[This post contains numerous spoilers for Star Crystal Warriors Go.]
Hello! My name is Brian Rushton and a few years ago I released In the Service of Mrs Claus, followed by playing every Choice of Games game at the time and writing an essay about it here:
Recently, I had the opportunity to work with Choice of Games again on the game Star Crystal Warriors Go. After the first author had completed 8 chapters, I was brought on to finish off the end of the game and to take it through beta testing. I had the first eight chapters and the design document at my disposal.
Playing the game, I realized that it was very fun and engaging, with interesting characters; that it was very similar to my own writing; and that it wouldnât take too long to finish. I was excited to come on board and to apply what I learned when I wrote that essay above.
The problem is, though, that itâs much easier to criticize than to produce. I can recognize a great game when I see one, but making one is a whole different matter.
I resolved to do two things:
- To make the rest of the game adhere as closely as I could to the authorâs previously completed material, and
- To give the game as much time as I needed, with no rushing or cutting corners.
Finishing the last few chapters
My first work was on chapter 9, the big finale of the âhumanâ side of the story, where you finally get to present your clubâs display at the school festival. This was a daunting task, because the author had set up 4 possibilities for what club you have, and each of those has 5 possibilities for what type of activity youâll be doing (competition, presentations, hybrid, etc.). And there are 3 possibilities for what kind of relationship you have with a rival club. So the first 2000 words I wrote were literally just the first pageâs description of your club.
While I tried to stick close to the previous authorâs vision, I also tried to add more options in this chapter that werenât direct pass/fail tests but instead different strategic choices, like who you want to spend time with or what accessory you want, as well as dialogue options with no stat effects at all (just roleplaying). I also gave it a âday of the weekâ structure with each day announced by the choice bar at the bottom, because Iâve noticed that giving a ârhythmâ or âcountdownâ to choice-based games increases enjoyment and sets a pace (just like seeing the remaining pages dwindle down in a physical book).
I also made a lot of dumb choices, so if you liked the book, thank Abby Trevor, my editor, for making a lot of helpful suggestions. For instance, I put a lot of events in this chapter where the Nightmares cause trouble, but I didnât let the player respond to it. Abby suggested that players would appreciate more agency, so I went back and added numerous scenes where you can respond by hunting them down or checking in on the Dream Kingdom.
I also put a lot of establishing choices in this chapter at first, setting your powers and traits, because mentally this was the beginning of the story I was writing. Abby encouraged me to make more testing choices instead, since this was the end of the big story.
I had friends test this along the way as I wrote. One thing I learned with Mrs Claus and my numerous parser projects was that testing should really take up almost as much time as writing. The earlier you test, the easier it is to fix problems. Saving testing for the very end is problematic if your game has an unfixably bad structure.
Fortunately, friendsâ responses were favorable to both the base game and my additional chapter. I kept my expectations low, since my first game (Mrs Claus) did so poorly that according to my monthly royalty results email it will be roughly 1000 months (literally) until it breaks even. But the response was encouraging.
For chapter 10, I again decided on a ârhythmâ/âcountdownâ by having five floors of an evil tower. I was heavily influenced here by Sailor Moon; I bought the Eternal edition with birthday gift cards and read it heavily (things like the evil mirror are directly inspired by it). While I struggle with relationships and romance (one astute reader pointed out that the moonlight dance ended up being almost an afterthought), fantasy battles and location descriptions are things that I deeply enjoy writing and this was a fun chapter to dig into.
One major concern I had with the âbase gameâ was something that I had indicated as a âno-noâ in my essay on CoG games: a lack of major branching or alternative goals. In my own earlier game, Mrs Claus, I had initially proposed three different âfinale chaptersâ to fight against the three main factions, but youâd only get one chapter depending on branching (like Choice of Robots). My editor advised that that might be too ambitious, but I overcompensated by including almost no branching in that game. I didnât want to make that mistake again, but with eight chapters already finished, it was too late to add major branching goals that would completely change the endgame.
So I added a âturning evilâ path and resolved to make it long and satisfying so that it would be clear to players that the game has actual replay value. As an author, one of the biggest mistakes you can do is adding complexity to a game but hiding it from the player. To a player, there is no difference between a linear game and a branching game that looks linear.
