MaG48HMML Developer Diary #1: A 48 Hour Detour

I have a history of letting small commitments snowball into massive undertakings. OH JOES! was a 2-month project that took 2 years. My YouTube playthroughs of the NES Mega Man games were a 3-week project that took 3 years. More recently, Make a Good 48 Hour Mega Man Level (MaG48MML for short) was a 2-day project that has lasted over 3 years and counting.

Logo by MiniMacro and rc, with input from gone-sovereign and eviemaybe.

MaG48HMML (48H for short) was conceived as a shakedown cruise for Megamix Engine, the game engine that would be the basis for the larger and more involved Make a Good Mega Man Level 3 (MaGMML3 for short). The plan was to hold a small level design contest during the summer of 2018, get the levels judged quickly, and crank out a simple game showcasing those levels. When submissions for MaGMML3 opened in the fall, the judge feedback and technical improvements from 48H would help participants make better levels. I was already a judge and devteam member for MaGMML3; I had no time or intention to get involved with 48H beyond submitting a level.

The main screen of the 48H judge EXE, circa February 2020. Relatable.

However, those plans fell apart. 48H attracted more contestants than MaGMML1 and 2 combined, the levels took years to judge (well overshooting the submission period for MaGMML3), and the final game was hardly simple. I became massively involved: I wrote ~75% of the non-cutscene dialogue; designed a weapon tutorial and a section of the fortress; provided extensive playtesting and proofreading; fixed (or attempted to fix) numerous glitches; made corrections or modifications to several art assets; programmed more things than I thought I was capable of, including what is arguably the biggest Easter egg of the game; conceived or helped plan countless major and minor aspects of the game; wrote copy, edited judge bios, commissioned artwork, and hosted downloads for the website; and helped revive the project when it was nearly dead—twice.

Also, I submitted a level.

Several weeks before the contest opened, I discovered that I had been added to the 48H devteam chat on Discord. I hadn't requested a devteam role, no one had asked me whether I was interested, and I knew basically nothing about the project. I didn't even have an official role; "any help you wanna offer would be appreciated" was the direction I got. I now appreciate what it's like to be a summoned monster in a Final Fantasy game. As long as I didn't have to do anything, I had no problem being on the devteam to discuss ideas. I spent enough time on Discord already; what difference would one more channel make?

What, indeed.

The rapid-fire conversations that transpired over the next month were riotously funny, and they shaped the foundation of the game. Almost nothing had been planned at that point, aside from the premise of giving contestants mystery boxes full of randomized assets with which to build a level in 48 hours, and we had a field day shouting out ridiculous ideas that somehow stuck. Of course, several ideas didn't stick (aimable Spark Shock, a Wily Assist weapon that shoots out a flustered-looking Dr. Wily at a 90-degree angle, GIANT CARDBOARD BOX WILY MACHINE - FOR PHASE 2 HE UNFOLDS IT), but the ones that did stick were agreed upon quickly and amicably.

Mega Man wins a free cruise, but it's secretly a distraction so Dr. Wily can do evil deeds without Mega Man coming after him! (Unanimous agreement.) Let's name the ship the Elroy, after MaGMML community member Cruise Elroy! (Unanimous agreement.) Actually, forget Wily; the villain should be box-themed! (Dissenting suggestions of Crystal Man from Mega Man 5 embezzling money from Wily, then Clock Men from Challenger From the Future, then these characters as part of a series of fakeouts leading to the real villain). The single absolute worst level should be on some dingy old raft floating behind the cruise liner, circled by sharks! (This elicited a "lmao".) Even in their rawest state, our ideas weren't too far off from their final form.

Sarge, the weapon tutorial host, evolved from Super Ball Machine Sr., a British high-society type. Sprites by Protty, circa February 2020.

