2018-Sep-17: Big Bad Game-a-thon 2018

Well it happened again. Just like last year, I was staff for Big Bad Game-a-thon and it was a blast. In fact, it pretty much followed the same formula as last year: an amazing time marred by one technical hiccup in the first night.

I was involved a bit earlier in the process this time: I did promotion, run selection, developed the scripts I used last year into more proper tools this year. It was neat to be there to see a marathon begin to come together before the actual marathon day. Honestly it felt strange. I mean, I've had entire multi-billion dollar infrastructures on my back professionally, but for some reason as I was doing run vetting it felt like if I made a mistake it'd just singlehandedly ruin everything. Quite a different feeling. However the truth was we had ten people vetting runs, and a consensus among us on what got in, so it wasn't just me.

I did my usual night time host shift, as well as some helping mod, hype chat up, etc. I also had four runs for this one which I admit may have been a little too much. Still though, I feel like I did well enough and everyone seemed to really like what I put on, so I'm not going to complain. Even if I accidentally laid down a pretty terrible joke during my Hachiemon run, heh:

Looking back at last year's post, I marveled that our event hit 200-400 viewers when we were expecting 50. This year we got on the Twitch front page, and got write-ups from big gaming sites. We started the event at 1,600 viewers: four times our cap from last year. That number grew all weekend eventually capping out at 6,200 on Friday night. I was just floored, and so was the rest of the staff. Despite reaching viewing counts fifteen times our max from last year, chat was chill and we barely had to moderate at all. It was really surreal and really welcome too. I had 3,500 for my runs, which I believe is the biggest event I've run in, barely topping SNES Superstars at 2,900.

In truth I felt like I did less this year than last year. I didn't suddenly produce a solution to a tech problem that saved the event, I didn't cut highlights within 10 minutes of runs ending. I mostly hosted, modded, and was there. Still though, by the end I was a weird combination of drained and pumped; it's left me on a high that's lasted at least for all of today. Better yet, I supposed my support was noted because the core staff promoted me. I'm one of the five official marathon/channel admins now. What do I plan to do with this? I don't know yet. I want to arrange some mid-year events, and have some ideas, but they're in the works. I'll need to escape crunch time at work to implement any of them.

Twitch VODs are done and here. Youtube VODs are coming but those take time to process.

2018-Sep-07: OStatus, ActivityPub, Mastodon, one year later

Over a year ago now I wrapped up an experiment in using GNU Social to interface with the Fediverse of Mastodon instances and concluded that if you wanted in on that party, you needed to be running (or join) a Mastodon instance. I ended up closing down my GNU Social instance due to a combination of concern at its lack of respect for Mastodon privacy settings, and social pressure from Mastodon instance admins over the same. I moved to a friend's Mastodon instance, and generally put the whole Fediverse thing on the back-burner.

Recently though Twitter's been making some bad decisions. I mean some real bad decisions. I'd grown interest in returning to the Fediverse as my main social watering hole, and apparently so did a lot of people. I was one of the first to return, a few days ahead of the giant wave that'd follow when the API changes broke most Twitter third party clients. That gave me time to settle in, look around, and discover what had changed.

I still wanted to run my own instance, but Mastodon had moved primarily to using ActivityPub for its federation protocol and, while it would still fall back to OStatus when it had to, I was right in my prior predictions that the feature split would continue to push GNU Social out of the Fediverse. Certain post types refuse to federate over OStatus at all and GNU Social instances are typically seen socially as troublesome and a lot of instances block them outright for privacy and security reasons. To be clear: I applaud this; in my prior write-up I had noted that Mastodon needed to go another way with its federation protocol, this is certainly "another way".

It would be unfair of me to not note that apparently ActivityPub support is coming for GNU Social but ehhhh... I don't have much faith that we'll see it; and if we do, I don't have much faith they'll support the full feature set to make them a first class player in the new Fediverse.

In any event, I wasn't about to set up Mastodon just for myself. My primary criticisms at the time were it was complicated to set up (especially if you don't like Docker, like myself), complicated to administrate, bloated, resource-heavy, and utilized way too many technologies to have coherent development direction. If anything those criticisms have grown stronger since Mastodon 2.0. It gets the job done and I'm glad it exists, but "Bloated mess" is just the start of the terms I'd use to describe it on a technical level. I thought I was stuck using friend's instances until I discovered Pleroma.

Pleroma is a sort of neo-GNU Social. It's written in Elixir, which is a sort of Erlang VM "thing" much like Jython is a Java VM "thing". It's built to be fast, lightweight, and simple to use and administrate. Pleroma's goal is mainly small instances: single users, small peer groups, etc. So in other words, perfect for my purposes. I resurrected social.voidfox.com (my first mistake-- maybe more on that later) and threw Pleroma on it. Setup was a little dicey because it's a very new project in the "Documentation happens to other people" stage, but I did okay over a day or so.

So how did things go? Well, there's some social concerns. Pleroma has a lot of missions and POVs that GNU Social does/did, and aims to have at least some kind of out-of-box parity with GNU Social. As a result, a lot of people who are into the Fediverse enough to know of its existence are wary of it. A lot of questionable instances seem to run Pleroma just because it's simpler. There's definitely a bit of social credit debt incurred just by running it, but not enough to lead to immediate instance blocks, and largely the debt is lifting as more and more instances run Pleroma and don't.... poop in the Fediverse pool?

