Structure and Interpretation of Scrandle 📄 posted at 16 August 2025 on new danboland

Scrandle started off as a game I wrote for a friend, which later became SCRAN GAME (I released it on itch under a pseudonym because I was ashamed of BRDA not being done yet). Now it’s a shoddily-written browser game that includes the entire data set in the initial javascript payload.

Since the donations for the next version of Scrandle have been very healthy, I’ve made good progress on Scrandle 2, and looking forward to getting it out there. But let’s have a look at this weird thing, the most popular thing I’ve ever made.

How Scrandle works

Scrandle is information-light by design. It’s supposed to be mysterious and stupid. It’s supposed to put you into a mindscape of uncertainty and doubt. There’s no way that a tub of cooked insects could defeat anything! And yet, why are you so scared? Could it be that even now there are people you don’t know making decisions you didn’t know existed?

The first of an array of misconceptions about the game is that each round represents a poll held somewhere where items were directly competing. This is not the case. How would that even work? There are over 2,000 items in the data set. The amount of comparisons between them is, let’s say, the 1999th triangular number. Do you think the FootyScran account has run 1.999M polls with enough eager voters to provide high-granularity data?

Each item has a poll associated which says “Scran” or “No Scran”. The “Scran” vote percentage is what is being compared. Nothing more.

This is mostly an exercise in data processing and data quality. To this end, I have endeavoured to make sure my scraper and all related tools are as janky as possible, with at least one manual data entry / validation step.

The daily generator tries to create a fairly balanced game, 3 easy matchups, 4 medium, 3 hard – order is shuffled. Over time I feel this has got a little stale, and having to put out X easy matchups a day means that certain items are favoured, giving the sense that the dataset is smaller than it really is. But at the same time, people aren’t really clicking the website to rate a 80% burger against a 85% burger 10 times a day. You want to see your friend Zsíros Kenyér.

The patchnotes on this day read ‘Fixed up a few small bugs’.

Friday Scrandle

Fridays, on the other hand, are themed and curated. The process varies from week to week. Sometimes I filter the dataset along some axis and sample it randomly until I get a good game, othertimes I try to write a function that generates a game a specific way. Sometimes, increasingly lately, I’ve been hand-picking each matchup in the friday.

It’s definitely work, and it’s insufferable treadmill work that I must maintain for however long, but the Fridays are for me. There is something a little depressing about some of these daily games, making a game that just spits out random data and leaving it to players, creatures with a finite amount of time on this earth, to figure out if it’s fun or not. Some well-meaning individual suggested that I could use a diffusion model to generate scran items for the game, and I looked at the stars in the sky, and for a moment I wished I could join them. But on Friday I can inject some personality, force the game to be weird, and be the meddler that I desire to be. I will rub peoples’ noses in how arbitrary and stupid the data is. I will continue to do this no matter how much people complain. The pulse in lompe perfect mirror match is a warning shot.

It’d be fun to make a “Friday Scrandle” style take on a roguelite that uses the (codified, played-out, manipulative – pick 1 of the 3) format to be funny/interesting rather than be an infinite and regular source of anaesthetic, slot-machine highs and progression. If I can figure out how to make it in a way that loves something instead of hating something… maybe.

On making streamer food

It is hard to consider a game more streamer-hostile than BLUE REVOLVER, beyond perhaps one where you rate the dicks of men* while pay-per-view wrestling matches play in the background. STGs require constant, consuming focus for at least 20-25 minutes, have little for most people to talk about on a run-by-run basis, and place a very harsh spotlight on mechanical/tactical skill. My next STG project, which I am also working on don’t worry, is an experiment in squaring this circle without compromising what I like.

On the other hand, I got so PO’ed writing Scrandle and found so little interest trying to post it anonymously to various communities that I just threw it up on a domain, no analytics, and forgot about it. Perhaps a year later, I find a small legion of weird girl streamers have latched onto the game somehow. Huh. Cool. Oh. Now Northernlion’s playing it. Huh. Cool. Joseph Anderson’s playing it. Huh. Cool. AvoidingThePuddle’s playing it. Huh. Cool.

Food preferences often reflect where someone wants to be, not where they are. I know people who would pick worms with a pint of slime before they picked a slice of Dominos pizza.

On the one hand, my life’s work, the convergence of my taste and ideals, the continuing of a spark lit playing Giga Wing, Warning Forever and Perfect Cherry Blossom when I was 14. On the other hand, a game where you click currywurst because, come on dude, currywurst never loses. One flounders harder the more I try to put into it, like fighting quicksand. The other was totally bootstrapped to 100K players from nothing without even thinking about it or lifting a finger. The framed picture of Albert Camus on my desk is smiling at me. (I’m being dramatic here for fun. BR is easily in the top percentiles of indie shmups as sales/revenue goes. Still didn’t make minimum wage on it though!)

You’ll go mad if you try to consider how much “value” big streamers are getting out of your work. Online video is the most monetizable substance on the internet and basically everything can be kind of mashed up and liquefied into it. Can’t really take it personally. Maybe just get your kicks where you can, try to see how you can get an advantage, and be glad it’s people having fun with a game rather than going The Danbo Situation is Crazy or whatever.

Remarkable how far you can get giving people what they need, I suppose. I theorised that the game’s popularity would create a Scrandle Effect for BLUE REVOLVER come the summer sale, and while it’s not a huge bump, it was a pleasant surprise. There might be an uplifting message here about chasing whatever has your interest and trying to join all the dots later. I built this out of my genuine curiosity, and now it at least shows an ad for my Real Work that I would probably have to spend 5 digits on, for as long as the game is popular. That’s positive.

Thanks for the donations

The other positive thing is the response to the Ko-Fi donation link. People have really been happy to toss over some money for server costs and development time, with maybe some food/rent/beer/Perfect Grade Unleashed Nu Gundam money on the side.

There have been some very high rollers! I’m really quite stunned, and very grateful. (I have stared at this sentence for a half hour trying to think of how to embellish this sentiment. Can’t do it. All I can think of to say is “hey thanks”, but I do mean it).

I am considering an extra mode for Ko-Fi supporters but haven’t decided on exactly how it’ll work yet. I feel a very strong urge to give donators something back, but don’t want to pollute the game with tons of paywalls or account login guff or whatever.

You didn’t hear it from me, but when the current Ko-Fi goal is reached, I will make at least one zsíros kenyér and eat it.

Scrandle 2 next steps

Stuff I talk about here isn’t public yet, but hopefully will be fairly soon.

I’ve totally rewritten the client for easy development of new gimmicks and one-offs. It looks and feels slightly different. Evil Mode is considerably more evil.

I’ve done a new scran scrape, trying to get absolutely everything I can to really shoot for the biggest data set possible. There are some really funny items here that I’m keen to see in the wild. There are So Many Cheesy Chips.

I’m building a dynamic scran rating system – scran without twitter polls can be rated based on the decisions of Scrandle players. This opens the game up for user submissions and all sorts of new data sources. While the game is probably always going to be based around stadium/arena food, I’ve no problem running whatever I like on the fridays.

I’m maybe a little suspicious of the dynamic ratings – will they be too boring, since Scrandle is a game with a confirmable “meta”, operating on its own data? You can’t blame random English footy fans for the ratings anymore. I have a feeling there’ll be enough noise and abstraction in the system that the data won’t be too predictable, and the many annoying emails I get confirm that the playerbase is broader and more diverse than I may have thought.

I think it’ll be cool. I do want to see more world scran, though. Mainland Chinese scran etc. I have no idea what those guys are eating at the footy.