Before long, we started discussing the contest rules. As a judge for MaGMML3, a soon-to-be contestant for 48H, and a professional editor, I felt a strong obligation to ensure the rules were as clear and comprehensive as possible. I volunteered to copyedit the rules document, which soon turned into me taking the lead on incorporating feedback and finalizing the rules. Once that was squared away, the contest opened and I got to work on my level. Regrettably, I forgot to pencil in a rule mandating that I win the contest, but that's hindsight for you.

To make the most of my 48 hours, I requested my mystery box on a Friday night after dinner. I had nothing planned for the weekend, I was in good health and good spirits, and I had a few friends on standby for playtesting. Conditions were perfect. My plan was to stick with the first box I was given (unless it was utterly unusable), wait to see what I got before starting work on the level (I felt like planning the architecture and aesthetics in advance was against the spirit of the contest), and craft a "normal" level that might feel at home in an official Mega Man game (not too long, not too wacky).

Using the "!box" command, I summoned this bot to give me what ended up being a very good mystery box.

It was more like a 30-hour level, really. I slept at least 6 hours a night, lost 20-30 minutes every time I compiled a test build, was waylaid by a series of asset issues (eg, monkeys falling through floors) that required devteam intervention, and spent entirely too long futzing with a laser gimmick that was long on possibilities and short on documentation. I also didn't count on how long I'd spend trying to get Rain Flush to block lasers (to no avail); getting the volume balance and loop point right on the music; and fine-tuning so many precision-oriented puzzles that were fine on paper but messy in practice, due to the nature of the assets involved. Fortunately, the nature-themed enemies and military-themed gimmicks made it easy to decide on an overall visual theme; and it took no time at all to pick Dyna Man as the boss, given how well his vaguely military aesthetic, incessant hopping, and use of explosives fit with the rest of the level.

There's plenty more I could say here, but in light of everything else, my participation as a contestant is almost insignificant. Those 48 hours were fun and memorable for me, and I'm proud of what I accomplished in that time, but my level is 3 minutes of entertainment in a game that takes upward of 20 hours to complete. My level came in 18th place, neither anyone's favorite nor anyone's least favorite, not sufficiently noteworthy or contentious for people to ever bring it up in conversation—except to comment on the music. The real legacy of "Base in the Boondocks" is bringing "Portrait of a Ghost Ship" from Castlevania: Rondo of Blood to the 48H jukebox and inspiring one question for the music trivia sidequest.

But hey, if you like using special weapons, my level is kind of a playground.

After submitting my level, I stuck around the Discord server for another 8 months. I provided advice and playtesting for other entrants during the rest of the entry period, and I continued to weigh in on devteam matters, but planning discussions soon began to taper off. By the end of March 2019, I was so busy judging for MaGMML3 and so little was happening with 48H that I decided to leave the server, with a standing offer to return if anyone ever needed me for anything specific.

Ten months later, I reached out to ParmaJon, the contest host, to see if I could help out with anything for 48H—hub design, NPC dialogue, etc. I was nearly finished with judging for MaGMML3, so I figured I could spare some time to help push 48H along to the finish line. The answer was yes, so I rejoined the server and started skimming through devchat to get caught up.

48H was in almost exactly the same place I had left it.

Granted, one judge had almost finished judging, and another was relatively close behind. Planning had begun on the weapon tutorials, some important technical stuff had happened with the game engine, the map function was being developed, and a couple tier hubs were in the preliminary stage of design. There had been some changes to the composition of the devteam, too. But we're talking about a game whose main content was thrown together 18 months ago in the span of two days, and we still didn't have screenshots to put on the website. From the perspective of anyone on the outside, 48H was dead in the water.

Preliminary Tier 1 mockup by Protty, circa October 2019.

Perhaps more accurately, 48H was rudderless and functioning with a skeleton crew. ParmaJon was a judge as well as the contest host, and judging was drawing focus from leading the project. Any planning that had been done was scattered across multiple Discord conversations and Google Docs. Several devteam members were simultaneously working on MaGMML: Episode Zero, which was nearing completion, and/or MaGMML3, which was still very much in progress. In short, hardly anyone was available to work on the game, and hardly anyone knew what to work on in the first place.