Aside from that, Pleroma, once properly configured, is a 1:1 feature match for all of Mastodon's core features, which is more than I can say for GS. Pleroma even offers a few features I see a lot of people wishing Mastodon offered, like the ability to auto-unfurl CWs. For my uses it's just about perfect.

As for my return to the Fediverse: I moved to my new Pleroma instance maybe 24 hours before the Twitter exodus wave hit and most instances, including my former home instance, strained under the load. A lot of instances closed registrations, some became unusable under the strain. I guess I was lucky in my timing for my own little move to social.voidfox.com. I did, however discover a fun bug manifest in re-using my old GS domain: anyone still following me from my GS days still thought they were following me, but my instance was not aware it should be sending my updates to them. I think Mastodon eventually re-establishes these follower subscriptions? But for some users their instances have not yet.

tl;dr: If you completely tear down and instance and then rebuild it from scratch later, consider using a different domain for it.

Surprisingly, this Twitter exodus seems to be sticking this time. I mean sure, a lot of people who checked out Mastodon/the Fediverse went back, but the growth this time is quite noticeable. It doesn't help that Twitter's stuck their foot in their proverbial craw several times over the past 6 weeks or so. Every time brings another small wave of users to check out what's on the other side. My own federated timeline is still slow enough to follow and I've met some neat people, but the traffic's definitely been growing!

So I think this time I'm staying. There's some worry about Pleroma's eventual project direction. It's in a weird place now where it's trying to be GNU Social 2.0, but still cater to Mastodon and its users. At this point though I'm pretty confident if it tilted the wrong way, it'd just get forked. It almost did during a recent development spat about new features vs GNU Social parity.

Time will tell; I'm quite happy with the revival of this experiment for now.

2018-Aug-31: New Blog, Now with static engine goodness

Dreamwidth isn't really what I'm looking for in a blog engine. I had switched to it hoping to do "social blogging", but the social part isn't happening so I may as well control my own content again.

I scratched together some python scripts to rig up a blog engine. This one does markdown and lets me embed random files into posts like this:

It's still rough around the edges but I like it. I've copied over posts from my Dreamwidth blog, backdated. The neat thing is my engine is Dreamwidth drop-in compatible (though I lose cuts obviously-- maybe I'll code cut functionality in). I'm also gonna throw this on github once it's a little more reasonable and not full of hardcoded conventions.

2018-Aug-19: On being on a speedrun marathon run selection committee

As part of staffing Big Bad Game-a-thon, I (among others) had the enviable and/or unenviable task of reviewing all the submissions to pick and choose what we wanted in the marathon. This was a first for me; I'd submitted runs to many marathons, but never been on the other side of the table.

At its core it's a simple task: watch over submission videos and decide if a given run would do well in the marathon or not, then review these feedback snippets and decide what gets in. Of course, the execution therein is a bit more complex: you have a limit to how much content you can fit into the marathon because most marathons have a duration decided before run selection, not a duration decided by run selection. As such, there's going to be a percentage of run-time you have to decline, and that percentage may be more than you would really want to decline.

That was the case with BBG. Submissions came in heavy at the end of our submission period and we ended up needing to decline roughly 50% of the submitted run-time. This is a pretty stark contrast from 2017, where we accepted 97%. As a result, the selection committee was forced to be pretty picky and things that would have not been looked at harshly in a marathon 5 or 10% over duration became really big deals. We declined quite a few good, solid runs unfortunately.

I think part of this is a lot of runners simply don't understand how a run selection committee thinks and does their job. I know I didn't until I actually had the task thrust onto myself. I hope to touch on some of the things I saw here and reveal some of my thought process to hopefully help other runners out.

I should note, BBG is nothing compared to GDQ, ESA, etc. We accepted half our submitted runs; events like GDQ accept less than 10%. We received about 100 hours of submissions, GDQ receives many times that. So take what I say here with a grain of salt and understanding that I'm talking about small marathons; I'm sure the process is entirely different for a GDQ.

Our Selection Process

The process for BBG was pretty simple: once submissions closed, we divvied up the runs and aimed to have three reviewers view and comment on each run. As a first pass, we merely added a few sentences of thoughts and flagged a run as surely in, surely out, or needing a second pass. This alone filled a good percentage of our schedule and I think this is where the "little things" really hurt some submissions.

To get through all the content we had, some reviewers watched runs on 2x speed, watched multiple runs at once, or relied on prior knowledge of a game to hop around the video and check run quality and commentary. So while I wish I could say every second of every video was watched three times over, that's not the case; there was just too much stuff to do that. I am confident, however, that every run was completely reviewed between the three reviewers for each.

We had to be fairly brutal in our vetting, with half the submissions needing to be declined. We actually did a second pass on any staff submitted runs with some enhanced vetting, because we wanted to be sure we could fit more submitters in, rather than just showcase ourselves.

After the first pass was done, the committee met and did three quick passes over the runs like so:

  1. Do we want all the "3 Yes" runs? Does anyone object? Any objections on a run bumped it to "Maybe", then the rest were all accepted whole cloth.
  2. Are we sure we want to decline all the "3 No" runs? Does anyone object? Any objections on a run bumped it to "Maybe", then the rest were all declined whole cloth.
  3. We sorted the "Maybe" list by its score ("Yes" is 1, "No" is -1, "Maybe" is 0) and started discussions and voting at the top.

