent the player from being able to see as much.

   At the time, I was planning on using old-style tiles to simulate the three-dimensionality of the dungeon, so creating mock-up art for each angle sounded quite tedious.

   Instead, I used a physical prototype. I took an old shoe box, printed out some dungeon wall image and taped it to the back of the box, then placed some wooden figures in it. Then I took pictures from various angles to get a sense of how it would feel.

   

   

   

   It took me only a few minutes to create, and I learned right away which one I liked best.

   There were also many ways I could show things on the screen, and even if I didn’t know what might be eventually on the screen in the end, I could at least start somewhere. So I created this mockup:

   

   And I found that doing so helped me figure out where I thought some UI elements would work best. The downside is that I realized very quickly that my mock-up didn’t anticipate the on-screen controls that would be necessary for the mobile version, so I tried another mock-up:

   

   The latest actual in-game shot shows that I tried a combination approach in the end:

   

   One of the next features I would have worked on would have been the in-game map on the HUD, and so I was already anticipating needing to rearrange and resize various controls, but I know exactly where I want that map to live.

3. Looking for reference art helped me create a cohesive, “real” world.

   I knew I wanted the game to feel light-hearted and family-friendly. I wanted to have stylized, cartoony art, so photorealistic art wasn’t what I wanted at all, nor did I want a gritty and dark theme.

   I’m primarily a software developer, but I know enough about art that having good reference material helps a lot.

   So I spent time looking into dungeons, sewers, caves, coal mines, and all sorts of things that could act as a dungeon hidden under a sleepy suburb.

   I was disappointed to learn that most real world dungeons were not much more than pits to keep prisoners in, but I was delighted to learn that the Paris has a sewer museum! I even learned that here in Iowa the Wastewater Reclamation Authority has a website that describes how their facilities work AND THAT YOU CAN REQUEST A TOUR!

   And I was very interested to learn of Derinkuyu, Türkiye’s underground city that was rediscovered when someone found a tunnel behind a wall in his home a few decades ago.

   Meanwhile, the other part of the game would take place in the house, so I tried finding art related to various rooms: bathrooms, dining rooms, living rooms. A website about the living rooms of various 90s television sitcoms helped. The living room in my game has a staircase leading up because of course it did: that’s how the rooms on TV look!

   When I found a neat cross-view of a house that had a dollhouse look to it, I knew that I wanted that look for my game.

   

   Besides looking for environment art references, I also spent time looking for character art references. Various cartoons, comics, anime, video games, and even clipart examples of children that I thought seemed promising ended up in a large document. I even had some animals, because I hadn’t quite figured out whether or not all of the characters in the game would be human.

   And with a game featuring people, I needed them to look distinct. Clothing store websites feature all sorts of child models, so I found quite a few neat looking outfits, which helped inform the clothing customization options that I added to the game.

   Now, I could have done whatever I wanted. The existence of games featuring sprawling dungeons despite the lack of a real-world equivalent means that it is already OK not to stick to reality. But having a basis for the “reality” of my game’s environments and characters helps sell the reality better than if I used my untrained artistic abilities to freehand everything with no good rhyme or reason.

   Something non-artists might not know: it’s not cheating to use references to inform your art. Use them!

4. Planning only one thing upfront to work on each week

   This one is kind of a mental health thing for me, but it is a deliberate change from how I ran my previous projects that I think worked out well.

   I ran my project on a weekly cadence. Each Sunday, I would write a blog post about the previous week as a development update, then I would plan the coming week’s sprint.

   In the past, I would kind of mix up my sprint plan with my larger planning, and so I would pick a collection of features to work on, usually features that I think go well together. That is, once all of these are done, I would feel like a significant piece of the game would be done.

   The problem was that sometimes one of those features big quite a lot to work on. Breaking it down into tasks, I’d maybe get a small subset of the tasks done.

   So when I would write my weekly devlog blog post, I would report my progress by saying something like the following from the Freshly Squeezed report of Sprint 42 of Toy Factory Fixer:

   Sprint 42: Training levels

   Planned and Completed:

   • Make sewing worker unique

   Planned and Incomplete:

   • Show tooltips during game based on triggers
   • Create floor training levels/tutorial

   Imagine weeks of saying “I planned to do X, Y, and Z, and I only did X” and you can see why I said it was kind of a mental health thing for me.

   Even if I didn’t mean to say I would get all of these things done in a single week, it still looked like I wasn’t getting things done, and it can be demoralizing.

   And of course, it is hard to predict when things would get done if I kept lying to myself. Not finishing these items means I can’t start working on the next items, and if I do this for weeks in a row, then if I was hoping to get the next item after Z finished by a certain date, well, I can’t even promise I’ll have capacity to start it by then.

   So for this project, I made sure to only plan one feature. Then, I either finish the feature and start unplanned work next before the end of the sprint, or I have only the planned work that is incomplete.

   By and large, it felt better. I no longer felt like I was behind, which is weird because I am my own boss and the feeling came from the accidental promise I was making by listing a bunch of planned work that I wasn’t actually trying to meet. It also helped me avoid deluding myself that a bunch of work would suddenly occur in a single week.

5. Creating devlog videos to attract more potential fans

   I’ve always written about my game development. This blog started in 2005, after all.

   But sharing my weekly devlog blog posts on social media only gets so much traction these days. Adding devlog videos might increase the likelihood that someone finds out about the games I am working on. It’s another opportunity for someone to discover GBGames and TDUMH.

   So I started making a weekly devlog video. Here’s my first one, published on October 3rd of 2023:

   Oh, that’s rough. You don’t have to watch it.

   YouTube said it has 20 views total.

   In fact, for the next few months, my videos would get about 6 to 35 views each. That’s not a lot, but you know, YouTube is very algorithm-heavy, there’s a lot of other videos being made continuously, my own videos don’t have amazing production values, etc.

   So I spent some time looking into best practices. I started creating custom thumbnails instead of using a random frame from the video that YouTube chose, and I even bought a new microphone to replace the ancient one I’ve had since forever (I think I got it when Windows 95 was the latest thing).

   With catchier thumbnails (I refuse to do a “stupid YouTube face” thumbnail no matter how effective) and a less tinny, much sexier voice, my videos started getting a couple hundred views as well as regular comments.

   Creating the videos each and every week was hard work, even with the low production values. I would write a script, create some video clips of various new features or progress I’ve made on the project, then record myself talking through the update. In the end, I would produce a video that is only a few minutes long.

   While I always tracked how long it took to write a blog post, one day I decided to track how long it took me to create a video from start to end, and I realized that it could take me about 1.5 hours.

   Remember how I said that I am working very, very part-time? I don’t have 1.5 hours to create a video each week! It eats into the time to actually make something for me to show off in the video in the first place!

   So I cut back to only creating a video once a month or two. One benefit is that I have more to show in a given video. Here’s the most recent one that I published at the beginning of this month, and you can tell that is definitely smoother, as I have gotten into a groove creating them in a consistent way:

   And another benefit is that while it takes me longer to make each new video, I feel more comfortable with batching all of that time together once in awhile.

   Obviously the videos are not bringing in thousands of customers, but I do think that it has helped me to increase awareness with a few people about what I am doing, much more than my blog and social media posts alone. As a bonus, it gives me practice with using tools such as OBS Studio, and I can also create animated GIFs more easily.

What Went Wrong

1. Not creating a real schedule resulted in a non-stop treadmill of development.

   Definitely my biggest mistake was not creating a schedule for this project.

   When I made Toytles: Leaf Raking, I created a road map. It was an optimistic road map in that I anticipated creating the initial release of the game in three months but the actual release didn’t occur until the eighth month of development.

   But the road map was still helpful to create as it helped me nail down as much of the project as I could upfront. I knew that estimates are likely to be way off, but I don’t think I appreciated knowing as much as I did about my project.

   When I worked on Toy Factory Fixer, I didn’t create a road map. I think I had read something about how road maps are awkward in modern development because it implies being able to believe in accurate estimates and also believing that you could know exactly what a project would look like at the end from the beginning.

   In Agile software development, you expect to be able to adapt to changing market conditions, changing stakeholder demands, and changing realities. The road map is a relic of old-style plans that presume you can nail down requirements up front.