With ParmaJon's blessing, I started developing a master design document, pulling together everything that had already been planned or discussed and filling in any gaps with my own suggestions. (I've saved a copy here, if you're interested.) I also suggested that we designate someone to make day-to-day decisions in ParmaJon's stead. It took two weeks for us to iron out the design document and less than a day to get CSketch on board as game design lead.

Suddenly, 48H was back on course. Devchat was more active than it had been since the contest closed, and the project felt exciting and relevant again. Although a few of my favorite ideas from the design document didn't make it into the final game (namely, Mega Man's cabin being like the Sky Room from Command Mission, and BomBoy from the Battle Network games running a mystery box shop), I wasn't precious about my suggestions. The purpose of the design document was to organize past discussions, spark new discussions, and give us concrete tasks to work on, and that's exactly what it did.

Preliminary Tier 2 mockup by Protty, circa October 2019.

Around that same time, I got a copy of the judge EXE and played through all the entry levels. If I was going to be involved in planning the game, it behooved me to know what the bulk of the game was going to be like.

It was going to be terrible.

In their submitted state, without Skip Teleporters or Beat or anything else that the final game offers to mitigate the difficulty, the entry levels were torturous. I could count on two, maybe three hands the number of levels I genuinely enjoyed. 48H's tight time limit amplified some of the pitfalls of amateur (and even professional) level design that annoy me most, and it didn't help that I was already pretty burned out on Mega Man after judging 170-odd levels for MaGMML3. Consequently, I was very vocal in devchat about introducing features that would sand off the rough edges and balance the difficulty of the entry levels.

One such feature, which we were already working on, was a series of weapon tutorials. What better way to give players an advantage than to help them fully understand the tools at their disposal? I called dibs on the tutorial for Homing Sniper—I had some ideas about how to elevate the training to something more than "mash the attack button to win"—and I made it a point to playtest all of the other tutorials as well. Aside from participating in all the usual devteam discussions (which now concerned the Box Cartel fights, possible postgame content, and whether to let players unlock extra checkpoints), my focus was entirely on the weapon tutorials during this period; everything else seemed to be under control and moving smoothly.

Concept art of Blocky and Square Machine by ACESpark, circa September 2020.

Once again, I decided to hang around the server for about 8 months (until mid-October 2020) before taking my leave. All the weapon tutorials were finished or just required finishing touches, and there wasn't really anything else I could help with. I was trying to cut back on Discord use and focus back on MaGMML3 as well. As before, I left an open invitation to call me back to the devteam if I could help with anything else in the future.

When I checked in with ParmaJon and CSketch a month later, it sounded like the game was almost finished, aside from judging. The devteam really only needed help with graphics and programming—two things that were best left to the experts. The only thing left that suited my skill set was NPC dialogue. It was still too early to start writing anything, but I could at least do some prep work.

Knowing that most of the NPCs were going to be regular stage enemies, I set up a sprawling spreadsheet so we could keep track of which NPCs were available, where in the game we wanted to put them, whether we needed to create custom sprites for them, and any notes or suggestions. By the time I was done trawling through the enemy/miniboss lists of every Mega Man game we were likely to pull from, there were well over 500 NPCs on the list, not to mention Duo, Reggae, and several other named characters who might show up. I even included Daidine, a spinning platform from Mega Man 5, because "The fact that this spinning platform has an actual name makes me want it to be able to talk".

Another month went by before I formally rejoined the team—not to write NPC dialogue, but to help design a section of the fortress at the end of the game. It's MaGMML tradition for each judge to make their own fortress level, but only ParmaJon was in a position to do so, and in a limited capacity at that. It fell to CSketch to design the first fortress level and co-design the second with ParmaJon. The third and final level was an independently collaborative effort involving the entire devteam, meaning that we all got to create our segments in isolation before smashing them together.

My segment was originally going to contain these assets. I glued two Beaks together so they'd fit in with the rest of the orbs.