For each run we had a quick chat on its merits and usually the three reviewers and the marathon admins came to a quick consensus on "Yes", "No", or "Backup/Bonus run", and anyone could object to that consensus if they felt it necessary. At some point in this process our votes became consistently "No" across the board, at which point we did another pass of "Are we ready to decline everything below this line?"

This method actually got us within our marathon time window the first try; but if it hadn't, we would have done another round of discussion in a similar manner: going through either the Backup/Bonus list or the accepted list and deciding if we should demote or promote runs to fit the time window.

The total time of the meeting was about 90 minutes. There wasn't really a need to debate much because most of us were on the same page. The meeting was so short because most of the leg work was done over the past week on each reviewer's own time.

Once we had an accepted and backup list, our next step was to get availability from each runner. There's two ways to do this: get it during submissions, or get it from those who made the cut after. We went with the latter option for BBG2018, which brings the benefit of allowing runners to submit without worry that their availability will damage their chances of getting in. From here, we can make a best effort to fit everyone, but if it doesn't pan out for someone we can promote a backup run in their place.

The Golden Rule for Getting In

Every reviewer is different, but to get a "Yes" from most of us in the initial reviews, we needed to be 100% sure the game, runner, run, commentary, and interaction of all of the above would fit well within the marathon. If any one part was absent, at least one reviewer likely gave a "Maybe" grade and bumped that run from the fast-track addition to the schedule. From there that submission went to the thunderdome of "Maybe" runs and had less of a chance of getting in.

Most of what pushed us to reject runs can be summarized with one rule: your submission should be as close to what you will do and what you want to show off in the marathon run as possible. If you're going to commentate the run, commentate your submission. If you're doing a showcase or non-speedrun submission, do a dry run of it as your submission video. If you say in your submission "I can cut 10min off this", try to cut 10min off it before submissions close. Etc etc.

I understand, especially for a small marathon, that sometimes it's not worth the time and effort until you know for sure you're in; but if you really want to get in, this is your best chance to do so.

That's really it. I'm going to touch on some specific points for the rest of this, but it all boils down to that one golden rule.

The Little Details

1.) Don't just submit your PB video

Seriously. This is rule #1 for a reason. Most PB videos lack commentary and most runners, when they realize they're on PB pace, go serious time. So your showcase tends to be a large chunk of silence if you're one of those types. Do a run where you do commentary as if you were in the marathon; even if you don't come close to your PB, you'll have a showcase of what your run will look like.

Even better: if you ran the game in another marathon, submit that VOD. There's nothing better at showing off how a run looks in a marathan than a video of... how a run looks in a marathon!

It may even be a good idea, if you're really attached to getting a run in a marathon, to cut a submission video when you're actively running the game. It doesn't need to be marathon-specific; just do a run with commentary before you put the game down after getting your PB goal and keep it around just in case.

About 70% of BBG2018 submissions were just PB videos. We declined a LOT of borderline runs because the lack of knowing how commentary would go pushed it over the edge into decline.

2.) If your estimate and comments don't match your video, explain why

One of the more eyebrow-raising things in a submission is a 20min video with a 45min estimate. You need to explain that. Is the run heavy on RNG? If so, what do you plan to do if you come in 25min under estimate? Are you likely to screw something up? Can you improve it if so?

On the other side of the coin; if you estimate 20min and your video is 45min, that's even more disconcerting. Why is your showcase of what you can do horribly over estimate? If you're 100% sure you can optimize later, note that, but keep in mind this isn't likely to make your run look good on selection day. Known quantities are good, "I can fix it later" makes reviewers nervous.

3.) Your comments are the first thing a reviewer sees, and will color their perception

You have a tweet-length (usually) blurb in which you can try to sell the reviewers on your run, don't spend it talking down on yourself or saying things like "It's cool if you reject this". If someone talked smack about their own run in the submission comments, I tended to believe them.

The best use of this is to explain why the run is awesome and try to draw attention to the best parts of your submission video. If you say "The skip in level 3 is really awesome too", the reviewer will definitely be paying attention and will remember the cool thing you did when it comes time to vote on the run's acceptance.

That said, don't lie. It's better to say "I won't have time to derust this fully" than to be deceptive about that and put on a bad run.

4.) If your video quality is bad, we'll assume it'll be bad on marathon day (for online marathons)

If your submission video has jacked audio levels, we can fix that in pre-run setup. However if your mic sounds like you're streaming live from your toilet bowl, your bitrate is negative two, or you drop more frames than you stream, we're going to assume this is a chronic problem and will show up in the marathon run. Even pointing out "My net isn't usually this bad but..." is worrisome because, if that was temporary, why didn't you just record your submission another day or do local recording?

You don't need perfect crystal clear 4k video with a $5,000 mic and mixer or anything, but if your submission isn't something you'd want to see in a marathon quality-wise, the reviewers probably don't either.

5.) You get no bonus points for submitting early

This is one I actually need to work on. I have this need to be the first person to submit to a marathon I'm really excited about. Don't though unless you have submission videos from prior events just ready to go. Take some time to stew on it, make sure your videos are good, you're submitting games you're ready to derust and commit to doing well in the marathon, etc.

No marathon is likely to ever close submissions early; you can submit on the last day and it's just as good as submitting on the first.

In future BBGs I'm actually going to lobby for randomizing the submission list before reviewers are turned loose on the vetting process; just to be sure early submitters don't get some kind of subconscious benefit.