   So I skipped the roadmap, and that project lasted a year. What I remember was that it was supposed to be a one month project, and each month I “hoped” that I would finally be finished with it.

   In hindsight, I don’t know why I ever expected to be finished each month because I do not remember doing any work to figure out the full scope of the project until near the end. Maybe it was because I believed my feature backlog was exhaustive and was always surprised as I added to it?

   For TDUMH, I once again skipped the road map, and I didn’t even pretend my backlog was complete. I basically came up with what I thought was a good first pass, and I fully expected that I would learn about the nature of the project as I worked on it.

   I could be responsive, changing and updating my plan regularly.

   And frankly? This works just fine. I could work on the project at the pace I am at indefinitely, and eventually the game would be finished.

   At the same time, I did nothing to try to figure out how to fit this project into a six month period, which is what I wanted it to be. But merely wanting Toy Factory Fixer to be a one month project wasn’t enough, either.

   So I had a conflict that I wasn’t addressing: my business expected me to release games more frequently, and my management of the project had me working as if I could take forever to release the project if I wanted to.

   At some point in the last few months I realized that I needed to figure out how much work was left on the project, and with my very rough estimates, I’ve decided what was left was way too long.

   And I think if I knew it upfront, I might have tried to make a smaller version of the game.

2. Not prioritizing the work well wasted my time.

   Even early on I knew that by making the game non-violent, I would need to replace all of the compelling combat-related mechanics such as health, weapons, armor, attacks, etc with something equally or more compelling. My goal was to emphasize conversation and knowledge-acquisition.

   The thing is, I didn’t know what it would look like.

   A branching dialogue tree wasn’t going to cut it. I didn’t want to make a rigid story or a visual novel. I wanted something more dynamic and flexible, that required the player to make more interesting decisions than choosing between a handful of pre-written options.

   In short, I wanted the conversations to BE the main game play.

   And yet, instead of spending time prototyping conversation systems and UIs and researching existing games that do interesting things with conversation mechanics, I worked on character customization.

   

   

   I justified it to myself. I really, really wanted the player to feel like they could see their friends and family, or maybe a fantasy version of who they wish their friends and family were, in the game. It seemed important at the time.

   And I worked on creating the first person dungeon view, first by trying to create tiles, and then creating a raycasting renderer, and then profiling it and debugging it to see if I can get it to be performant each time I made a change that slowed it down.

   In hindsight, I should have realized that the first-person aspect of the game, as nice a thrown-back as it was to the kinds of old-school RPGs I wanted to remind the player of, was not necessary or fundamental. A top-down overhead view would have been faster and easier to implement. Character customization turned out to be tedious work, which I didn’t anticipate when I started, and perhaps for this project I should have created concrete, unchangeable characters if only because it would have been faster.

   And the main reason I wish I had done those parts of the game so much faster is so I could focus the bulk of my time on the more interesting game play that I was trying to create with conversations and personalities and knowledge acquisition. Instead, after 20 months, the systems in place feel clunky and limited, and I knew I still needed to expect to spend considerable time on experimenting and figuring out this core part of the game.

   It would be one thing if I spent only a little time on non-core things. Spending 20 months on a project might have been more palatable if I had spent a large amount of that time coming up with new game play, but as it is, I feel like I could have used my time better.

3. Not quitting earlier meant missing opportunities to publish more frequently.

   This one is kind of related to the previous points. If my goal is to release a game in six months, I should hear alarm bells in my head when I hit the six month point and not have a game anywhere near finished.

   In fact, those alarm bells should have been heard long before six months passed, because I should have an idea how much of the game was finished versus how much is left to do and be able to figure out how confident I felt about getting the remainder of the work completed in the time left.

   Instead, as I said before, I could work indefinitely on my project, inspecting and adapting as I go, and given my limited capacity, I wasn’t investing time to inspect the health of the project as a whole. I did not treat the six month deadline as a real deadline so much as a soft suggestion, which didn’t help give me a sense of urgency or a reason to worry.

   Toy Factory Fixer’s initial release was in 2021. I didn’t start TDUMH until 2022 with an expectation that I would release it in 2023. Even if I had succeeded, that is already a long time between game releases, at least for my business goals. Taking longer than six months means even more delays for accomplishing those goals.

   Maybe if I would have put the project on hold or canceled it in 2023, I could have used the time since then to figure out a smaller scoped game. Maybe by now I would have released at least one, perhaps two or even three more games.

   Better late than never, but I should be more protective of my time. I am getting too old too fast, so it isn’t just my business goals I’m worried about.

4. Waiting so long to use freely available UI elements made my game look amateurish

   When I started working on the project, I slapped art together to get something on the screen as quickly as possible.

   The main menu’s buttons, for instance, looked like this:

   

   Just colored rectangles. They are functional and easy to create.

   I also tried to add a little bit of detail to them to make them better, but it was still very basic:

   

   Between taking screenshots and making videos, I had a lot of footage of the game with these placeholder UI elements.

   Which is unfortunate, because the entire point of publishing footage of the game was to get people interested, and I was basically saying, “Here’s this ugly amateur thing for you to not waste your time with.”

   It was only recently that I decided to use buttons from the free Kenney UI Pack, and after adjusting some font colors and sizes, the buttons on the screen looked way better.

   

   What’s weird is that I had no problem using existing UI art. I already used icons from Game-icons.net in various places, and I already knew about free assets from Kenney.

   So I’m not sure what took me so long to get around to it other than a sense that other things were more important to work on.

Lessons Learned

1. I should take advantage of good art assets earlier.

   I should definitely start my next project with great looking icons and buttons, especially if they are going to be displayed on the screen a lot and they’ll be some of the first things seen in screenshots and videos.

   While I still like providing my own art in general, and I don’t think I should fill my game entirely with free sprites and icons, using fantastic, ready-made resources such the Kenney UI Pack makes the game look way better than I could on my own and in a much faster time period. I should really need to justify not using them rather than the other way around.

   And if game UI icons and assets make sense to bring in early, what about sound effects and music? I usually leave those until late in the project, and perhaps it makes sense to have placeholder audio at least.

2. If I want to release more frequently, then I need to make a schedule.

   Open ended projects are a luxury I can’t afford.

   As I said above, it isn’t as if I couldn’t keep working on my project the way I have been. I know each week I make a little progress, I can inspect and adapt as I go, and one day I’ll work on the last bit of progress and call it done.

   But my business strategy ostensibly requires me to release my games much, much more quickly. Until I can increase my capacity to work on my projects, my projects will need to be scoped much, much smaller so that I can get them done in a reasonable time.

   I still expect that I’ll need to inspect and adapt, that I can’t possibly anticipate everything, but I’ll have a better idea of what needs to go into a project plan for a small game than a big one.

   So simultaneously knowing as much about a project as I can will help me create a realistic schedule, and a schedule with deadlines should help me limit how much a project can be.

3. With my limited capacity, I need to ruthlessly prioritize the work.

   I like to be purposeful, which means that each project I work on isn’t just a random collection of cool-sounding features. When I create a game, I like to identify some core part of it that is the reason it needs to exist now instead of some other project.

   For TDUMH, I wanted to make a non-violent dungeon crawler, and what I should have done was figure out exactly what I needed to focus on to make it work. I already mentioned above that not having combat in the game isn’t enough, as I needed some compelling game play to replace it.

   If I were to start again today, I think I would realize that I need to prioritize prototypes related to conversation and knowledge-acquisition mechanics as early as I could. Maybe I could have decided then that a party-based RPG would be too big of a project, but what if I applied the same focus on conversation and knowledge-acquisition to a game that consisted of nothing but a small room and two or three people, and maybe a few objects?

   I might have been able to put together a game in a much shorter period of time, and I could always use it as a building block for my next project.

   Since I instead worked on things that shouldn’t have been prioritized, I still need to figure out compelling game play because right now I just have customizable characters and a dungeon to traverse but not a whole lot else.

   Even if I had more capacity, I still would want to prioritize my time much better, if only to make sure that I am releasing games sooner than later.

   For the early stages of my next project, I should spend time trying to identify the dependencies between different features and systems. I did a light list for TDUMH, in fact, but it wasn’t something I paid much attention to, and well, here I am wishing I had done more.