To start creating our segments, we selected one of the unused mystery boxes from the entry period—I picked mine at random from the master spreadsheet, but others chose the specific one they wanted. As development progressed, I started noticing a fair amount of overlap in the assets we all were using for our segments. In the interest of showcasing a wider variety of assets (and giving players a reprieve from too many segments where they were required to shoot the terrain), I set up a spreadsheet so we could keep track of our progress and which assets were used in which segments. I threw out my box and rerolled until I got something novel that I could work with.

First reroll: "Your choice of a destructible block". Pfft. I was trying to avoid shooting the terrain. Next.

Second reroll: Venus Waterfall Spawner. Ah yes. The janky gimmick that had become a running joke, on account of it showing up in an alarmingly high number of boxes and being the bane of the programmers' existence. NEXT.

Third reroll: Wave Man Jet Ski, Cricket, Bikky Bomb, Tamp, Nombrellan. I could work with this. And so I did.

I modified existing art assets for my sublevel. If I don't use my custom slope tiles from MaGMML3, who will?

Development of our so-called Final Box levels was slow but steady. I provided as much feedback on everyone else's levels as possible, going so far as to fire up Mega Man V for Game Boy and thoroughly examine the physics of the bubble floor gimmick, which was implemented differently in Megamix and causing playability issues. By the beginning of February 2021, my Final Box level was finalized. Two months later, I hadn't started any new projects, and I was having second thoughts about sticking around for that other thing I had offered to do.

"If we need dialogue for any of the weapon tutorials, I can help with that," I wrote. "If we reeeeeally need help with other NPC dialogue, I'll consider it, but I had previously mentioned the possibility of un-volunteering myself if we had enough people to cover everything. I'm not as confident in my ability to write snappy dialogue for random NPCs, and it's been very challenging lately to work up the energy to get back to my creative projects. But if it's the difference between the game launching on time and the game dragging on even longer, I can give it a shot."

After a little bit of discussion with the devteam and with my wife, I was persuaded to try my hand at NPC dialogue after all. Nothing had been written outside of the weapon tutorials, so I had my pick of every tier in the game. I started with Tier 5, the minigolf course—I enjoy minigolf and had been watching a lot of Holey Moley at the time, so that was the natural first choice. It took me all of four days to pick out and arrange the NPCs, write dialogue for them, and implement their sprites.

I write only the most sophisticated dialogue.

A little over a week later, I was ready to mute the 48H server and get back to working on MaGMML3 unless someone pinged me for something. Any last requests before I disappear again? Well...okay. I guess I can write a little more dialogue.

I gave the devteam a few options and let them decide which tier I should do next. They chose Tier 3, the dining hall, which was perfect because I love talking about food. This one took longer—slightly under a month—in part because I got a little more ambitious and asked for both programming help (for the Hologran gag) and custom NPC sprites for a few enemies from Mega Man 7 and 11 who needed to be redrawn in an 8-bit style.

I use only the most sophisticated placeholder sprites.

Afterward, I announced that I had it in me to do one more tier. I went with Tier 9, the ice skating rink. I love ice and snow aesthetics; figure skating is one of my favorite sports to watch; and I thought it'd be good to break up my tier claims so that if someone didn't like my writing style, they weren't stuck with me for multiple tiers in a row. I started work toward the end of May but didn't finish until early July, due to life craziness and some unexpected graphical and programming needs.

For example, correcting a mistake that turned background decoration into foreground footwear. Oops.

As my work on Tier 9 was nearing completion, I realized I had finally hit my writing stride. I could probably manage one more tier, if not a few more tiers; despite how exhausted I was from multiple multi-year Mega Man projects, I was having fun. However, I wanted to balance "helping finish the game" with "hogging all the fun stuff," so I asked the devteam about who else actually had an inclination to write NPC dialogue. Half a dozen people were interested, but burnout and busyness were very apparent. Spade_Magnes was working on Tier 6 (ballroom) and had finished Tier 2 (cargo hold) after picking it up from snoruntpyro. The main deck and passenger cabins were at least partially reserved for CSketch and snoruntpyro. Otherwise, everything was up for grabs, and it really didn't matter who took what. We all just wanted to get the game out the door.