6.) If you want a run in, submit it. Really. Don't worry about it

A few people contacted me, or came into general chat, after the submission period ended and talked about how they wished they could have submitted game X but didn't feel it'd get in and didn't want to waste our time. No really, if you think it fits, submit it.

It's our job to vet your submissions. It's your job to make a case for them; go for it. At worst we reject your submission out of hand for being wildly off marathon topic. It's fine. Marathons largely gauge their success by how many submissions they get. It makes us feel good to have a ton of content to choose from!

Also, you don't know what the marathon staff is planning. Maybe you acidentally fit the bill for a feature they were considering. Maybe you'll have them invent a feature for you.

Obviously don't go submitting Super Mario World to a kusoge marathon, but don't be shy if you think you got something that fits.

As noted, this all boils down to giving time and care to your submission video, and being unafraid to sell yourself and your run.

2018-Jul-30: Okami

I streamed a blind playthrough of Ōkami last week. I'd been wanting to play it for years, but I knew it'd be an extremely special and perhaps even emotional affair, so I waited until I felt I was in a good place to commit to it. This was one of the few games I'd dread starting and then getting bored of it and leaving it unfinished. With some of the games I managed to finish on stream over the past six months, I figured now was a good time-- even if I was worried about chat spoiling things.

I'd been warned it was a long game, and I think I kind of rushed it. It weighed in at 33 hours over 7 sessions. Honestly I usually only stream 2-3 hours at a time, but this was so engaging and addictive I was doing 4, then 5, and finally 6 hour streams before the end. I'd originally bought the HD remaster on PC to play, but ran into rather severe issues with it, so I ended up emulating the PS2 version-- this was not an ideal solution but it served well enough.

I'd gone in knowing two things: 1.) That the game was strongly linked to Japanese folklore and even a bit of spirituality (a thing I have close ties to myself) and 2.) The existence of The Sun Rises, which was a song I came across years prior and felt a strong attachment to, though I didn't know anything about the context in which it was presented in the game.

My chat was actually very respectful of my desire to experience the story completely blind. In general my rule is "Let me know if I'm about to miss a permanent miss-able, otherwise let me find things myself," Their desire to shield me from failure was perhaps a bit overbearing at times, but the game is fairly easy on its own anyway. So I got to at least experience the story without spoilers, and I'm glad for it.

I knew right from the start, when you conjure a river into being with a swipe of your brush that the game was going to be something amazing, but it wasn't until I finished up in Kimiki Village and headed out into the world map proper that I was left awestruck. For a game developed in 2006, that's saying something too. The art style is just amazing, and once I purified the first Guardian Tree, well... I was hooked on the game. Everything about the game just has so much care put into it, down to the small cutscenes you get when you feed animals.

The characters are all so memorable too. I think the garbled quasi-dialogue sound effects actually help to give them personality. Everyone you meet just has this unique and vibrant style about them; when you find yourself feeling remorse for the passing of a possessed sheet of paper, you know the character design is top notch. I especially liked the Canine Warriors, as short as their time in the limelight was.

I downed Orochi for the first time around the 12 hour mark and, even then, the game could have ended and I would have felt like it was a complete experience. However it kept going, sending me further afield to Sei-An City; a locale I came to know well across the next 12 hours and the 2nd chapter of the game.

Here though, the game punches you in the gut. This bright, vibrant, cheery game full of humor and references and wild takes suddenly gets serious with the introduction, and then brutal death, of the Queen of Sei-An. It was an almost "out of nowhere" experience, to have one of these identifiable characters suddenly torn apart by the chapter's antagonist; the cheery vibe and happy-go-lucky humor was gone as the stage set for the final battle of the arc. This chapter ended in a massive duel with a nine tailed fox at the end of the game's longest dungeon, and one of the most desperate and intense fights in the game thus far. And again, it could have ended there, but there was more.

As the nine tailed fox falls, she gives mention to the game's true antagonist: Yami. This is, in many ways, almost a Zeromus situation: where you defeat what you think to be the primary antagonist only for them to mention a new, larger foe that up until then had been completely unmentioned. However in this case there were hints of a bigger evil, so it's not quite out of left field. Part three begins with Amaterasu being thrown into the distance icy north, where evil is said to have found its source.

Here the story bares itself: how Amaterasu came to end up on Earth, how demons came to terrorize the world, and the nature of the gods within the game world. As you save the people of Kamui from an icy death at the hands of two clockwork demons holed up high atop a mountain, spewing ice and snow down below, you come to learn of the Ark of Yamato. It is here that both the Celestials and the demon first set foot on the world, having arrived in the ark from the Celestial Plane eons ago.

In the course of hunting down the demons, you also learn the truth behind the legend of Amaterasu: how her first death was not at the hands of Orochi 100 years ago, but she traveled forward in time to assist you in your hour of need, and died from the wounds sustained in a battle with the twin demons as a result. It's a little Deus Ex Machina, but an amazing twist to the legend you're told over and over again through the game, and even get to witness yourself via the same time muckery. You are the hero of 100 years ago, the hero of 100 years ago saves you from death in the present, you're both one in the same-- a thing Amaterasu knows, but since she does not speak in the game, the player never comes to hear it.

