Fail fast — you’ve probably heard this advice before. It’s very common in art, indie development, or even startup culture. The idea is that if you try enough things, surely something will stick. Spending too much time or too many resources on one thing in particular without knowing that it will work ahead of time is a road to ruin. Instead, ship quickly. If something works then improve it, if not drop it like a hot potato and move on to the next project. On the whole it’s good advice. I’ve received this advice countless times and even given it now and again. In late 2016 I decided to put this advice into practice; it’s now 3rd quarter 2020 and I’ve not shipped. There’s no lesson or moral here, only a story of failing to fail fast.
In 2014 I had an idea. Everyone has those, but this was a good one. I prototyped it. Afterwards I remained sold on it, so I continued work. Another developer even took a stab at it, though certainly not in quite the way I envisioned. At some point in 2016 I was forced to admit that doing this idea justice was beyond my means. It required a lot more art than I could afford, plus there no doubt were some unknown unknowns in play regarding the endeavor of publishing a commercial game. I had briefly considered trying to crowd fund. Unfortunately my starting network would have been approximately zero; living in near-monastic seclusion has its drawbacks. Instead I decided to try a different idea (we’ve all got loads, right?) that I could finish quickly on about $500. I could make every mistake and discover all those unknowns without much risk. It would in all likelihood fail, but hopefully it would at least recoup what I spent. I’d additionally have some sort of following when I resumed work on the previous game.
The idea in question could scarcely be simpler: a text game! Along with cloning Pong, making a simple text adventure is one of the frequent assignments given to the newest of the new game developers. The engineering required is fairly minimal. Indeed, I was fresh off a game jam in which I wrote a simple finite state machine to power a visual novel retelling of the Faust story. The concepts were all top of mind. What could be simpler than that? Setting aside my low level fixation — my previous work was in plain C with embedded Lua for scripting support — and working in the web stack! JavaScript undeniably has its warts, but the UI system I wrote in Lua needed years of work to even match the feature set provided by CSS. Further I didn’t have to compile anything and porting to other operating systems was a breeze.
Everything was coming together. I believed in the idea. I’d even made a small prototype of this in early 2014 but set it aside because I thought it would be too simple. The new technology choices promised to make development smoother. Even the visual style readily presented itself. Full of purpose, I drew up an outline and began work. The first commit happened in January of 2017, which can fairly be called the start of the project: a medieval kingly life simulator called Arbitology: Dei Gratia Rex.
The keen-eyed reader may have noticed something amiss in the last paragraph. What sort of visual style would be necessary for a text game? Parser-style Interactive Fiction is certainly something that exists. My original encounter with that was Zork on an Apple II compatible, but the IF scene is certainly richer than that. For better or worse modern games — even text games — demand a certain amount of graphical content. Indeed the game that most influenced Dei Gratia Rex is the beautifully-illustrated King of Dragon Pass, and comparisons would almost certainly be drawn between the two. The technology chosen makes adding images easy enough after all, so for a medieval text game it just seemed obvious to go the skeuomorphic route and make the game an interactive illuminated manuscript. I thought it should be fairly easy to raid the public domain for the miniature illuminations, majuscules, page borders, and all the rest. For those things I absolutely needed to be custom, I had even found an artist on DeviantArt who fit the bill perfectly in terms of style. Perhaps this was a bit harder than printing ‘West of House’ to the console, but it wasn’t too bad right? Wrong as it happens, but more on that later.
As I was doing the initial work for DGR, I also was trying to teach myself what I needed to know in terms of business and law. Some people ignore all these things. I did consider doing that as well, but, dear reader, I am not lucky. If something has a chance of going wrong, it frequently does where I am concerned. Having been burned a number of times over the years, I now take every reasonable step to insulate myself. Programmers are an arrogant tribe. Hubris is actually one of the three virtues of a great programmer per Larry Wall, and it’s one to which I aspire. To that end I just assumed I could handle all of the petty legal and business stuff. I got a certain distance but ultimately admitted defeat. Mistakes in code typically cost zero dollars to fix, but legal matters wrong turns out to be quite expensive. To cut to the chase I needed some expert help. Getting contracts, trademark, business formation, etc. handled ended up costing $4,500. This was actually cheaper than what most attorneys who do this sort of work charge. I checked. Unfortunately getting the trademark handled at this juncture was premature; that set up on ongoing liability that cost more money. And it turns out that using the registered agent address as the primary business address was something that was done that I had to fix later. All that is to say that I spent nine times the intended budget upfront in startup costs and had other recurring costs on top of that. I still had not properly sorted art, music, or anything else game-related.