Spin Attaxx claimed Tier 10 (water park), and I put Tier 8 (library) and Tier 4 (engine room) in the queue for myself. My plan was to keep churning out NPCs until we ran out of tiers or someone told me to stop, whichever happened first. Tier 8 took me only three or four days, and I knocked out Tier 4 in a single weekend. I was bolstered by positive feedback from the team; they seemed to like what I was doing with the NPCs, and any critiques were basically always constructive and beneficial. This part of development was easily my favorite: the project had a sense of momentum; I was fully in my comfort zone; and I felt good about what I was doing, because of both personal satisfaction with my work and recognition from the devteam.

My next claim was Tier 1 (lifeboats), which was a one-day project. As with all the tiers, however, I would go back later to finesse the placement and creation code of the NPCs and polish the dialogue. Tier 1 was attached to the main hub, which was being updated frequently for one reason or another, so I had to make sure no one else was working on that Room file in GameMaker Studio when I was. The last thing I wanted was having my local changes overwritten in a dreaded merge conflict when pushing my work back to the master file, so I suggested we implement a system where anyone who wanted to work on the main hub had to announce it first, in case anyone had unpushed changes. This system mostly worked.

At the very least, I never encountered a merge conflict with myself.

Now that I was mucking about in the main hub, I started thinking about what players' first impressions of 48H would be. After the intro cutscene, you step onto the deck of a gigantic cruise ship and have free reign to explore. Where do you start? Is the ship easy to navigate? What areas might you overlook? Player experience had already been on my mind (see: the Giant Telly in Tier 3 who checks in on your emotional state after playing "Megatroid", the Puyoyon in Tier 4 who subtly reminds you to visit the costume shop on your way out, all the NPCs in Tier 1 who try to prepare you for the disasters you're about to deal with), but this is where I started taking on focused projects to make the game more player-friendly.

We had discussed adding some info-oriented NPCs, so I created Iota to explain what 48H was all about, a Jamacy in Tier 2 to warn players about a level with broken ladders, and multiple NPCs to guide players around the ship. I created a Junk Golem NPC to provide hints about where to find new sidequests and how to complete them. To make the main deck less overwhelming to new players, I followed through on a plan to lock some of the cabin doors until later in the game.

Because ParmaJon didn't want a traditional shop, players needed ways to stock up on E/W/M-Tanks that didn't involve grinding in the entry levels. I added an emergency M-Tank dispenser to ensure players wouldn't get stuck on the Box Cartel fights, when the entry levels are blocked off. I also created a Mad Grinder who would supply free E-Tanks and W-Tanks based on sidequest progress and Energy Element collection, respectively. It's all too easy to chug E-Tanks instead of applying actual strategy, so I wanted to encourage players to try special weapons, pursue sidequests (and their rewards), and not get hung up on any one level in the early game, and then to shower them with tanks in time for the endgame and postgame.

So much for this early attempt at programming the Mad Grinder. Math is hard. Let's go shopping.

Furthermore, I suggested we reinforce Joseph's explanation of Junk by stationing an NPC outside his cabin who would give you Junk to trade in. I also argued strongly against starting the game with a mandatory scavenger hunt to find Beat before accessing the levels; to me, that felt like a contrived and needlessly restrictive way of ensuring players would explore the ship and obtain a special weapon that the levels weren't designed around. Although I was overruled both times, I found other ways to influence the player experience. For the Beat search in particular, I moved the invisible barriers to more organic locations, and I revised the generic "I shouldn't go that way" message to a fairly blunt reminder/hint about what you should be doing.