After that, the Ark opens, sending you into the final battle with Yami. As an eclipse begins, you as the Sun Goddess are stripped of your powers and left for dead at Yami's feet-- and then a miracle happens: Issun, who had been your smart-alec, smart-mouthed companion up until the Ark, paints a beautiful rendition of the Sun Goddess, running across the land to show everyone her glory, and they come to believe again after generations of darkness without the gods. The belief gives Amaterasu power, restoring her ultimate form and then...

And then the damn song began to play. This random musical tidbit I'd adored for nearly 10 years was the final battle theme, and I never knew. I lost my darned mind. It's a fitting end really, to vanquish the darkness that Yami had brought with the power you'd had all game to simply conjure the sun at will. And then a flurry of offense during which you are practically invincible, and the beast is dead.

Truthfully I thought The Sun Rises was the game's overworld theme. I was confused when I got to the end and hadn't heard it; then the punch in the chest when it begins to play as Amaterasu stands up and prepares to lay the smackdown on the darkness was very, very moving. The entire game was very emotional for me really; very beautiful.

I think at this point it's the best game I've ever played. I don't know, that's a tough call; it's definitely one of the ones most dear to me now.

2018-Jul-16: More retro RPG rambling

The other night I was musing on wanting to find a series of RPGs I hadn't discovered yet, like Deep Dungeons. I may have, kind of?

While futzing around with some old Big Blue Disk dumps, I came across Dark Designs: Grelminar's Staff. This is apparently a sort of hybrid early Ultima/Wizardry style game authored by John Carmack before he went to form Id Software.

The game is a shareware sized chunk of gameplay, as would be expected on a diskmag. The interesting thing though is there's six of them, spread out across the life of the diskmag publication. As far as I can tell three were authored directly by Carmack, and three were written by a different developer using Carmack's engine. All six together would probably combine into one fairly long gameplay experience. An Etrian Odyssey in game length maybe? (Are we measuring game length in EOs now?)

The down side is getting all six of these things running. Platform availability across the series is a little strange. Chapters 1 and 2 are available on DOS and Apple II with both e and GS versions, Chapter 3 drops the DOS support, and Chapters 4 through 6 regress to only offering the e version for Apple II support. That latter most point is a real shame too because the version tailored for the GS offered a pretty significant increase in graphical fidelity that just ceased to exist for the latter half of the series; probably due to the fact that Carmack was not part of the development of those chapters.

Having toyed with the games a bit, they're definitely a bare bones affair: an Akalabeth to the eventual Ultima that followed. Still, there's some neat little features here. You can, I believe anyway, export your party from each chapter into the next, playing the entire series with one group of adventurers. Of course, to make that work you have to make sure you have platform consistency, and for that you are essentially locked onto running an Apple IIe or Apple IIGS.

The bump in graphical fidelity for the first three games is too good to pass up; unfortunately emulating the Apple IIGS is a very sketchy affair. KEGS is your best option on a Windows system, and it's not a very friendly tool at all. At least I now know how to handle ProDOS 6? I guess that's a plus?

I may stream these some time, though I suspect their austerity would turn off viewers. There's no music, no sound, no animations. It makes Wizardry 1 for the IIe look like a modern AAA title. Still it's charming.

2018-Mar-13: My delve into Deep Dungeons

As of late I've been pretty heavily into dungeon crawlers, or I guess the proper fully accurate term is "Grid based first person dungeon crawlers". You know the genre: you're dropped in a maze in first person and have to navigate your way around either with the aid of a top-down map, or by drawing your own. The big examples are Apple and MS-DOS titles such as Wizardry and Bard's Tale.

Well, there's a ton of less well-known titles for consoles. The main examples that were known in the US at least were Arcana and Shining in the Darkness (both amazing games by the way and I wholly recommend them). Popular in Japan, and never seen in the western world was the Deep Dungeon series. Over the past month I've been chipping through all four games of the series on my Twitch stream and at this point feel like I've reached the point where I almost have to ramble extensively about them.

First of all, Deep Dungeon was developed by HummingBirdSoft, a Japanese Real Estate company that branched out into video games for some strange reason. They had been developing games for three years when they made Deep Dungeon, but Deep Dungeon was their first game for the Famicom Disk System. Honestly, the game could have easily been a standard Famicom cart and probably would have fared better for it, but alas.

As far as dungeon crawlers go, Deep Dungeon is both very generic and very basic for its first two iterations: you only have one character to worry about and only battle one foe at any given time. 99% of battles involve you mashing A to attack and hoping you outlast your opponent. There's some semblance of a magic system in the form of purchasable items that are not consumed on use and can inflict damage, silence, sleep, etc. At first glance there's really not much here.

Since this is practically a novel, I'll cut each game into its own clicky link.

Upon booting up Deep Dungeon I: Madou Senki, there's quite a lot that renders the series forgettable on top of the very spartan gameplay: dungeons are mazes of monochromatic walls with the occasional door, your chance to hit any given opponent is terrible and you'll often die simply due to droughts of solid blows on monsters, the story is your bland and generic "the princess has been kidnapped, save her!" fare, and then there's the horrible, screeching, ear-piercing music. Seriously, for the entire first three titles of the series, I streamed with the game audio off and other music playing in its stead.

The actual design of the game bakes in several cases of pointless difficulty. Enemies can find you in real time, and often will if you stop moving. Any time you stop to update your map or figure out where to go, you'll likely be forced into a battle. Many enemies in the latter half of the game have hard control abilities like sleep. Since you're alone, being controlled is usually deadly. Death drops you back to the title screen, so that just means progress lost since your last save.