4. I should continue to seek out opportunities to promote my work.

   People have asked me why I still try to sell Toytles: Leaf Raking after all these years. Why don’t I just drop it and forget it and move on?

   And my main objection to that reasoning is that it isn’t as if lots of people saw the game and thought, “Nah.”

   So I try to promote it and sell it because people by and large haven’t had a chance to reject it yet.

   Like most indies, obscurity is my main obstacle to success.

   For TDUMH, I found that sharing screenshots, posting weekly progress reports, and publishing videos means I am giving more potential fans a chance to find out that my games exist.

   Some of those people become fans who start to ask about details, but I don’t know where the next fan will come from. Maybe they’ll find my blog post because someone shared it. Maybe they will be perusing YouTube and the algorithms will manage to place my video in their feed at the right time. Maybe they’ll see one of my screenshots posted on social media.

   Most of my time is spent on game development, but it doesn’t take much time at all to capture in-development screenshots that I can then easily use in blog posts, videos, and posts on social media. There’s no excuse to not tell people what I am working on more frequently and in more ways.

   Plus, making games with an audience is a bit more motivating than listening to crickets.

5. It’s easy for me to keep my nose to the grindstone.

   Each week, I plan the work for the next sprint, and then each day I succeed more often than not to make steady progress.

   On average last year I worked about 7.5 hours per week, and this year I’ve been trying to hit 7 hours per week as well.

   As much as I would love to make more time for development, given my life priorities and obligations, this pace is doable for me. I rarely find myself exhausted or burnt out by it.

   Maybe that sounds silly, but the hours add up, and each hour I work on my game is an hour I’m not spending time with my family. Until I quit my day job and free up a significant source of hours, this pace is my reality.

   That said, I have shown that if I set myself in a direction, I can continue plodding along until I’m finished. I stick with things, and I can keep focused on the same task. These are good things!

   The challenge is to make sure I check in on myself to ensure that direction is still a good one and course-correct as needed.

   I’m used to focusing on the current sprint’s tasks, then planning the next sprint. I’ll need to start looking at any given sprint as part of a whole, asking myself how far along do I think I am and how much more is there to do. And I’ll need to be honest. If I already have X tasks to do in Y weeks, and I discover an X+1th task, it very likely means I’m working for more than Y weeks if I think it needs to be done. Which means I’ll need to be sure that it is something that “needs” to be done and not just a really appealing nice-to-have.

   These frequent project health check-ins should help me lift my nose up from the grindstone periodically, mainly to make sure I’m using the right grindstone.

Conclusion

As excited as I was about The Dungeon Under My House, it ended up being too big of a project for me at this time.

I’m not new to game development. I’ve been creating games for many years. Still, this project reminded me of my limits, and if I am honest, it humbled me a bit.

I do believe that if I kept going the way I had been going, the game would eventually be finished, but it would take me way too long to get there. At least, too long for my overall business goals.

The opportunity cost is too great when I could be releasing smaller games and building an audience during that same time. So as painful as it might be to have to part ways with this project for now (and potentially forever? I hope not), it’s not a hard decision to make.

After 20 months of development, even though I won’t be shipping this project, I am glad I can take stock and glean some insights from it.

I know this analysis will help me greatly with my next project. If I could sum up the lessons, it would be that it isn’t enough to keep moving forward because I also need to ensure the destination isn’t too far away.

I hope this post-mortem helps you, too. Let me know if it did!

Stay curious!

—

Want to learn when I release updates to Toytles: Leaf Raking, Toy Factory Fixer, or about future Freshly Squeezed games I am creating? Sign up for the GBGames Curiosities newsletter, and get full color PDFs of the Toytles: Leaf Raking and Toy Factory Fixer Player’s Guides for free!

Now that my first Freshly Squeezed game has been out for a few weeks, I thought it would be a good time to look back and see what lessons could be learned for future projects.

Toy Factory Fixer, a turn-based toy factory management game in which you hire and manage workers to ship toys, was a project I started in December of 2020, the first of my Freshly Squeezed Entertainment line of games.

As I said in the my first progress report post for this project, Freshly Squeezed games were meant to be quickly created within a month and given away for free.

The idea behind giving away free games is that I want my games to have as little friction finding their audience as possible, and if enough demand exists for a particular game, perhaps I will create a “deluxe” version for sale. In other words, rather than guess at what random strangers might want based on trends and fads, I’m trying to find and get faster feedback from the people who would be interested in playing the kinds of games I am creating.

So I’ll start by saying that Toy Factory Fixer was meant to be a one month project, with a target deadline of December 25th, 2020.

Ha.

I ended up publishing the game in the Apple App Store and on Google Play on December 14th, 2021, over a year after I started.

For the mathematically inclined of you, you probably noticed that it definitely took much longer than a month.

But this is a post-mortem! Let’s talk about what went right, what went wrong, and what lessons can be learned from the experience.

What Went Right

1. Paper prototypes helped me narrow in on the concept and game play early

   At the end of November, right after my latest published update of Toytles: Leaf Raking, I already came up with the general idea for Toy Factory Fixer. So right away, I had the concept on paper. Originally, it was a Christmas-themed game, with the idea being that the in-game shipment deadline was Christmas Eve and Santa needs those toys.

   But how would the game play? It could be anything.

   When you have the proverbial blank slate, there is a lot of potential, with many directions to go in, and it can be quite exciting.

   But you can’t stay in this amazing realm of imagination where anything is possible. At some point, you have to make decisions, which means cutting off some options as you choose others.

   Within days, I thought about a tower defense-style game, with a conveyor belt and workers (and machines, but those didn’t make it into the game).

   

   
   Now, I could have started coding and creating digital art, but that is an expensive thing to do upfront, especially in terms of game design. It’s faster and cheaper to use paper and markers.

   And doing so helped me to figure out the answer to some early design questions related to what the player would do within the game and how the various parts would work together.

   Should workers be able to do things automatically or require the player’s micromanagement?
   What obstacles would be in the player’s path?
   Should there be a turn limit?
   How does a worker get the parts to put together a Good Toy?
   How do toys on a belt interact with other toys?

   While not everything explored through this phase made it into the final game, it did help inform implementation details. For instance, I knew that I had to figure out what it meant to have multiple toys share the same location on the conveyor belt early on, and when it came time to actually write the code, I had already thought of a few solutions thanks to the quick paper prototypes I was able to come up with and ended up picking one that I thought made the game better than I had originally anticipated.

   I love the benefits of paper prototyping, and I wish I had done more of it, but for the amount I did, it definitely had a huge impact in terms of speeding up development and design work.

2. Light-weight project planning helped me stay on task for the long-haul

   I’m a big fan of Agile software development, and so I am habitually used to focusing on value-delivery, working in short sprints, breaking down work into bite-sized pieces of value, baking in quality, and keeping the stakeholders/players in mind.

   I know some indie game developers work with a simple list of tasks, or maybe a list of must-haves and a list of want-to-haves, and it works for them.

   And maybe my spreadsheet is just a glorified collection of such lists.

   But from the beginning of this project, I created a backlog of features and tasks, then for a given week, I took a subset of the backlog and tried to get it finished. Then I repeated until I was done.

   

   And that’s oversimplifying it, but that was the secret that kept me going for a year of very, very part-time development outside of the day job, the family time, and other obligations.

   I basically had a small set of things to do at any given time, and at the end of the week, I assessed how it went (usually by writing a sprint report blog post), took a few minutes to plan the next week’s sprint, and repeated.

   The slightly longer explanation is that I estimated how much work any given set of tasks was going to be when it is in the backlog using t-shirt sizes (Small, Medium, Large), and when I planned the sprint, I broke down those items into more detailed tasks so I knew what needed to actually be done. Then whenever I was doing game development, I’d look at the spreadsheet, pick a task, and focus on it until it was done.

   And since I usually didn’t get more than hour of game development at a time, I found great success with small and specific tasks rather than high-level, vague descriptions that leave a lot up for interpretation even though I’m the one who wrote it in the first place.

   I had a daily habit of putting in game development, and my project plan helped keep each day’s game development session connected to the previous one. Even if I went a few days without doing any game development, keeping my project plan front and center meant that it was easy to get back into it.

   I am not saying that everyone should do what I did. I’m merely saying this is what I did, and it worked well enough to keep me focused until the project was finished.