I was more successful about suggesting changes to the sidequests, which were being developed relatively quietly—I didn't know they existed until May 2021, and there weren't any real opportunities to get involved with them until July. After I tried out the music quiz for Funky Fresh Beats, I pushed to include a screenshot with each question, to make the sidequest accessible to deaf players and to anyone playing with the volume off. Because screenshots were too much of a hassle to implement, we agreed on text hints, which I wrote. I also made a few suggestions that stuck for the Mutual Attraction and Poltergeist quests; for the latter, I even recolored the ghost sprite (being obligated to use a canonical color for the Rotom Pokémon it was based on) so that it wouldn't blend in with the background as much.

Storyboard for the Acolyte Joe sidequest cutscene by ACESpark, circa August 2021.

Love Survivor was the sidequest I influenced most. For a very long time, the cruise ship was going to have a casino, either as the location for Tier 7 or as a standalone area (in the space that the Kickboxing Club now occupies) with minigames to play. The original premise for sidequest, then dubbed "Red or Yellow", was to help fashion boutique co-proprietor Bol'o raise money to pay off an enormous debt to a mafioso called Don Loath (a tip of the hat to Lex Loath from The Misadventures of Tron Bonne). Doing so would require cheating at a casino game where the objective was to guess whether a light would turn red or yellow (an inside joke apparently inspired by a Vinesauce video poking fun at a game called Color Fun).

The casino was scrapped in January 2021 to reduce the devteam's workload, and it was several months before we discussed relocating and redesigning the sidequest. Tier 3 (dining hall) had some unused space that was inaccessible to the player, which could easily be remodeled to accommodate a quest. This inspired me to outline a new quest, which made it into the final game with only one change: Don Loath (who never even received a character design) became Master Reddorgold (whose name is not a reference to "Reddit Gold", but rather a nod to the sidequest's original title). A kitchen stealth mission wasn't my only suggestion, though; I also pitched a Donkey Kong Country–inspired level using dining carts in place of mine carts.

My mockup for the revamped Red or Yellow (now Love Survivor) sidequest, circa July 2021.

A week later, I formally committed to writing all remaining NPC dialogue in the game. I was the only person adding to the main hub, several passenger cabins were still vacant with nothing planned, and nobody else had claimed Tier 7 (art gallery) or Tier 11 (sky deck). I was least enthusiastic about Tier 7 and had been hoping someone else would take it, though. The tileset contained sculptures and paintings (by eviemaybe and Protty) that I wanted to acknowledge in the dialogue—except half of them referenced things I didn't recognize, and by that point it was taking all my brainpower to write what I knew, let alone what I didn't know. And there was a major deadline approaching: first test build of the whole game for our internal playtesters.

I had about one week to power through Tier 7 and add as many NPCs as possible to the main hub and cabins. Nothing said I had to rush, but the longer it took me to get my work loaded into a test build, the less likely it was that people would see it. This was especially relevant for the deck and cabin NPCs whose purpose was to guide the player. Tier 11 was still in the process of being tiled, so I didn't even bother with that one yet. I'm not entirely satisfied with the dialogue I produced during this time, but "not entirely satisfied" was to become my mantra for the duration of the crunch leading up to the game's initial public release.

One of the few bits of dialogue that I discarded wholesale after running it by the rest of the devteam. This joke (conflating a famous sculptor and a famous kaiju) was too layered and esoteric for its own good.

By the beginning of August 2021, new playtesting builds were being released about once or twice a week (soon to become once a day), a release trailer was in development, and I was rapidly running out of time to finish everything on my to-do list. The team wanted to ensure a summer release, given the cruise ship theme. Summer, as I pointed out, would last until September 22. They wanted it by the end of August—when I would be unavailable for several days. Eventually we settled on end of August as our internal deadline, with a few days of buffer before the release date of September 4 that was ultimately announced to the public.