Additionally is the story of the warp cloth. One of the reusable magic items you can purchase from the shop is a cloth that will spirit you back to town when used. What HummingBirdSoft chose to do to balance this ability was give it an approximately 50% chance to warp you into a wall (read: kill you) on use. This is both maddening, and confusing as Deep Dungeon I allows you to save literally anywhere, at any time. The only penalty for saving before using the warp cloth is the wait you are forced to endure as your game state is written to the floppy disk.

So with all these questionable design decisions, what kept me playing Deep Dungeon I? Really, the map and enemy design. What few foes there are are charmingly rendered, and the maps have a certain inspiration to them. As you wander the halls of the eight floor dungeon, you immediately get a feel for what kind of area you're walking around in: the cemetery is a big open area with small mausoleum rooms flanking it, the underground castle is a big tapering courtyard leading to both an open throne room and a series of dungeon cells, the underground river is a spidering labyrinth of waterways and sluices. Considering maps consist exactly of open tile and wall, it's impressive each floor has a unique feel that conveys a specific location. The level design may very well have been the MVP of the original Deep Dungeon team.

Your ability to progress is strongly gated by level. Your equipment is largely moot as, if you progress in the manner I did, you'll pick up the end-game gear on levels 4, 5, and 6 long before you have the money to buy the strongest gear in the shop. As I mapped each dungeon floor to 100% completion before proceeding to the next, I expected to be on par or ahead of the level curve, but was still forced to grind on floors 2, 5, and 7 before I could survive beyond. As a result, I entered the arena for the final battle on floor 8 at level 17 and rather handily defeated the boss in an anticlimactic battle that also consisted of mashing A until the foe died.

All in all, Deep Dungeon I: Madou Senki was more a mapping puzzle for me than an engaging RPG.

Deep Dungeon II: Yuushi no Monshou is a direct continuation of the story of Deep Dungeon I and is much the same as its predecessor. Once again you have an eight floor dungeon and once again you have a solo hero, solo foes, battles consisting of mashing A, and ear-piercing droning music. The music is better in chapter two of the series, but it's still not good. In fact, that's kind of a motto for Deep Dungeon II: everything has been improved to varying degrees.

Instead of forcing you to walk all the way through eight floors of dungeon every time you return to town, town this time is in the middle of the eight floors, with two four-floor sub-dungeons branching up and down. You are expected to go up first, then down after you've reached floor four. Nothing is stopping you from going down first, but the enemies are powerful and most rewards are locked behind keys or key-like objects you only get by going up.

In this chapter you are also provided stat points to distribute on level-up. These points give you flexibility between Strength, Agility, and Luck. Truthfully the only valid choice is Agility. Anyone who finished Deep Dungeon I probably put most of their points there, in hopes that they don't see a repeat of the abhorrent 50-75% miss rate present therein. This turns you into a god boasting your own 75+% dodge rate, and you'll of course also never miss your own attacks. You'll even dodge spells!

As a result, Deep Dungeon II is the easiest of the four games. You gain level 4 or 5 in the first floor of the dungeon, pump Agility, head up, and never really have a problem to speak of as long as you kill everything you encounter. The game boasts the same end-game gear as the first chapter, and you get it at about the same point, so there's not much in the way of gear progression.

The dungeon layout is the same interesting, somewhat inspired design as the original, except this time in the form of a castle and its catacombs. The castle almost seems to be upside down as you explore it: four inaccessible towers standing obvious in the corners of the map until you reach floor four, at which point you can enter those corners and begin descending back down into the inaccessible corners of the map. It's in these upside down towers that most of the progression items are hidden. Once at the top, you find a pit that leads to a shaft all the way down to the basement where you can begin exploring downward.

Once in basement floor four, you find what you may believe to be the final boss, and it's here that Deep Dungeon II sets itself apart from the original in a real way: there are multiple endings. You can kill this foe, at which point you'll discover the open-ended bad end that allows you to continue the game. Only by finding a hidden item and giving it to this foe does he reveal the location of the true final boss. Still, this is a small additional detail; Deep Dungeon II is pretty much an identically formatted continuation of Deep Dungeon I.

One can really consider Deep Dungeon I and II to be one game spanning two disks. Deep Dungeon III: Yuushi he no Tabi, however, is an entirely different beast. It's the first Deep Dungeon on the Famicom itself, which brings with it a change to the saving system: you can only save in inns in town now, not anywhere.

There a slew of additions to the game: a real magic system, and actual party with multiple classes, multi-foe battles with the enemies arranged on a 3x3 grid with rows that can or cannot be reached by certain attack types, multiple dungeons and multiple towns, and different styles of dungeon. Deep Dungeon III was set to be a sharply superior title than either of its two predecessors. Sadly it's easily the worst.

There's four party member slots and four classes (Hunter, Priest, Wizard, and a Hero class unique to the protagonist) so the natural decision is to run one character of each class. This is probably the most balanced party with the strongest chance of success, though dropping the Hunter for a second Wizard is also compelling. The magic system is bare bones, but is at least present. Priest spells include generic defense buffs, then a small, medium, and large heal. Wizard spells include offensive buffs and damage spells for single target, one row, and everyone. That's about it.