3. Light-weight project planning helped me adapt to changes discovered mid-development

   I’m not going to pretend that games are special in terms of the need to deal with changing requirements, since so many software (and non-software) products also deal with it.

   But quite frequently I would add a feature, playtest it, take a step back, and realize that the player is going to need more feedback to know what is going on.

   That is, mechanically something might be in the game, but if the player gets no indication, it can be easy to miss.

   So sometimes I would discover partway through the work that there was more work to do.

   And no, I wouldn’t describe this as feature creep.

   Let’s take the time I added the ability to earn money when you ship a Good Toy.

   When you ship a Bad Toy, nothing happens (and maybe I should have added some negative feedback there like I’ve thought about doing for months, but ah, well). When you ship a Good Toy, it looks like nothing happens even though you earned money. If you were paying attention, you would see the money value is different.

   But as a player, you aren’t paying attention to that part! You’re trying to pay attention to everything else going on in the game!

   So I did a few things. One, I changed the value of everything by an order of magnitude. Workers used to cost 10 Money, and then they were 100 Money. Why?

   So that way I could more easily make it obvious that the player’s Money interpolated as it changed from one value to another. There is something satisfying about seeing numbers change over time instead of instantly changing at once.

   Of course, once I did so, I realized that it felt too slow when the numbers change a lot, so I even tweaked it to interpolate faster with larger changes in value.

   And I had the money the player earned fly from the shipping chute towards the Money balance at the top of the screen.

   These were small bits of polish that really helped sell what was happening to the player, and I didn’t anticipate them when I started.

   But I had a plan! Didn’t these changes get in the way of the plan? Didn’t they slow me down?

   No, that’s another thing about Agile that I like. You learn things as you work on a project, and that learning should inform the work.

   So my light-weight planning tool was easy to modify on the fly. I could be days into a given sprint, and I might realize that a given feature had a few extra unidentified tasks in it before I could say it is truly complete. I have the option to add those tasks to the sprint or create a new feature in the backlog if I decide I’ll work on it in a future sprint.

   

   It’s mere seconds of work to update the plan, and I can get back to work.

   Now, I could argue that the downside is that I allowed for scope to balloon for any given task, but I’m alone in this project. I didn’t need to argue or justify or push back against someone else’s requests. I just had to make a judgment call on whether or not I should add work. And while I often did, which led to the year-long project, I also had quite the backlog of new ideas that never made it into this initial release because I was trying to keep scope down.

   I think it is easy to let the light-weight part mean that I didn’t put on my producer’s hat often enough, but it did let me make changes on the fly fairly easily while keeping on top of it all, and I think the game is much better for it.

4. Having testers meant finding major issues and getting feedback right away

   At one point, I put out a call for testers, and two people replied.

   And they made a huge impact, both in terms of game play and in terms of basic functionality.

   When you are targeting mobile platforms, you have to deal with regular cycles of sometimes seemingly-arbitrary changes to those platforms. They aren’t changes I care about other than the fact that they might break how my games work on those platforms.

   So imagine my surprise when I was regularly testing my game on my Android device only to find that a tester couldn’t get past the main menu on his Android device. He said it looked like the game was frozen. And after adding some debug code, we verified the game was not frozen but was, in fact, working quite well except for the tiny detail of touch inputs not getting processed.

   WHAT?!

   Weeks later, I finally figured out that the major culprit was that Android 10 somehow handled touch input differently from my much older Android device, or at least differently enough that libSDL2 wasn’t really behaving the same way. My workaround was to exit out of a message pump any time I process any input, which allowed my code to detect when a player pressed and released a button. It otherwise wouldn’t notice because the press and release happened at the same time and so the before and after input state would look the same.

   Separately, the testers frequently told me what was confusing or what worked well, and a lot of their feedback made it into the game in one form or another (further adding to the scope, but I think for the better). Some of their feedback is still in my backlog as things I want to add to the game but that didn’t make it into this initial release.

   Also, it is quite gratifying to have testers tell you that the game is fun. When you’ve been doing your own testing of an in-development level over and over and over again, falling asleep at the computer (and worrying that the game is boring enough to make players also fall asleep?!), it is so great to hear that someone else is enjoying what you’re making.

   I probably could have been in more frequent contact with the testers, but I was also trying to be mindful of the fact that they were volunteering and had their own obligations and time constraints. As things stand, though, I am really thankful for their efforts.

5. Little bits of polish went a long way towards making the game look interesting

   I already talked about the money interpolation thing above, but at some point I realized that the game has a lot going on but looked like nothing was going on.

   Still frames and screenshots are fine, but animated GIFs and video tell quite the story. Except they were boring-looking stories.

   And at some point, I had accepted that this project was taking a lot longer than a month, and I wanted to give it the best chance to find an audience, so I thought it was important to add some “juiciness” to the game.

   I added animations, such as having the toys lean back as they are moved forward on the conveyor belt, or having the workers move their arms when they are finished separating a Bad Toy, or having the worker’s eyes look at a nearby Bad Toy.

   Here’s what it looked like in January:

   

   And here’s what it looked like in September:

   

   My favorite animation is a toy arcing into the air and back down to land on the conveyor belt with a thud. There’s a small dust ring that appears, and the squash/stretch of the toy as it moves it subtle yet works really, really well. And the sound effects really sell it.

   Most of this work didn’t feel like it took too much time to implement, and some of the effects were purely aesthetic, but a lot helped communicate to the player what was happening. Seeing floating numbers appear as toy parts were added or taken away from the inventory worked way better to let the player know that the quantities changed.

   And it made the animated GIFs look way, way more entertaining.

What Went Wrong

1. I stressed myself out with my arbitrary one month deadline…for months

   Freshly Squeezed games were supposed to be quick. I wanted to create one-month prototypes. But I also wanted them to be polished and finished enough that someone could say “Yeah, that was a complete experience…and I want more.”

   And every week after the one-month mark was another week that my game was overdue and in which I failed to do the “simple” task of creating and releasing a game.

   I especially felt bad after three months had passed. That’s 90 days! I should have had a Minimum Viable Product by then!

   No one was promised this game in one month. I had no customers or stakeholders who paid for it. And yet, I felt pressure to get it done.

   I mean, the longer this one game took to make, the fewer games I could make for the remainder of the year. And without even one published game, I couldn’t find out if the Freshly Squeezed “give away polished prototype, hope people subscribe to learn about future games” idea was likely to work. The entire product development strategy felt like it was in jeopardy! I needed to ship!

   Now, it turned out that I had underestimated the amount of work it would take to make something I would feel proud of releasing. A key part of Freshly Squeezed games was that they were not going to be shoddily-made, and it turns out that quality takes time.

   And as I wrote throughout the year, game development time was not something I had in abundance.

   The game ended up taking as long as it did. I don’t like saying “It’ll be done when it is done” because, you know, I did want people to play the game before I turned 80-years-old. I don’t have the luxury of working on games forever.

   In a well-run Agile project, if there is an arbitrary deadline, you work on the most valuable thing and have something to ship as soon as possible, and you build on it so you always have something functioning and ready to ship. And you expect that a lot of items in your backlog won’t make it into the final product by the deadline, but you know you got the key things in, AND that there is no other optimal way to do it that wouldn’t require inhumane working hours.

   I wanted to have both that arbitrary deadline AND the ability to keep expanding the scope of the work, and it doesn’t work that way. I probably should have realized early on that, no, this project was going to take longer than a month, then figure out what my minimum release criteria was as early as possible.

   Of course, the version that did get released feels like it is already fairly small in scope. Anything smaller, anything I would have released earlier feels like it would have been a much lesser game for it.

   Perhaps going forward, I limit myself to a maximum of 90 days for any one project, and I hold myself honestly to that deadline. It would require me to spend more of my time upfront exploring the basic game play, and only then can I polish it up.

   But it definitely requires me to rethink my product development strategy going forward.

2. No real design document meant a lot of wasted energy wondering what should and shouldn’t be in the game

   I’ve never spent time creating massive 300+ page design documents, of course.

   But in the past I have found that even a 48-hour game jam went more smoothly when I made use of a living design document.

   Unfortunately, I didn’t use one for this project.

