April 1st Jokes, Part 2: Parodying NFTs

On making statements through jokes, the off-cliff method of learning, and lucky breaks.

Table of Contents

The ineluctable pull of online discourse

When I decided to “stop whining and go do something” about the sorry state of the web, it was (partly) due to seeing how much time fannish folks (me included) were wasting on increasingly less-useful discourse¹. This was honestly a great decision, and I wish I could say I then never discoursed again. Old habits, however, die hard. Even with my newfound worksona (who has thankfully stayed away from fandom wank), I sometimes manage to drag myself into digging way too much into the controversial topic du jour–a pull that is especially hard to resist when that topic sits at the intersection of my special interests.

“This is important. Everyone is wrong on the internet.”²

Historically, the biggest culprit (and the star of this story) has been NFT discourse. At the time, I already wrote a way-too-long article (sorry sorry sorry) about my feelings. The tl;dr: as it often happens with discourse, people on all sides missed part of the picture, or at least flattened some aspects of the problem.³ But while the arguments on the other side of the divide can be frustrating, nothing beats the frustration of your side getting a particular aspect of the issue wrong.

In particular, one argument kept riling me up: the confident assertion that people online wouldn’t spend money on “meaningless JPEGs”. As a fandom person, I simply knew deep in my guts that this was not true⁴. Seeing people continuing to parrot this talking point (and NFTers rightfully shooting it down) was driving me up a wall.

Finding the right cliff to die learn on

While the NFT discourse raged, the Famed FujoVerse™ Holiday (April 1st) was quickly approaching. As I mentioned in part 1, one of my favorite aspects of building whimsical April 1st jokes is using them as an experimental ground for skills I haven’t had the chance to develop yet. At the time, the need to monetize was becoming more and more pressing, so throwing myself onto a rapidly-approaching deadline would be an excellent way to force myself to learn how to handle money online⁵.

One question remained: what could one sell that would delight our audience and make for an “appropriately hilarious” joke? As I asked myself this, the NFT discourse continued popping back into my mind (and Twitter timeline). I don’t remember when I finally connected the two ideas⁶: if I was so pissed off at the current discourse, why not show that people would indeed buy meaningless JPEGs of things they care about? Thus, the idea of selling absolutely meaningless shipping certificates was born, and RobinBoob with it.

While this story might make RobinBoob seem like a unilateral dig at my own side, there is more to it than that. RobinBoob is also a statement about the (typical of NFT spaces) conflation of financial speculation and the desire to collect things you care about. Buying a ship certificate would never make someone rich. However, it bought something way more important: a fun time on the web, and an excuse to connect with your broader community⁷. That is why, with all due “oversimplification warnings”, many in fandom buy (and sometimes sell) things.

The part where engineering

With the premise out of the way, let’s talk a little bit about the technical details. The RobinBoob’s website itself isn’t particularly complicated. Don’t get me wrong: the code is an absolute mess⁸, but that’s more a byproduct of time and knowledge constraints rather than of intrinsic complexity. Long story short, I found a NextJS + Stripe starter on GitHub and got to hacking, learning what I needed to learn as I tried to decipher its inner workings.

A lucky break appears

The hardest part of building RobinBoob was figuring out how to build a list of characters, and how to avoid character duplicates that would make the whole game a lot less fun. The obvious solution was to scrape Archive of Our Own for this data. But while this was overall feasible⁹, we also were very short on time. We could try, but there was no knowing how many characters we could get in that short time.

As I was exploring the various options, however, an incredible coincidence happened: for the first time ever, AO3 released a dump of its own database. This included all character tags, and the information needed to deduplicate them¹⁰.

And so the RobinBoob project was officially a “go”. What can I tell you? Someone out there really loves fujoshi and good jokes.

Woman vs deadline: getting nerdsniped

There isn’t much more to the story (that I remember) aside from a few anecdotes. For example, we found out that the lazy hack carefully-designed method I was using to associate a purchase with character pictures would stop working when someone bought a certificate with 6+ characters. This was a few hours before release. And so the RobinBoob sprint marked the first time I did not sleep on March 31st (a 3-year streak we’re hoping to break in 2024).

With all of this said, there were some aspects of this rush that were overall avoidable: the biggest culprit was that I nerdsniped myself into figuring out default pictures for our character selector.

Eventually I realized I didn’t need to get those images in, and could instead use our classic “historical sexyman” fallback (that is, a picture of the Once-ler). However, this feature did made for some amazing code sleuthing, and I’m still proud of having figured out how to makes it work.

Here’s the step-by-step logic:

1. I used AO3’s data dump to figure out which characters had the most fics (since I knew I couldn’t get to everyone).
2. Having ordered the characters by number of fics, I scraped AO3 to get the name of their fandom (a piece of information that’s not in the database).
3. Once I had the name of the fandom, I scraped the search page on fandom.com¹¹ to find its wiki.
4. I used the MediaWiki API to find pictures of the character on the wiki of their fandom.
5. I had my program upload a few of them to my own storage and pair them with the character in the database.

…and that is all! While the code I ended up with isn’t particularly complex, the hard part was figuring out how this could be done with no idea whether it could be at all. The sleuthing included looking through various image search APIs, then wondering whether MediaWiki had one and finding out that indeed it does.

Eventually, thanks to this fully-automated method, I was able to add some pizzazz to our character search¹². The results aren’t always perfect (and sometimes are completely puzzling), but, hey, it was all worth it for the memes.

See you next week with another April 1st deep dive: FujoBoard!