This game spans sixteen floors instead of the eight from the first two titles. These sixteen floors are spread across four dungeons, flanked on either side by small explorable towns. Each dungeon has an obvious, explicit theme as well as what I found to be an implicit theme regarding its source of difficulty:

The game utilizes backtracking very effectively, forcing you to return to prior dungeons only once or twice for additional items and plot elements. These trips back also reveal hidden passages to increase the speed at which you move between dungeons and towns, and showcase how much your party has grown since you were last in that dungeon. The path to the final town actually weaves through the Underground, 15 levels after you were last there; it's really neat.

However the game lacks any semblance of balance or difficulty curve. Having 100%ed the map for the Underground and bought all available gear in the first town, I entered the Tower and was immediately flattened. I was forced to grind for a bit to survive and reach the 2nd floor, where I was once again flattened, and this repeated all the way to floor six. The Tower probably represented about two-thirds of my playthrough despite being only 3/8th of the total dungeon floors.

Once I finished the Tower and entered the Cave, however, I was much too strong for anything in the first floor, and was able to proceed all the way to the third floor without so much as a worry, only to start getting instantly flattened again. Once I got strong enough to survive, the fourth floor and the Cave boss were a joke. The Castle, however, began to flatten me yet again. After several more levels of grinding I was able to barely scrape my way through until the final floor, where the difficulty is cranked to 11 in the form of a teleporter maze with a super-powered tier of enemies unique to only this floor.

The problem with survivability here is two-fold: first, if your Hero class party member dies at any time, you instantly game over. Unfortunately your Hero is also the only one capable of wearing the best armor and boasts a significantly larger health pool than any other member, so he's also the only viable choice for the party's tank. An instant death spell cast on the leader can instantly send you back to the load screen.

Second, depending on where you are in the game, one of your classes is fully useless. At the start of the game it's your Wizard. He's weak and can only cast one damage spell that hits one enemy for laughable damage. One hit can kill your Wizard practically through the entire game as he gets literally no armor or health. Around the mid-Tower, your Wizard gains the spell that can damage an entire row of foes and quickly rises in power just as the Hunter falls. With the hero out-damaging the Hunter on single targets and the Wizard nuking entire rows, the Hunter struggles to find a place in the party; though he is an additional MP battery for between-battle heals, having exactly three Priest spells at his disposal. Once you reach the Cave, the Hunter is nigh-useless against the high physical defense of the foes, and your Wizard is critical for spamming hit-all spells in every fight. Sadly your Wizard is no more durable, still falling in a single hit.

Deep Dungeon III is easily the hardest game for these reasons, and easily the worst. What's funny is I played through the game using a translation patch that also added some difficulty tweaks like making success rolls more reliant on stats than RNG, preventing critical misses that not only fail to damage the foe but cost you an additional turn, and made it so Priests cannot fumble a heal. Anyone who makes it possible to fizzle a heal in an RPG is just a sadist, in my opinion. Despite these quality of life improvements, the game is still utterly unfair.

In its defense, at least the warp item doesn't have a 50% chance to kill you any more. Now you can purchase multiples of it, and it has a 50% chance to instead break, and leave you in the dungeon.

That just leaves Deep Dungeon IV: Kuro no Youjutsushi, easily by far the best title in the series (and the only one I left the audio on in my streamed playthrough). If Deep Dungeon III was an entirely different game from the prior titles, Deep Dungeon IV is on an entirely different planet. There's seven towns linking eleven maps that are both above and below ground. Most of the dungeons are bite-sized one-floor affairs that don't wear out their welcome and probably add up to eight full 32x32 dungeon floors, if I had to guess.

Deep Dungeon IV keeps the multi-character party but takes away both stat points and control over your classes. Instead you find two allies through the course of play who are branded as the party Wizard and Priest. The party hero (who can die in this chapter without an instant game over) is a paladin with solid physical capabilities and some curative magic, then you have Caleph the Wizard and Erick the Priest.

Gone also is the 3x3 battle grid. Instead you have a Bard's Tale style system where up to three types of enemies can appear in any number from 1 to 5. Each group is represented as a single sprite but each foe gets a turn in battle. I think it works much better than the clumsy grid system.

The travel system in Deep Dungeon IV is unique, and has the distinction of being the first thing toward which I can attribute that compliment in the Deep Dungeon series. When you exit a town you may choose which field you want to visit; some towns link to other towns and certain fields, other towns link to different fields. This results in a requirement to backtrack through certain fields a few times to get from place to place, but doesn't require you to move through the entire game any time you want to go somewhere new.

On the difficulty matter, Deep Dungeon IV is the only Deep Dungeon to not have battles "find you" in real time. In addition it's the only Deep Dungeon game that didn't see me die to the first encounter I found when I started. The difficulty curve actually rises gradually, keeping the player on their toes but rarely overpowering them outright. That is, presuming you do things in the right order.

The game is open-ended and non-linear. Right from the start you can easily visit five of the game's seven towns. Each town except for the first one gives a quest that is not directly related to the main plot but is not entirely optional either (I believe the final MacGuffin to reveal the way to the end will not open unless all the quests are done, but this is only a hunch). These quests are a far larger source of experience than normal battles, and it's entirely possible to flee from all combat and level up purely on quests, though you'd be a bit weak as a result. Towns six and seven lie in a different direction than the first five, and would require a bit of training to get there safely.