   Instead, I had random notes on sheets of paper, random notes in my development notebook, and the random ideas and features I threw into my project backlog.

   My project plan is in a spreadsheet, and one tab had a vision for the project, but it was very much focused on the purpose it had as a Freshly Squeezed game, as well as some guiding principles such as a desire to make the game family-friendly.

   So, when it came to making decisions about what to put into the game, I didn’t have much of a North Star to guide me.

   I had a lot of ideas, but they were not together in one place, which made it hard to think about them cohesively.

   I used different individual tools separately, and to great effect. For instance, to get the look of the workers in the game, I found images of Christmas elves online. I found a lot of reference art, which helped inform the creation of the sprites for the elves (hey, that’s almost a pun…).

   I doodled and sketched out ideas, and some ideas made it into the game partly because that prework helped me figure out how to make it real.

   But it definitely felt like I was randomly deciding what features were key and which were optional, and it would have been nice to codify that decision even if it was after the fact. That way, when I was revisiting an area of the game months after I last touched it, I wouldn’t have to remember why I chose not to do something or why I went in a certain direction.

3. I spent time on polish before I spent time on nailing the game play down

   I think reading and watching videos about adding “juiciness” to games got me thinking that my game looked lifeless, and I needed to address it.

   Some of the work was quick, such as making the toys animate as they move down the conveyor belt.

   I’m very proud of the polish I put into the game, but added up, all of this polish and “juiciness” took a substantial amount of time.

   Which is fine, except that even after a lot of polish work, I kept adding major game-changing features since the game felt incomplete without them.

   Now, I’m not sure if it is necessarily wrong to polish existing features before moving on to adding another feature. After all, if I intend to ship any day, having a prototype that looks like it was hacked together is less appealing than a prototype that looks like a finished product.

   But I think the fact that I didn’t have a clear idea of what needed to be in the finished product until almost a year later made it easy to keep extending the work by adding animations, particle effects, and more.

   They were always justified, of course. The animation and sound effects act as powerful and immediate feedback for the player. They aren’t just eye-candy. They actually help the player significantly.

   But I think being more deliberate about fleshing out the core game mechanics first would mean it is cheaper to try exploring a few ideas before settling on them.

   Adding polish basically means I’m baking the features into the game, and they become harder to change.

   

   I can always iteratively add polish, revisiting things instead of trying to get it right the first try.

   My first pass at income generated from shipping Good Toys was functional and non-obvious to the player when it happened. My second pass added an animated amount that flew towards the player’s money balance at the top right, but it was hard to read as it went too fast and if it went slower it would be distracting. My third pass was to change it to floating numbers that disappeared after a short period of time.

   I think purposefully identifying potential iterative passes for polish might help balance between focusing exclusively on functionality and spending too much time on making things look and feel good before that functionality is all in.

4. Being very, very part-time meant not a lot of time dedicated to experimenting with game design

   Rereading all of my progress report blog posts, one common theme was that I wish I could dedicate more time to game development.

   I lamented the lack of hours. A lot.

   So, I apologize if you got tired of reading it.

   But besides forward progress being slower than I would have liked, a lack of time meant I had to limit the amount of exploration of the game design space I could do.

   Game development is iterative and incremental in nature. You make something, then you build on it. And sometimes what you build is only obviously wrong after the fact, when you can experience it in motion, so you throw it away and try something else.

   If you could experiment with game play once a day, you could learn a lot about the game you are making, and you can find all of the dead ends right away, and you can find the more entertaining experience sooner. Any one option is cheap to try.

   But if your experiments only come once every few months? Now your changes are expensive to make. A choice to try something is more likely to become permanent simply because there isn’t time to replace it with something better.

   For example, at one point early on, I was worried that it was tedious to direct individual workers to craft toys. And frankly, I still am. But how much experimentation can I do with such a core part of the game, especially when I thought I was almost done for most of the lifetime of this project?

   So that mechanic stayed as it was.

   I did create the Go/Stop buttons partly as a result of tester feedback, so there were some things I thought of that eventually did get experimented with, but the point remains.

   If I wanted to release a playable game in a relatively timely manner, any game design experiments would have to be focused and quick. Maybe instead of fully exploring a design aspect, I try two proposed ideas and pick the best one. Or I try one and see if it works well enough, then move on.

   Much better would be to focus full-time on game development, but until my circumstances change to make that possible, I’ll need to figure out an approach to game design that let’s me maximize design exploration in as little time as possible.

5. I didn’t pay attention to release criteria until near the end

   I laugh today at how optimistic I was about how soon I expected the game to be released. First, it was by Christmas. Then it was by the end of the 2nd month. Then the 3rd month. My birthday in July? Before a year of development was done. At least by the end of a year of development!

   I didn’t have game play for months, and it was almost six months into the project before I felt comfortable handing it off to testers to try. Why did I think I was almost done when I didn’t have any levels to play in?

   There is a difference between a wish (“I’d like to release this project within 30 days”) and a reality-based plan to accomplish it.

   I somehow fooled myself into believing that I was somehow going to finish the game just because there was a day I said I was going to be done. Even if that day changed every time I missed yet another deadline.

   Much of development early on was infrastructure and scaffolding. I created kind of a template for a menu system that all future Freshly Squeezed games will use. But that didn’t help me get a game done quickly.

   Later, I was solving platform-specific issues with newer Android OS versions, and then I was dealing with iOS build issues. They were important, but they delayed work on the game itself quite a bit.

   And I kept adding scope to the game. Sometimes it felt legitimate, like adding animations to help communicate what was happening in the game to the player, but sometimes it was plain old feature creep.

   I tried to separate my backlog between must-haves and would-like-to-haves, but the must-haves list kept growing in scope.

   I had 25 items in that must-have backlog when I started, and ended with about 95, with another 40 or so ideas that I decided not to include.

   At some point in late October, I decided that what was possible to do in the game good enough, and I created a hard list of things I needed to do to finish v1.0 of the game. The biggest “real” task was deciding on and creating the number of levels I would ship with, but the rest included some key areas of polish, options to toggle sound and music, and creating an in-game help.

   Having that release criteria helped me focus on finishing in the final months. What would it have been like if I had decided earlier on the nature of levels in the game? Maybe that design document above would have helped inform other features.

   But the point is that I spent quite a long time thinking I was almost done, only to identify a significant amount of work late that I could easily have seen if I was serious about releasing a publishable game as early as I would have wished for.

   I’m not saying that a plan would have been perfect, that I would have been able to predict with 100% accuracy how long the project would have taken. But I am saying that I had a plan, and I could have paid attention to it more to give me clues.

What I Learned

1. Prioritizing the work is important enough to make time for

   Each week, I took a few minutes to plan the next week’s work. Sometimes it took more than a few minutes.

   With limited capacity in terms of hours, one might think that every minute of development counts, and I should spend less time on project planning. Simply work on game development until the project is done.

   But having limited time meant it was even more important to invest some of my time in planning. I always had a direction to go in, I always knew what I was working on next, even if that work was “figure out how to solve this particular problem.”

   My weekly planning cadence produced a lot of benefit.

   That said, I think I could stand to wear my producer hat more often. I overscoped a lot of my sprints, which meant that while I was making steady progress, I was also using any given sprint backlog less as a prioritized list of the most important tasks and more like a sub-backlog.

   Now, I’m generally fine with the work being fluid. If I learn things about a given chunk of work, if I come across defects, I throw them into mid-sprint backlog. Fine.

   But if I only get about two major features done in a given sprint, what sense did it make to add five of them to my sprint backlog? Again, wishful thinking served me poorly.

   What it did was give me permission to say “Well, any work unfinished will just carry over to the next sprint.” Which gave me permission to start work on some items without thinking through them, since, hey, I was just going to work until I thought it was done.

   Which meant I sometimes spent too much time on a given item of work. Or I allowed myself to work on a task that probably shouldn’t have been a higher priority, such as the amount of time I spent trying to create a background image for the game despite the fact that I am not an artist by trade and that all the time and effort I would put in would just result in a slightly less bad piece of art.