My eagerness to get this project out the door had finally caught up with me. Leading up to the crunch, I kept taking on small assignments to ease the burden of more specialized team members. The artists were busy, so I cleaned up the subtle color inconsistencies between the special weapon icons, and then loaded them into a tileset so I could add them above the weapon tutorial teleporters. The programmers were busy, so as much as possible, I tried to figure out how to code complex NPC behavior for myself. I was already spending all my free time on 48H before the deadline was announced; the only way I could crunch any harder was to start sacrificing sleep, aspirations, and quality—and I won't even go into the Big Life Stuff that started vying for my time. The weeks leading up to release were physically, mentally, and emotionally brutal, and they turned my experience with 48H extremely sour.

My key mistake was not communicating to the team just how much I actually had planned for the initial release. I wanted to bring my perspectives as a professional editor, MaGMML3 judge, and fangame designer to the parts of the game I hadn't yet seen. I kept coming back to seemingly finished projects to polish things up and incorporate playtester feedback. I was still working on NPCs for the main hub and cabins, but because I had added any NPCs in the first place, people assumed I was done already. I hadn't even started on Tier 11, which would end up taking over two weeks to finish. And that's to say nothing of all the Easter eggs I was working on, or the Butt Mode cheat that I got involved with at the last minute.

Early drafts of the Butt Mode script ended up mangling the text in unexpected ways. I ended up spending a lot of time polishing butts.

As we entered September, mere days away from release, show-stopping technical issues began to arise. A key devteam member's computer died at an incredibly inopportune time. We kept discovering game-breaking issues and applying fixes that we didn't have time to properly playtest. The only person who could set up the online leaderboards on our usual server was unexpectedly unavailable. On top of that, a couple sidequests and noteworthy cheats were still being developed, and I still had a lengthy to-do list. No way was this game getting done on time.

By September 1, the devteam was discussing the possibility of delaying the release. By September 3—late enough in the day that it was already release day in certain time zones—we came to an agreement that a delay was necessary; the question was, how long? A few devteam members—myself included—voted for "until it's ready", but the final decision was to not keep the public waiting, saving any nonessentials (which included most of my to-do list) for an eventual patch. Release was pushed from September 4 to September 6, just long enough to lock down the most pressing technical issues.

As the guy in charge of the MaGMML website, I did as much as I could to prepare in advance for release day, such as gradually adding screenshots to what would become the download page. As soon as the judge scores were finalized, I commissioned Phusion—who did website art for MaGMML1, 2, and 24H as well as intro cutscene art for OH JOES! and MaGMML1 Remastered—to once again create art celebrating the first-place level for the download page. Once the story was fully locked down and we knew exactly what features the game would have, I wrote copy for the download page for the devteam to review. By release day, all we really needed were download links and one final sanity check before going live.

Art for the 48H download page by Phusion, celebrating Mystic Museum, the first-place level.

Release day was hectic and exhausting, owing to miscellaneous setbacks and delays. After the download page went live, I waited just long enough to confirm that there were no immediate issues with the download links or the game itself, and then I dropped off the face of the planet for the rest of the day. While other devteam members were celebrating, I was taking all the time I could to decompress and recharge before getting back to work. For me, the crunch wasn't over.

Given the general public's tendency to quickly find ways to break a brand-new game, I correctly guessed that we'd need to release a patch to 1.01 almost immediately, and then another small patch to 1.02 within a few days. If I acted quickly enough, anything I had planned for initial release could still be a part of people's first experiences with the game. After that, there was no telling how often we'd release additional patches, if at all, so it behooved me to get everything done ASAP.

While I labored away at the rest of my to-do list, I checked in regularly to see what people in the MaGMML Discord server were saying about 48H. Compared with the initial public response to OH JOES!, it was a joy to discover that all my time and effort had apparently paid off. People were sharing screenshots of NPCs I made that they liked, posting video playthroughs containing positive feedback about my design contributions, and talking favorably about the game as a whole...at least, until they started discovering the sidequests.