Finally, I worked on the ending chapter. I wanted to have different âtracksâ for the endings: one track for the âgoldenâ ending (where you basically become a God), one for the ânormalâ ending, one for the âyou diedâ ending, and one for the âevilâ ending (I ended up combining the latter two, with some variation for each).
One thing I had learned in playing all CoG games was that you need a properly satisfying plot arc, which includes a denouement. I also hadnât added as much relationship material as I could have in the game, so I wanted to make the ending essentially nothing but relationship material, giving it a ârhythm/countdownâ where you see all the main characters one at a time before concluding the game.
Again, Abby was an enormous help. Her suggestions greatly increased the game size. It was 100K when I received it, and I planned on writing just 50K more. With her suggestions for extra scenes, we ended up at 250K.
Brushing up the rest of the game
How could 3/11 chapters be 60% of the game code? Well, itâs not; I also went back and added more to the early game.
Based on my analysis of CoG games, I identified several âweaknessesâ in the mechanics of the early game:
- Setting choices often mixed skill increases (permanent, powerful) with opposed stat increases (weak, subject to change). A very common pattern was three conversation responses where two changed stats and one was the âsmart choiceâ that increased Insight.
- Testing choices often mixed skill checks and stat checks. One particular case was that Courage could be used as a power to fight things, but not caution, which meant that players that played fearful characters would get locked out of fights.
- There werenât many choices that impacted secondary stats like festival prep and investigating your dreams.
- Some choices made it difficult to tell which stat was being affected (especially strength and finesse).
I went through and revised dozens of options. For the âcorrectâ insight choices, I removed them and sometimes made them static insight checks outside of choices. So, there are five or so places in the game where you get additional text just for having high insight. I added extra choices to make sure there was equal representation of different personality traits.
I also went through and took out all the times that courage could be used as a skill, and instead added more options for other traits.
To help boost secondary stats, I went back and added several vignettes in earlier chapters, about one per chapter, where you can choose which secondary stat to focus on (like picking between researching nightmares and prepping for the festival) as well as choices that focused on a single secondary stat (like sending emails to prepare for the festival).
For the difficulty in telling the difference between stats, I worked really hard. I took input from beta testers and my own numerous playthroughs (every author should really, really play their own games often! If itâs not fun enough for you to play, how will anyone else enjoy it?), and anytime there was confusion I reworded the choice to be more clear. In very difficult cases, I would just straight-up use the stat name in the choice. I renamed Strength to Combat Strength, because all the exercising scenes felt like they could be strength or finesse, but strength was only used in combat. But still testers were confused. So, with Abbyâs help, I added the âvisible stat testsâ mode and end of chapter checkpoints. I also added explicit guidance in a help menu on what each stat did, when it was useful, and recommended target numbers. Then, with Abbyâs enormous help, we ran a lot of randomtests and structured the game difficulty to make it reasonably possible to succeed.
Difficult decisions
There was one big flaw remaining: this was a pass/fail game. All of the objectives could be pursued and won at once. So either you win, or you donât. This is as opposed to games like Choice of Robots, where you have to commit to distinct paths like âmaking robot companionsâ or â leading an armyâ, or Creme de la Creme where you can only be in one club and have numerous other decisions that differ from one playthrough to another.
This is bad. In my essay on CoG, I pointed out how what people like in these games isnât difficult stat checks but difficult moral decisions. This is one reason, I believe, that romance is so effective. Having multiple romance options means focusing on person to the exclusion of others, in time if not in affection, requiring multiple playthroughs.
But this game had pass/fail baked into its structure. And time and time again in playing low-rated choicescript games, I saw how authors had realized the same problem, that the whole game could be beaten in one go, and decided the game was too easy. These authors (including me with Mrs Claus) try to fix it with hard stat checks, but then people just lose over and over and get bad endings and end up hating the game. I ragequit my replay of Mrs Claus where I romanced Claus herself but couldnât resurrect her at the end. I wrote the game and canât beat it!