   

   And if I really wanted to make a game quickly, even if I was going to be fine with it taking longer than a month, why didn’t I take advantage of the t-shirt sizing I did to create a rough roadmap or something similar that would help me understand just how much time might be needed to work on the game? I might still underestimate it, but I would probably be able to recognize that I wasn’t “almost done” for so many months if I could see that even the initial work I could identify upfront was going to take more than four weeks of work.

2. A little polish and juiciness goes a long way

   I think it is safe to say that Toy Factory Fixer is my most satisfying game to play out of all of the games I’ve ever created.

   Most of the games I have worked on in the past were 48-hour game jam entries, and my efforts were focused on putting together a functioning game quickly and not necessarily on how it looks. In fact, many of those games were silent, with at most some sound effects thrown in at the last minute.

   Many years ago, I read a quote from a game developer saying that the lion’s share of game development is in the interface, and at the time I thought, “Ok, yeah, so there’s code and there’s art. Duh.”

   But then around that time while discussing how Model-View-Controller works, I had a realization that the interface to the underlying game system can itself be very complex. For example, when you click on a unit in a real-time strategy game, there is a lot of visual feedback. Sometimes the unit gets a little circle around it to make it clear which one you are currently looking at. The menu might update with the unit’s picture. A sound effect might play, like a worker’s bark. And when you click on a button to issue a command, the buttons will change to indicate which one you selected.

   And all of this isn’t strictly necessary in terms of the bare minimum to issue that command to that unit. That is, the underlying game simulation probably has no idea any of these clicks are happening.

   But the player’s experience? It is greatly enhanced. It is a lot less confusing, and it can often be a key source of the entertainment of the game on its own.

   When I added transitions to the screens of Toytles: Leaf Raking, I found that I loved how something as simple as fading in and out made the game feel better to play. The game is mostly static still, but it’s better than it was, even if only slightly.

   With Toy Factory Fixer, the toys went from merely traveling down the conveyor belt to feeling like they had actual weight and momentum. Functionally, they still moved one tile at a time, but visually and audibly, they arced into the air, landed with a heavy thump, then got pulled down the conveyor belt each time it moved. The toy parts flipped in the air as they soared towards the inventory.

   And when the toys got so much work done, I felt I needed to enhance the workers. They soon talked, blinked, looked at nearby toys, and waved their arms. You got a sense of their personality.

   And floaty numbers made it easier to see when you earned money and when you gained or used items in your inventory.

   Juiciness wasn’t just animations and particle effects, though. I had a list I created in July of various opportunities to add enhancements to the game, and some were more useful than others in terms of player feedback, and less than half of the ideas made it into the game.

   Some ideas were silly and meant just for fun. I always wanted to give the workers random names and irrelevant stats, such as “Laziness: 7” or “Disgruntled: 2”. Apophenia is a powerful thing to take advantage of, but it will have to come in a potentially future update.

   Other ideas that didn’t make it in might have helped communicate the state of workers, such as animations that indicate what they were doing, or communicate the lore of the game, such as letting the player see a description of individual toys.

   But even with the few enhancements I did implement, the game came alive, felt great to play, and seems to delight players.

   If I learned anything, it’s that I want to try to focus on adding personality to the game earlier rather than as an afterthought.

3. The best of intentions isn’t a substitute for an actual plan

   Between assigning myself too much scope in any given sprint and allowing the game’s overall scope to grow throughout the project’s lifetime, I kept fooling myself into thinking I could do a lot more than I was demonstrating.

   I think feeling like I was supposed to be finished yesterday meant that I was mindful of keeping scope down, but I always felt like I was trying to figure out what was the actual minimum amount of scope I could feel good about.

   After one year, the game has 3 workers, 2 toys types, 2 toy sizes, 4 level layouts with 2 shifts each, and the option to play with a hard deadline or not. And a bunch of polish.

   What would this project have looked like if I was able to actually focus on game play for a month? Maybe one worker type? One level? Would the general design be quite a bit different to make up for the lack of variety of things in the game? And how little juiciness would it have? And is that fine for a short prototype?

   I think wondering about the above is fine, and in fact, it is probably what I should have been doing if I was serious about getting something out quickly. What wasn’t fine was pretending I was trying to make something quick while also not actually operating like I was.

   My original plan was to create polished prototypes, with a guess that a game might take up as much time as a 48-hour game jam entry might. After all, I have shown myself capable of creating a decent enough game in that time, and having that time spread over a month should be fine.

   But none of my games look and feel as good as Toy Factory Fixer did, which required an order of magnitude more hours to develop.

   In hindsight, I think with my current development capacity, perhaps a one month project is too short to explore the design space and develop something I’d be proud of.

   But I definitely can’t pretend my next project will only take a month if I just want it badly enough. I have to actually do the work of managing scope and/or my schedule.

   A project that I allow myself to work on until it is done is going to operate very differently from a project that that has a real deadline.

4. Slow and steady can still cross the finish line eventually

   Looking at my records, I put in some game development time every week until the game was released.

   My best week was in May at 11.25 hours, and my least productive week was in December right before release at 1.25 hours, which makes sense as I was spending time on writing blog posts and email newsletters. My next least productive weeks were 2 hours each. Otherwise, I generally averaged 6.1 hours per week, which is a bit more than my 5 hours per week goal.

   This project took 299 hours in 2021 and another 32 hours at the end of 2020, for a total of 331 hours.

   It is about the equivalent of a little over two months of full-time effort, although when I was a full-time indie game developer I found that the most I could put in on any given day was about 5 hours, so let’s say it likely would have taken me between two months and four months if I was focused full-time.

   But I wasn’t full-time, so it took me a lot longer, which required a sustained effort over 13 months.

   Don’t get me wrong. I much preferred those days in which I was able to put in multiple hours, especially when solving a big problem. All things being equal, five hours in one day allows me to focus and work faster than five hours spread across a week. I didn’t need to leave breadcrumbs and notes for myself so I knew where to pick up the next day.

   But on the other hand, having a daily cadence of at least some game development time made it easier for me to keep each development session connected to the previous one. Those weeks in which I went multiple days without any game development made me glad I took notes about what I was doing (and curse myself if I didn’t), even if I caught up by putting in a lot more hours than usual. Otherwise, I wasted time trying to remember what I was in the middle of.

   I think that daily cadence, with a weekly goal, did two things for me to help me finish a long project part-time.

   One, I was always focused on outcomes. Whatever I was working on needed to contribute to the finished product.

   Two, so long as I was focused on outcomes, I just needed to put in the hours. Those outcomes weren’t going to happen unless I was sitting in front of my keyboard working on them. And a steady pace was better than crunching and needing to recover, both for my health and my family’s.

   Maybe 20 years ago, I once calculated that after you account for sleep, hygiene, mealtimes, and an 8 hour day job, I still had four hours with which to work on game development if I so chose.

   Today, it feels like there is a lot less time, as there is cooking, cleaning, and chasing after kids, among other things.

   I found I watched a lot of TV shows and movies with my family. I attended baseball games, helped coach soccer, and volunteered. Every so often I played video games, either alone or with someone else.

   And yet I still finished a game because I just kept working at it, day after day, week and week, until it was done.

5. Player feedback as early as possible is huge

   I wrote about how important polish and juiciness was in terms of player feedback, but the biggest influence of my decision of what to work on came from other people playing my game and telling me how confusing it was.

   Even now, the v1.0 release of the game could probably use some more clues. I’ve seen players try to tap on things that aren’t interactive, or wonder what to do, or get frustrated when it isn’t clear how to get a better grade.

   But it was worse mid-development, and I knew because my testers informed me.

   Outside testers had access to devices that I didn’t, which helped me identify major defects with the Android version of the game early on. When it came time to actually publish the game, it was a relatively smooth review process because I already tackled the requirements earlier.

   I would periodically make time to play the game myself, preferably on a mobile device, and there is value to doing so. I sometimes noticed things that weren’t operating correctly, or weird graphical glitches, or game-breaking bugs. And sometimes I would notice when I wished I had something in the game, and then make it happen, such as the indicator that tells you which production run you are currently on.

   

   But working on the game everyday has the downside that I’m too familiar with it. I know how to play the game. I won’t run into the problem of wondering what to click on or what my goal is at any given moment because I already know.

   But reading the detailed reports from my testers helped me to see where someone who isn’t as acquainted with the game ran into trouble, and often it’s a matter of the game not providing feedback.

