GROWBOTICS postmortem

Also posted on my blog at

I won’t be addressing all specific criticisms here, but I have been paying attention and appreciate everyone who took the time to write reviews, comments and emails. It’s been valuable for me to have this much attention from constructive and dedicated people. I’m going to discuss my intentions in some detail here. Most of them didn’t work, but hopefully they’ll be a teensy bit interesting anyway.

GROWBOTICS already feels like something I made years ago, rather than something I made a couple of months ago. That’s probably good, but means even I don’t understand it as well as I did while I was making it. It’s in-character for me in that it’s far more concerned with themes and structure than plot or characters. It’s out-of-character in that it goes broad where I’d usually prefer to delve deep. Also, in places it makes me look as though I have a sense of humour, which can’t possibly be accurate.

A competition deadline was a useful motivator and I’m glad I was able to experiment and put this out there, but I always knew it would be a hard sell for a lot of reasons. Including not being a good fit for story-focused judges, which I certainly wasn’t trying to be. Just scraping into the Top 40 is probably fair given what I was up against.

I was hoping more for mixed response than flat out bad response, and that at least some people would engage with the themes. I did receive positive comments, mostly in regard to the work put into presentation and the range of possible combinations, so I can count some success there. The theme side was far less successful.

So, let’s get into it.

The title

First thing’s first. I know if there’s any kind of karmic justice in the universe I’ll eventually suffer for the name GROWBOTICS. It just stuck, and all-caps was an additional indulgence. Then again, I didn’t want it to sound like me. It’s more like something created by a marketing team who probably thought they were being extremely clever. We’ve no shortage of popular products with silly names and odd capitalisation, after all.

One of the endings mentions the competing product iBuild, so maybe that’s the sequel in some alternate universe where I’m foolhardy enough to make a sequel. I hear it’s far more user friendly.

Creating in context

The item combination system in GROWBOTICS dominates everything else, and what framing story exists is extremely stripped back. Contexts are vague and characters are not well defined. In some cases sections rely on tropes to make the location easier to imagine. Context was important to me but I wanted it to come from you as much as from me. So I described a location in as few sentences as possible. Just the seed of an idea for someone to latch onto and flesh out their own setting. Maybe to think about their own creative processes and influences.

Relying on people’s imagination was demanding a lot of players, many of whom would have preferred me to take them on a more specific ride. I’ve recently been interested in the idea of game design that demands as much of the brain as it does of a computer, and not just in a puzzle-solving sense. It’s not a simple goal, and I suspect I’ll be thinking about it for a long time even if it does lead to a lot of creative dead ends. This seems like a decent place to fail at that idea, being about a product that’s supposed to foster accessible creativity and probably can’t.

The conclusions are all jumping off points rather than endings proper. Some of these do act as mini-essays and suggest a specific idea, but these are still simplistic and incomplete. Maybe even straw man-like. They aren’t meant to be taken at face value any more than the sillier endings.

As most people probably gathered, endings don’t depend on what you create, just on your location and how you feel about what you produced. This does reflect my ideas around creativity, the context and your own emotions about it matter far more than what the thing actually is. Locations came in six categories that determine the type of ending. I don’t personally think of any as good endings exactly, but you can think what you like.

Location categories and endings, if you really want to know:

[spoiler]Category A: library, cyberdwelling, greenhouse, blanket fort, crypt, root system
= sleep endings (identical dream sequences)

Category B: shag pad, wizard’s tower, undersea garden, icecream truck, torture chamber, imaginary ballroom.
= eating or drinking endings

Category C: penthouse office, secret lab, spiderweb, treehouse, monster lair, cloud palace
= power (superheroes) or weakness (alien abduction?)

Category D: motorcycle clubhouse, airship, druid grove, stuffed animal teaparty, gothic mansion, religious cell
= revenge or a sort of romance

Category E: dressing room, moonbase, lighthouse, dollhouse, dark fortress, artist’s studio
= shame (hide the evidence) or pride (share on social media)

Category F: gallery, detective’s office, moonbeam, gingerbread house, nightmarescape, surreal dreamscape
= concepts of ignorance or insight[/spoiler]

There isn’t much logic to why a location is in a particular category, beyond having a diversity of options in each, and I do think that was a mistake. At the very least I could have colour-coded them or something. In the end it might have been better just to have one location per category. The initial idea was to provide enough variety that there would at least be somewhere for most people to be willing to imagine themselves.

Combining Objects

In the iOS/Android game Alchemy combining items really is the whole game. It’s not particularly deep, but it still reveals a lot about the person who made it. Society, personal politics and biases are baked into the system in an unavoidable way.