So, many people have said Star Crystal Warriors Go is âeasyâ, âstraightforwardâ, etc. (mostly in a positive way). This is completely intentional. Yes, a really easy game isnât ideal, but âhard gameâ isnât better; whatâs better is multiple compelling, competing objectives, and I just didnât have the structure for that. Adding in the dark path was my best option, since it both gave one major branch and also encouraged players to see âfailingâ text, of which there is a ton in the game, and which most players donât see as they do their best to win.
The other difficulty was that there were tons of opposed stat checks for success in the game. I put in my essay that those were bad, and that opposed stats are better suited for static game checks (like, just changing the descriptions of things around you and what people say). The biggest knocks on this game have come from those checks, and I agree completely. However, they were too baked into the game to remove without disturbing the overall structure, and I didnât want to mess up the authorâs vision. A lot of authors (including past me) rely on these because otherwise every check is a skill check; this is, again, solved by having more choices that choose between different strategies or scenes or other kinds of branching, instead of having a linear succession of pass/fail. [As a side note, resources are another good way of adding complexity without opposed stat checks. Money in Night Road, for instance, gives you a lot of fun options.]
The role of beta testers and testing
My editor Abby, my final update coordinator Mary, my friend testers, and public beta testers all contributed a ton. Once beta testing feedback came in, my choice to not touch the authorâs earlier work went away, because I will change anything in a game if enough testers want it. The testersâ voice is, to me, much more important than my own. Numerous things were added. The early chapter had very few descriptions and many people found it bare-bones and too rapid-pace. But testers loved chapters 6-8 by the original author with all the lore drops. So I took the writing style of those later chapters and added more descriptions to the first chapters, like the city streets or your bedroom.
I myself wanted transformation to be cool like in the shows, so I added the âcatchphraseâ, more outfit options (including the embroidered tunic, since I wanted a more ancient and primal option for non-feminine players), and expanded each transformation scene. I also liked how Sailor Moon kept getting cool accessories that featured in her transformations, so I added powerups for people who maxed out a stat (or at least got it up high).
Testers wanted to know more about the brief mention of a dead mother, then wanted more, then wanted even more. Even in reviews now, people want more. If thereâs every any expansion in the future, Iâll look there.
Testers also wanted to know more about Polaris. The line about them not needing food during the day came from a tester concerned about if Polaris was eating enough.
I love testers, and I really appreciate everyone that helped!
Overall outcome
I told myself that since Mrs Claus was perhaps the lowest-selling CoG game of all time, that either this game would do better than that, or become the new worst game of all time, bumping up Mrs Claus to 2nd.
Of course the whole game concept and most chapters were written by another author, which makes the success really about Holly and not me. But I completely retooled the mechanics, so I kept a close eye on what people thought of the mechanics and endings.
Iâve been very happy with the outcome. I donât know how sales are looking; itâs possible theyâre bad. But the game has 8 reviews on steam, all recommended, and got two positive reddit threads and a mention in another. Itâs picked up a lot of ratings on the choice of games app. And response on here has been very positive. I see people who love the lore and the characters (which is good because I love what the other author did there, too!) And I see people who liked the visible stats test and end-of-chapter checkpoints (I did that!).
People definitely see the flaws. Most ratings are a 3/4, a 7/10, and so on. Iâm okay with that. I think the core was solid, and I think I boosted it, but there are still structural issues, and so with the mixed reviews I nod my head and think, âyeah, this person gets itâ.
Seeing everyone be so kind and positive has been really refreshing. Mrs Clausâs failure has been a burden on my shoulder for half a decade, a shame I recount at parties (âDid you know I published a book? It was the companyâs worst-selling book of all time!â), so Iâm glad to be out of that shadow.
I also really enjoyed collaborating and hope to do it again. Iâve started writing a short CoG game based on going to a Brazilian Jiu-jitsu gym, because my sister is obsessed with that, but Iâd like to try my hand again at either another collaboration or a new original game further down the line (after I finish a parser game Iâve spent years on).
Studying every past game definitely helped. It didnât grant me godlike powers to be the next Harris Powell-Smith or Kyle Marquis; writing is still hard, and you canât reduce it to a list of checkboxes. But there is at least a list of checkboxes of what to avoid. No matter how sales go in the future, Star Crystal Warriors Go feels like a success to me, for both authors, and Iâm grateful to everyone who made that happen!
donât give up