   Early on, the way turns worked was that they would be continuous until you either hit the Stop button or you opened a menu. Then, it stopped advancing turns until you hit Go. This made sense to me.

   But I got feedback from a tester that it was confusing that the game stopped. If you open a menu and it stops the conveyor belt, then closing the menu should start it, right?

   I didn’t initially want to change the way things worked, but as I played it and felt it was tedious, I decided to implement the suggestion.

   Which introduced a problem in that you had less control over the turns advancing. If you open up a worker’s menu and told them to craft a toy, then immediately the next turn starts. But what if you wanted to tell three workers to craft toys, and you didn’t want to waste turns?

   So I added the Stop button to every menu, which gave you the option to take a temporarily paused turn advancement into a stopped one. Once you were ready, you could hit Go.

   And in the meantime, I added a visual indicator that told you if the game was paused or stopped.

   So it was a lot of work sometimes, but the feedback I got from testers helped me see gaps in feedback from the game. Sometimes the solution I came up with was different than what was suggested, and sometimes I had to cut a suggested feature for lack of time.

   But testers helped me take this game to another level in terms of ensuring it had the best chance of delighting players.

Summary

Between Google Play and the App Store, I can see that I’ve had over 90 downloads of Toy Factory Fixer since I announced it.

I didn’t do a big launch, send the game to reviewers, or anything like that. I only announced it on my mailing list, on my blog, and on social media.

What is most interesting to me is that people are still finding it and downloading it weeks after the announcement.

I don’t know if the game or the posts I made about the game led to it, but Toytles: Leaf Raking sold a couple of copies last month after no sales for many months, which is a great outcome I am attributing to Toy Factory Fixer.

So far, even though periodically people do go to my newsletter’s signup form from my game’s menus, only one person signed up for it, and it was someone I know, so it remains to be seen if strangers will decide to subscribe.

The game took longer to make than I would have liked, but I am glad I put in the effort I did to make it understandable, entertaining, and enjoyable.

I once again validated an Agile approach to product development while also recognizing areas of production in which I could improve in the future.

It’s not exactly a complete success in terms of what I set out to do originally, but this first Freshly Squeezed game is published, getting played, and is acting as an ambassador for GBGames to the public.

Thanks for reading!

—

Want to learn when I release updates to Toytles: Leaf Raking, Toy Factory Fixer, or about future Freshly Squeezed games I am creating? Sign up for the GBGames Curiosities newsletter, and get the 19-page, full color PDF of the Toy Factory Fixer Player’s Guide for free!

Ludum Dare #33 is months old, and the next Ludum Dare is about to begin. It’s about time I wrote a post-mortem about the game I submitted.

The theme was “You Are the Monster”, which, as usual, caused a lot of panic among the participants who worried that it was impossible to make a game based on that theme in two days. And once again, despite such worries, there were almost 2,000 entries submitted, so people found a way to do it.

I ended up making a fairly compelling, somewhat humorous, non-violent strategy game about being an imprisoned evil using its influence to escape. I called it Free Me, You Idiots!

What Went Right

1. Quantity Over Quality: Fleshing Out Concepts and Designs Early

   It’s easy to want to get started right away. You only have 48 hours, and every second counts. But I also knew from experience that it helps to spit out as many concepts as possible early on, then pick the one that seems most promising. You are more likely to have a good idea picking the best of a thousand ideas versus the best of ten ideas.

   So I spent the first couple of hours thinking, doodling, and discussing with people on IRC.

   The theme had a lot of obvious and not-so-obvious applications. The obvious choice is to harvest ideas from classic movie monsters, such as vampires, werewolves, and ghosts. It’s direct, and you could have monsters who are genuinely scary or you could have them be misunderstood by the populace.

   The next-most-obvious choice is to think about metaphorical monsters, such as people who have the most evil and disturbing thoughts. It could be a psychological thriller in game form.

   I thought about characters, origin stories, and settings, and I had a lot of fun thinking up ideas as varied as space aliens and beings frozen in the Arctic for centuries.

   I went with the idea of the incarnation of evil imprisoned in a tree for centuries. With no previous story, I was free to write my own with no preconceptions.

   I fell in love with the idea, which really helped me drive the development of the game. If I hadn’t spent time upfront thinking, I might have settled for a merely OK idea, and I might not have found myself as passionate to work on it.

   I spent quite a bit of time thinking about this concept. I threw out a lot of ideas early on and managed to find the core of the game. I had originally thought about having a bunch of stones surrounding the tree that you were imprisoned in that had to be removed by the nearby villagers under your influence. I mocked them up and played with the idea for awhile.

   

   

   

   The idea was that rock-gathering was an activity for the villagers, and the villagers would grab these forbidden rocks.

   I’m glad I threw it out because I ended up having entities that did hardly anything but explore and pray. I would have run out of time without a way to remove those rocks and finish the game.

   I think if I had more than two days that those rocks would still be in the game and the entities would be mining, chopping down trees, and more, but because I had a wealth of ideas to choose from, discarding some didn’t feel like I was gutting the game of its essence.

2. Iterate Like You Mean It

   In the previous Ludum Dare, I spent a lot of my time on animations and art, but not much at all on game play. By the end of that compo, which had a theme of “An Unconventional Weapon”, I had a monster that could chase around the player when you got his attention either by shouting at him or by crossing his line of sight. I wanted the monster’s head to turn separately from the body, and there was going to be this idea that the monster can trip over itself when it is moving in the opposite direction that it was looking. I spent a too much time creating the complex animation system when I could probably have gotten away with something simpler, and as a result, what I submitted wasn’t anywhere near a finished game.

   I didn’t want to make that mistake again in Ludum Dare #33. Even so, I didn’t start writing code until the compo was well underway. While I was worried that I wasn’t going to make everything I wanted by the deadline, my approach meant I was always going to have something to submit.

   My first focus was on interactivity. If the player can do something, I can build upon it, even if I have to cut most features or change the direction of the game design completely.

   It meant I was also playing the game frequently. I would make a change and immediately see how it felt. I can identify things that may be buggy or hard to understand, and I can fix them immediately or prioritize them.

   For instance, when clicking on an entity in the game, an arrow appears over them. This was a simple bit of polish that I thought seemed important enough to add early on. With a bunch of entities moving about, this arrow meant it was easy to identify which entity you were looking at.

   But moving around the world was a bit jarring. You click, and the camera centers to that location. I knew I wanted to add interpolation to smooth the camera’s movement, but I also knew that what I had was passable and could be fixed if I had time for polish later. So the working yet jarring camera movement stayed throughout the rest of the game’s development.

   What was exciting was that by making small yet meaningful changes, it allowed me to add a relatively complex economy without much effort.

   

   The game features a few variables that are interdependent and make it feel quite rich to explore. For instance, the player has Evil Energy to use to influence villagers. Evil Energy replenishes one point every second.

   The number of followers influences the amount of maximum Evil Energy a player has. More followers meant more maximum Evil Energy, which opened up options for the player.

   When you have followers, they will pray to you. When the total amount of prayer time exceeds 10 seconds, the player gets an extra point in Prayer Energy, which can be spent on upgrades and ultimately on the ESCAPE option, which ends the game in victory.

   I originally didn’t intend to add so much because I was worried about spending too much time on balancing all these variables, but without all of these different values and their interactions with each other, the game felt very simplistic.

   I’ll admit feeling worried that I might find that the whole thing was unworkable and would result in nothing but wasted time. Learning about and playing with game economies from the book Game Mechanics: Advanced Game Design by Ernest Adams and Joris Dormans gave me some confidence I could pull it off by adding pieces of it throughout development and building upon them iteratively. Each addition was relatively risk free and easy to fiddle with.

   I was very pleased with the results, and so were the people who played the game. I was surprised by how many actually played it all the way to completion!

3. A Goal for Being Cool

   I realized that back in pre-20 Ludum Dare compos, it was easy for me to play almost everyone’s games because there was less than 100 entries. There would be a burst of game development activity for a weekend, and then maybe a week of playing and rating games, and then going back to regular life. Plus, back then you didn’t need to rate games to get your game rated.

   Today, with thousands of entries, it takes much more of a commitment. Your game won’t show up in the random sampling for entrants to rate if you don’t also rate other games. It’s a fair system.