Gender’s an easy one to talk about, which isn’t intended to pick on the creator of Alchemy specifically, just that they’re reflecting something societal and often unconscious. In Alchemy, the object “Man” is used as a stand-in for humanity in general, and is used in a wide variety of combos. “Woman” is a more derivative product. She can be created in two possible ways. Man + Milk (suggesting women as nursing mothers, or more crudely as people with tits), or using some combinations of diamonds, flowers and perfume (women as the receivers of romantic gifts, presumably from straight men). Once you have “Woman” she can be used to create three products total and one of those is a baby. That’s not counting Man + Woman, which produces Sex. Another heavily loaded combination when you think about it.

I’m not happy about it, but in a system where a lot of people are expected to intuit the possibilities it’s perhaps unsurprising that things end up reduced to stereotypes. It makes concepts feel naked and I was interesting in exploring that. I was very conscious of how exposing it was to build the GROWBOTICS combo system. I know it says something about me when you punch in Belief + Identity and produce Confusion. Whether that’s meaningful to you is another question. I could only work with my own associations and hope something resonates, or alternatively that it feels good to be surprised.

GROWBOTICS isn’t about finding all possible combinations, as in Alchemy or Doodle God. Rather, it’s intended to be about creating one thing that you (or a character you imagine) can attach a strong feeling to. And I didn’t want to suggest how you should feel about any of these objects, or cheaply categorise them for you. Some seem obviously negative, like a sense of shame or a broken bone, but even there I don’t want to tell you what that should mean. It was particularly important to me for the system to remain neutral on objects with a lot of loaded meaning, such as a gun or a sex toy.

I know a lot of people wanted more consequence from their creations, but as with the settings, I was more interested in the associations people naturally have with particular words and objects. Take the solenoid value as an example. It’s one of the more technical products in the game and might not mean much to a lot of you. Maybe you don’t even know what it is. But for me it reminds me of my anxieties around not being a very practical or hands-on person when it comes to building and fixing things. I have a lot in common with my father, but this is one of the ways we differ. I’ll never build a house, or create an amazing garden with a gravity-fed irrigation system.

I can’t know the stories of people who play, but I know they exist.

Of course, it’s hard to know how I’d feel about creating a solenoid valve in GROWBOTICS. The biggest weakness from my perspective is being forced into this extreme binary when it comes to emotions, and what it means to be satisfied or unsatisfied with a result. After having such roughly sketched settings you spin off into endings based on rather extreme, concrete reactions. But maybe it can at least prompt something when you inevitably disagree and see the ridiculousness in what these characters do.


The combo system is more complex than it strictly needs to be for where this is aiming. It’s probably hard to see why I would put myself to this much effort. The personal-level reason is simply that I’ve been unemployed and it felt good to exert control over something but there is a more design-based reason. It’s impossible for me to represent a machine that can create anything, at least without some particularly clever procedural generation. But, if you make a system complex enough that people aren’t seeing the edges it’s effectively infinite.

That doesn’t completely work here since you can see the starting components and understand the scope of possibilities in a mathematical sense. But my hope was that people would spend a short enough time with the system that it still felt surprising. These kinds of things are a lot less interesting at the point where you just start trying every possible combination. For me, a game like Alchemy starts fun and interesting, seeing where my ideas do and don’t match up with the designer’s. But at the point where I stop feeling inspired and become more rigorous about testing combinations it loses the magic.

Reading the Manual

The manual is the core of the design for me, though it wasn’t presented clearly enough for most people to see where I was going with it, or in an exciting enough way to make people want to engage with it. I’m sorry; I just naturally like manuals personally! Live and learn.

To clarify, if you decide to read the manual it usually locks the system into one of four possible “modes”: physical; audio/mathematical; visual/perception; or conceptual. The most obvious influence of being in a particular mode is a restricted list of four starting essences, whereas if you skip the manual all sixteen will be available and there’s a high chance of invalid combinations. So the manual creates a more manageable system and was supposed to train people in which objects will combine successfully. If you understand the basic combo rules, and stick with essences from the same mode, it should never fail, but learning the system is a bit of an ask. I did deliberately include a PDF cheat sheet to shortcut the process if desired, but I wish I’d put more effort into in-game hints about how the mode system works.

Mode also influences the text in the setup sections. For example, if you’re in visual mode the button will say “That doesn’t look right” where in audio it will be “That doesn’t sound right” and so on. Similarly, the location descriptions focus more on these aspects, perhaps describing the colourful landscape in one mode, and how quiet and peaceful it is in another. If you skip the manual those parts of the location description are randomised.