"Where is Tomothy Daddy?" "What other level is spicy?" These kinds of questions started dominating the chat, and I had no idea what they meant. When I wrote the Junk Golem's sidequest hints, I often had to go off of secondhand information from the devteam and an outdated sidequest planning document; I had yet to play many of the quests for myself. A few days after release, I started playing through all 16 sidequests, and suddenly my priorities shifted. After playing Credit Where It's Due, a tricky scavenger hunt, it was clear to me that players needed more hints, better hints, and reminders of critical information that was given once and never repeated—and they needed them now.

Unbeknownst to me, the day I played Credit Where It's Due was the same day version 1.02 was slated to be released. After outlining my proposed changes to the devteam, I scrambled to implement them—the team wasn't expecting to address any scavenger hunt complaints until version 1.1, but they were happy to include my edits in this patch if I could finish in the next hour or two. The dialogue that I hammered out is serviceable, but there's minimal personality and no humor to it. The patch went live shortly after I pushed my changes, and I haven't heard a single question about Tomothy Daddy since.

I dunno; maybe this IS funny. The word "drat" kinda makes me giggle a little.

Most of the team basically went on break after version 1.02. I kept crunching. I devoted almost every free minute of an entire weekend to Goody Two Shoes, a riddle-based scavenger hunt, first playing through it (which was exhausting) and then planning and implementing a complex network of hints and reminders about where to go next (which was beyond exhausting). I had to dig into the code to understand how each step of the quest was triggered, figure out which NPCs would supply hints, write appropriate hints for each step, program the hints to only show up at the right time, and then test all of that. I also tidied the existing dialogue for the quest and fixed up the black splotches that lead you to your destination—in "Running Down a Drain", for example, the trail inexplicably went cold for several screens, as though you overshot the target.

I pushed my changes on September 19, just over a week after 1.02 was released. Although I loosened my pace a little bit, I stayed focused on getting through the rest of my to-do list as fast as reasonably possible. I scrapped a few ideas that I was now too tired to bother with, touched up the weapon tutorials, added new NPCs and tinkered with old ones, implemented cabin numbers to help players navigate the main hub, and refined/expanded the Butt Mode script.

The farther we got from the release of 1.02, the more the 48H devteam server felt like a ghost ship. People were burned out, busy with other things, or no longer checking in because their obligations to the project had been fulfilled. A few others were working on updates alongside me, but by the end of November, it started to feel like the project was adrift again. I prodded CSketch for a time frame until the next patch, and that set into motion a flurry of activity, which ultimately led to me taking the lead on a couple rounds of targeted playtesting and getting the updated game ready for public release. Version 1.1 went live on December 18, 2021—which means we had been sitting on my changes to Goody Two Shoes for three months. I haven't heard a single question about any of the riddles since.

If you look back through my social media posts, you'll notice I didn't promote 48H until version 1.1 was released. As far as I'm concerned, 48H wasn't done until 1.1. I wasn't about to encourage the general public to play a game I was still actively helping to develop and beta test. It bothers me deeply that most people have experienced a version that was rushed and incomplete, and that the earliest iterations are the ones immortalized for future generations on YouTube and Twitch.

Tweet tweet, game's complete.

Now, though? I recommend 48H whenever there's an opportunity. I'm proud to share this game with others. I'm proud of the entire devteam for making this game happen in the first place, let alone making it look so professional and polished; proud of myself for the quantity, variety, and quality of work I contributed; and proud of my fellow entrants for pulling off over a hundred MaGMML-worthy levels within (or just barely over) the 48-hour time limit, even if I'm prone to complaining about them. I enjoy seeing people enjoy 48H.

As of this post, devteam discussions are infrequent and rarely pertain to the game itself, chatter about the game seems to have died down on Discord and YouTube, and bug reports have all but ceased. Effectively, this ship has sailed. Of course, no game is ever truly done, least of all one that can easily be patched. In fact, a modest number of changes have already been pushed to the GitHub repository in anticipation of another patch. I have one or two more Developer Diaries to follow this one, some updates I might like to add to the MaGMML wiki, and plans to start livestreaming the game once I can commit to a semi-regular streaming schedule. It's been a long voyage, and I expect it'll be awhile longer before I return to shore.