   After the compo, I was on vacation from the day job and so I spent the next few weeks playing games. I set myself a Coolness goal. I figured that if I played a handful of games a day consistently for the next couple of weeks, I would be able to hit it.

   And when I rate games, I like to give feedback to the developer. I don’t just post “Nice!” or leave no comment at all, so I wasn’t going to allow myself to speed through reviews in order to quickly increase my Coolness. I was going to play games, give them a fair shot, and give an honest review with constructive feedback.

   

   I only got 63% Coolness, but it was enough to earn me a bronze medal. I don’t understand how that percentage is calculated. I know I didn’t rate anywhere near 63% of the nearly 1,199 games submitted, yet it seems like I could have hit 100% with only a few more handfuls of rated games if my math was correct.

   But I know that if I hadn’t set a Coolness goal, I would have either made the mistake of not rating enough games, resulting in my own game not getting rated by many participants as in past recent compos, or I would have spent a lot of my waking hours doing nothing but playing LD games, ignoring my other responsibilities. I think I had the balance mostly right.

What Went Wrong

1. No Polish and No Sound Effects Again

   In an effort to speed development along, I did the opposite of what I did in the previous Ludum Dare compo and ignored animation and polish as much as I could. It wouldn’t matter how cool something looked if the player had nothing interesting to do.

   As a result, the game lacks life. The entities move around without turning or animating, and there is no indication when a player successfully influences someone. The grass tiles had markings which were too subtle and resulted in a sea of green that made it difficult to tell if the player was actually moving the camera.

   Now, it wasn’t completely bad. The prayer bubbles added a nice touch, as did the Evil Energy ball in the HUD scaling quickly when it replenishes. The flavor text had some humor and lore.

   But no sounds, no animations, and no special effects left the game feeling like a lot was missing.

2. Buggy Engine Code

   There were a few problems with the custom engine I was using. One was something I saw in the last Ludum Dare in which tiles would seem to separate as you moved the camera around. It turned out to be a floating point error that resulted in me accidentally scaling the tiles so they would sometimes be one pixel smaller in width and/or height. I must not have pulled in the fix when I was putting the project together, so I easily fixed it, but it still took me some time to figure out what exactly I was missing.

   Early on I noticed that entities seemed to walk over the tree. They should look like they are in front of the tree when standing in front of it and behind the tree when they were behind it.

   So why were they always appearing on top of the tree no matter where they were?

   I recently ported my code from libSDL to libSDL2, and sprite rendering was done differently to accommodate scaling and rotation. Unfortunately, I introduced a Z-ordering bug which, after spending time on figuring out what was going on, I decided to ignore for this compo rather than spend more time trying to fix it.

   The risks of using your own code instead of a ready-to-go engine such as Unity or Game Maker is that infrastructure bugs like this are not only likely but also might be something you feel inclined to fix immediately. If Unity had a bug, I’d work around it because I’d have no other choice. If my own code has a bug, I’d feel like I could fix it while working on my game and might underestimate the effort it would take to do so.

   So while it wasn’t fatal, I had buggy infrastructure code that slowed me down and prevented me from making a better game.

3. Missing: Some Features and Any Challenge

   One of the upgrades in the game is CAUSE FEAR. Do you know what it does?

   If you guessed it caused the villagers to be afraid, you would be wrong.

   It does nothing. Why?

   Because I never got around to it. I had ideas for what CAUSE FEAR would do, but since so much of the game didn’t get made, such as the mining and wood cutting I mentioned earlier, there was no real reason to cause villagers to be afraid. Maybe it would make them pray harder for a bit, and so CAUSE FEAR was similar to trading Evil Energy for Prayer Energy but in an indirect way.

   I had a few ideas, but none of them got implemented. Unfortunately, I left the option in even though it did nothing. Whoops. B-(

   Similarly, because there was no real conflict in the game, there was no challenge. There is nothing in the game that prevents you from converting every villager, getting all of the energies you need, and winning the game.

   I wanted the good villagers to try to convert back your followers. I wanted different kinds of villagers, such as priests and acolytes, and have their presence force you to change your tactics. Perhaps if your followers got numerous enough, the other villagers would go to war with them.

   Something. Anything.

   But instead, there’s nothing. There’s enough economy and variety that it seems to hide this lack of conflict, but without conflict, I can’t really say I have a complete game.

What I Learned

1. A Design Document Is Key

   Throughout the compo, I set myself goals. I set an initial goal of settling on a concept within two hours. I set a goal of having interactivity early on instead of waiting until the end. I set a goal for rating games after the compo.

   Each goal gave me some control over the outcome. I was able to focus on what was needed and ignore what wasn’t. While the lack of polish hurt, it wouldn’t have hurt nearly as much as having a fairly incomplete experience.

   What really helped in setting goals was having a design document for the project. Now, I’m not talking about a 300 page document that no one reads. I’m talking about a living document that changed throughout the 48 hours I used it to help guide me.

   Everything I thought of went into that document, which allowed me to assess what was a priority and what was nice to have and what didn’t need to be there at all.

   I could see what was left and compare it to how much time I had left, and I could make intelligent decisions about feature cuts and what had to be there no matter what.

   I credit Hybrid Mind’s Ludum Dare #29 timelapse for In the Black for showing me how effective a design document can be for even a short 48-hour project. Thanks, Dave!

2. A Game’s Economy Can Have a Big Impact on the Game’s Appeal

   I was glad I decided to risk adding more to the economy of the game. Evil Energy and Prayer Energy play off each other in an intuitive way, which makes the game more compelling.

   I think people kept playing because throughout the game there was always something to aim for. Converting a follower resulted in an upgrade being made available, which required Prayer Energy to attain, which in turn allowed more followers to be converted more easily, all with the aim of using ESCAPE to end the game.

   I could have had a simpler economy. If I eliminated Prayer Energy and made things more directly available, it could have worked, but it might have felt too simplistic.

   What I’m not happy about is that the richer economy hides a lack of challenge. I don’t want to use economic designs to act as a bandage on a fundamentally broken game.

   But I did learn that doing a decent job designing an economy can result in a great return on investment.

3. Keep Your Nose to the Grindstone

   During the 48 hours of Ludum Dare #33, I slept, I ate, I showered, I talked with my wife, and I blogged.

   But most of my time during the 48 hours was spent on design and implementation.

   According to my records, I did a little over 24 hours of game development between Friday and Sunday. Just over half of the time spent on the compo was spent doing actual game development, and I’m pretty sure that because I lived and breathed this project, I was even making progress while showering or eating.

   Even with the failings I mentioned, I can see that this laser focus was the real reason why I was able to get a decent game finished.

   I’ve done compos in which I was distracted, and they didn’t always end well. Either I was dealing with poor health, in which I couldn’t sit at my desk for long periods of time, or I was going to a party or a soccer match.

   For this Ludum Dare, I made sure that I was 100% focused on the compo with no outside commitments to keep me from doing anything but game development that weekend.

   It was grueling and exhausting, but that concentrated effort made things move along so much more quickly than if I had spread out the development effort with frequent and potentially long breaks.

   I’m not saying breaks are bad. I took breaks.

   I am merely saying that my mind was focused on game development and not on how to interact with other human beings in a social situation or how frustrating it is when your team is losing due to the same blunders they always make.

   I rested when I needed to rest, but I didn’t allow myself to procrastinate or do much of anything other than participate in this compo.

Summary

The results of this compo were very encouraging to me:

Out of 1,199 entries, my game was rated in the top 36% overall and top 8% in innovation. My only rating that was in the bottom half of the group was for graphics, and even then it was at the top of the bottom half. B-)

I credit my ability to focus that weekend on game development almost exclusively, and all of the tools that allowed me to leverage that focus, such as my design document and setting milestones.

The main complaint from players is that there was a lack of music and sound, followed by the noticed lack of challenge. For future compos, I’ll want to focus on not only interactivity but also adding real conflict.

I may want to experiment with adding audio iteratively. Normally I add it at the end, and an informal survey indicated that a lot of other game developers add it at the end as well. I wouldn’t want to spend time on something that might get thrown out before the end.

Still, the idea is that if I can have a playable game early on in a compo, I can also have a playable game with audio early on in a compo, too.