This meant the setup parts were written in an unnatural, disjointed way. Sometimes including words that don’t come naturally to me but better suited a particular mode. This was much harder work than setting up the hundreds of combo possibilities, even though it’s a subtle thing no one probably noticed, and just made me look like an even worse writer than I am. But I was fascinated by taking this mechanical approach to language and forcing everything into these restrictive frameworks.

The manual’s modes were meant to represent different teaching styles, but in a simplistic and artificial way. Maybe like what becomes of a lot of complex ideas once they’re stripped back and packaged into products by entrepreneurs.


I won’t discuss all of the underlying ideas here. Let’s just say there’s far too much depth for something that feels like a throwaway toy, and it can’t possibly support it. The ideas would have been better off split into multiple games if I wanted more clarity of purpose, but then again clarity wasn’t the goal.

One of the things that surprised me was how inclined some people were to interpret my writing as sincere. Did you think the happy endings were genuinely happy? How would they look tomorrow or next week?

Beyond that it’s certainly fair that the purpose was kept obscure. There’s a lot hinted at about creativity, and the role of large companies in that, but no real conclusions.

It’s complicated, I think, so you’re seeing the result of some of the spaghetti in my own head. I’m wearing a FitBit right now and don’t have any kind of high ground when it comes to talking about hype and brands, but there’s always tension in that. The aspects actually worth it versus the things that are just about a status symbol or joining a bandwagon.

When it comes to creation tools, I want accessible options for a broad range of people to create things and share them with others. I love that Super Mario Maker exists. But sometimes I have doubts and fixate on the limitations. It’s an unfortunate side-effect that Nintendo has such a heavy influence over how people think about designing a platformer. It makes me wonder about all the thing people might come up with if they were learning in some other way.

It’s possibly appropriate that I spent a lot of time here butting up against the limitations of Twine, even though I do love Twine.

Themes that include limitations, and how easy it is to be unhappy with things we create in general, meant putting people through an unsatisfying experience by design. The original concept was also far more intentionally frustrating. Focusing on the choice between skipping the manual and being dumped into a confusing system that fails most of the time, or following manual and taking the easier route (but being more restricted).

I moved slightly away from that concept over time, eventually adding the mid-game option to check the manual after a few failures. There’s no logical reason to prevent someone from changing their mind, after all. The little remaining indulgence is a rare chance of getting a faulty manual and being stuck without any help available. For any judges who encountered that I probably deserve whatever score you gave me.

Of course, there turned out to be more frustration outside of the parts I intended. It’ll make me think a lot more about how to implement any help systems I might write in the future (and encourage me to test them a bunch).

Harlowe and Twine 2

A few technical notes for those who are interested.

I’m not sure if I’ll be working with Twine again, at least without some additions such as the ability to iterate through an array. Harlowe in particular caused me a lot of problems. I like some of what it’s going for but it doesn’t feel complete enough at the moment. There are so many ugly hacks here. By the time I properly understood what I wanted, and that this wasn’t the best format for it, it was too late to change. This is less a criticism of Twine than an acknowledgement that it wasn’t best suited to my needs in this case.

I didn’t have an efficient option for determining the outcome of a combination and selecting the appropriate icon. I was able to improve speed somewhat from my initial attempts, but it’s still not always enough. I know that on some browsers GROWBOTICS is too slow to really be playable, and I feel terrible I wasn’t able to ensure a more consistent experience for everyone. Sound is also unreliable, but that I always expected and at least isn’t critical.

To try and speed things up, I had to create hundreds of separate passages to represent the combo possibilities. Once you have a story that large it takes about five minutes for the standalone Twine program to open, and there’s a massive risk of errors that take a heap of testing to find. As much as I set myself up for some tedious work this was far worse than it needed to be.

Still, given all that I’m happy with how well it came out after thinking I might have to abandon the project entirely. A little bit of clunkiness isn’t out of place for this machine, I just wish it wasn’t quite so much.

My coding skills have a way to go, but they’ve improved a heap in the last few months. So that’s increasing the number of tools available to me, and bodes well for having more options in the future.

Future Plans

I wasn’t planning to jump straight into more IF creation (or similar projects), but of course as soon as I said that I immediately started getting new ideas. Whatever I do next, expect it to be very different. I’m particularly interested in collaborating more with some of my talented writing friends so I can focus more on the aspects I’m suited to and interested in.

I was surprised by the number of people who were interested in further development of GROWBOTICS to produce something more robust and interesting. I’m not sure what that would look like yet, but I’m thinking it over. It would certainly need to have different aims, but I’m not opposed to that.

Ah, that was something I hadn’t picked up on–I wasn’t sure why the first time through I had four choices and then I had lots. I did have a much smoother experience when I read the manual.