Speaking of art and law, my original idea was in fact terrible for a lot of reasons. There are many digitized manuscripts online. Most of them are provided by European institutions. In the U.S. a scan or flush-on photograph of a manuscript that is in the public domain is always in the public domain; a ‘slavish copy’ as it is called does not attract a new copyright. Not so in Europe! Worse, the specifics vary on a per-country basis. The whole thing was a minefield. In the end I opted for only original work, despite knowing that this would cost more. I reached out to the artist I had previously discovered. Things seemed to be going well. They were receptive to working on it and only proposed to charge a bit more than I expected. At a certain point after preparing an art brief, emails stopped getting answered. I took the hint. Thanks to the intercession of one of my friends, I found someone else eventually. I’m beyond pleased with the current situation, but it is fair to say I was rather demoralized at the time when all of my initial plans regarding art fell apart.
I kept working through all of this. Systems came together. Content was written. Unfortunately these things ended up looping. Most of it was hubris (see above). Simple text games are very simple; it’s almost a tautology to say that. However once you go beyond simple, text games become fiendishly complicated. When you are running a simulated kingdom to generate events to feed into an interactive narrative game or find yourself reading doctoral theses and other actual scholarship to try to avoid anachronisms, things are actually rather difficult at times. I assumed I could power through all that. Sometimes I could, other times not so much. And then of course I had to add a strategy layer with clickable provinces on top of all the rest. Before I knew it this text game had at least met, if not exceeded, the original idea’s complexity.
Art was finished by the second half of 2018. The opening phase of the game was completed and I continued to write content. And systems. And more content. And so on with no end in sight. Costs continued to accumulate throughout all this time, and even began increasing. It turned out that in my location I actually do need a physical address to make every relevant authority happy. In November of that year I ended up signing a lease on an office, since using my home address was not a good idea for many reasons. Living in the Appalachian part of Tennessee, the costs are low. That’s part of the social contract here: everything is cheap, but it’s utterly impossible to make money outside of a few limited careers. Even taking the low costs of the region into account, the deal I got was abnormally good: an all-access pass to GDC rents my office for twenty-three months. I couldn’t really afford it (and still can’t for that matter), but I also couldn’t afford not to accept the offer. Being able to work in a dedicated work space was also a huge boost to productivity. Good progress was made during 2019. I got the Steam page up and started marketing. Things were looking up, except that there was a bit of impending doom as the first hard deadline approached at the end of the year: unfunded liabilities.
I blew right through that deadline of course. As 2020 began, I had about 3 months of money remaining, taxes due in a few months that would empty my coffers, no music other than some terrible work I did myself, and the realization that I still needed a few more art assets. Then COVID-19 hit and I could no longer go to the office. I still can’t if I’m honest. We had a lockdown with 30 active cases in the county, but with more than 600 active cases we had open restaurants and bars. The stimulus funds landed just in time, but it was not going to be enough. In order to stay afloat I sold everything I owned that had any value: an impressive Magic: The Gathering collection including a piece of power plus lots of retro gaming gear from my childhood. I kept my computer of course. The venerable 2010 machine is legitimately worth less than some of those pieces of cardboard.
The next deadline is harder and there are no ways around it that are not very expensive. Far more than I can afford in any case. I have to launch by April to avoid that, but the sooner the better. I still don’t have music sorted, though I have a few leads. I still need some more art, but new laws that appeared in the interim may complicate matters. Marketing has been a mixed bag. I’ve tried a few things — some worked and others didn’t. The Steam page has been able to attract a little over 1,000 wishlists in about a year. That’s not enough to break even, but it is enough to indicate that there is interest. I don’t need to sell a million copies. I don’t even need to sell 10,000; I’d be fine at 8,000. That would be competitive with a fast food job. There is still some hope to reach that, I think.
I’ve failed to fail fast. The all-in external costs have exceeded $10,000 when they were meant to be $500. The fourth anniversary of the initial commit happens in less than 4 months, but the project was meant to last less than 12 months. I have some indications that the game will not totally flop, but I do not as yet have any indication that it will be a success either. Part of this was pure arrogance on my part in underestimating the task. Another component was that ever-present foe named ‘scope creep.’ Yet another aspect was scrupulously doing things the right way when cutting corners may have been wiser.
Time will tell if this story’s conclusion is failing to fail or merely failing to fail fast. This is normally the point where I’d drop a call to action. However in the spirit of indiefuck I’ll instead end by restating for the record that failing fast is a task which can itself be failed.
Robert is the owner and primary developer of High Tower Games. He is currently working on his first major commercial release, the medieval narrative simulation game Dei Gratia Rex. You can find High Tower Games and Dei Gratia Rex on Steam, Twitter, and the web.