This open-endedness presents a small problem in that you won't really have an idea of where to go, though the game is small enough that walking around the world and talking to everyone is about a thirty minute affair. If you're truly lost, you can just resort to that, and it should only happen once or twice in a playthrough.

One interesting thing about Deep Dungeon IV is the rate at which you find companions. Presuming you follow the natural progression of events, you could go half the game solo, only finding the way to your first companion when you reach the fifth town. Your second companion is locked away in a tower that stands as either the third from last or fourth from last dungeon depending again on order. This isn't really a problem as your hero character is quite self-sufficient for some time. It's only when paralyzing monsters appear that you will be thankful for a second body. Furthermore, by the time you should have Caleph and/or Erick, you will note a sharp increase in difficulty as a hint that hey, maybe you should go back and find them.

Compared to the first three games, Deep Dungeon IV feels vibrant and alive. The monochrome dungeons are gone, replaced with mazes of verdant trees, craggy rocks, and detailed brick passageways. NPCs have things to say and quests to complete; their dialogue changes as you progress through the game. The plot evolves as you solve quests and meet your companions. It's pretty neat stuff after three games of monochrome dungeon with one page of dialogue with a boss every four floors. You'd probably need a notepad to keep track of all the hints and details NPCs give you.

The one black eye Deep Dungeon IV has comes in the form of its final boss. I'm not entirely sure I didn't miss something, but the final boss comes with the ability to cast the strongest offensive spell in the game, and instant death. The former can be stymied with either the Priest or Wizard silence spells, but the latter seems to ignore being silenced. After several attempts that started with my Priest getting killed, I finally lucked out and had Caleph be the fall guy. After one member died to boss seemed uninterested in continuing to cast death, much like Atomos from Final Fantasy V. So in short, after slogging all the way through the final castle, your success or failure is decided by whether or not it's Erick that goes down to instant death. Kind of ended the game on a sour note.

However, Deep Dungeon IV has one of the cooler things I've seen in an outro for a game this old: every quest you finish, whether it's a mandatory quest, a side quest in town, or a secret quest you stumble onto by wandering into a room that happens to contain and angry dragon that attacks you, is showcased for a moment during the credits in a sort of chronicle of your accomplishments. It was a neat topper for a good solid RPG.

In total, all four games took me about 70 hours to complete. If I had to rate them in order, it would go something like IV, II, I, III, with III a significant distance behind the rest and IV a leap ahead as well. All in all they're solid games if you like dungeon crawlers. They're just not anything unique, and can suffer from shallow design and monotony until the fourth iteration.

Next I may try the Bard's Tale series...

2017-Dec-25: Neugier and the quest to become the potato

I'm an odd type of speedrunner. Three to four times a year I go hunting for strange offbeat games no one has run (or usually even heard of) before, to route and showcase for major marathons.

With SNES Superstars 2018 set to be announced soon after New Year, I went looking early for my offbeat offering for next year. I found it (or one of them anyway) in Neugier: Umi to Kaze no Kodō. This game is Wolf Team all the way through: beautifully thought out and executed with the requisite amount of Wolf Team jank present in the mechanics and gameplay.

Of particular interest here is the ranking system. At the end of the game you're assigned a rank from 1 to 20, with an appropriate title attached. My first playthru got me the rank of "Normal gamer" at 14. As I began to develop a speedrun route I passed through "Masterful Player" at rank 8, then "Beyond Champion" at 4, "Over the Top" at 3, and finally "King of Kings" at 2. Quite the superlative! However rank 1 eluded me. At this point I was completing the game in 20 minutes, down from 2 hours for the rank 14 play.

I did some research and found from game's one FAQ and its translation patch notes that rank seems to be based purely on in-game time and that rank 1 was translated in the English patch to be "Couch Potato". I was now playing on the Japanese version though (as I was doing proper speedruns), so I wanted to see what the title for rank 1 was in the Japanese version. I set out on a stream with the explicit purpose of getting Rank 1, calling it "The quest to become the potato".

It turned out to be a five hour stream as I honed my route and shaved off seconds. About ten attempts in I cut the 20:00 mark, which I thought for sure would give me the desired result. To my dismay I was once again assigned the rank of "Rank 02: King of Kings". Okay fine! At that point I knew ranks 2, 3, and 4 were on one minute increments: sub 21, sub 22, and sub 23 respectively but apparently the illustrious Potato title requires yet more. Back to the grind!

Finally while practicing the Stage 4 boss, I discovered a strategy that cut an entire cycle off the fight. I cut 27 seconds off my personal best and sat with bated breath as the credits scrolled. My chat intoned "Time for potato?!" as the azure cloud-textured text scrolled by and then, there it was...


   RANK 01

Chat exploded. One of the viewers volunteered the translation: layabout, good-for-nothing, deadbeat... Yeah okay, Couch Potato is a good translation for this. A bit kinder maybe. In the afterglow as I tore down stream and cut the video of the run a thought occurred. While I don't consider this to be an optimized play at all, it seems to me they calibrated the ranking system to align almost perfectly with a good speedrun of the game.

I direly want to know the story behind this. Did someone at Wolf Team speedrun the game? Did they sit around on lunch break trying to finish an alpha of it the fastest and then calibrated rank 1 for whoever did so? More importantly, has anyone gotten rank 1 in this before? I peeked around and couldn't find much, but I also acknowledge that I really can't search, for example, Nicovideo well because of language barriers.

In any case, sub-18 in-game time is possible, and I'm curious if it's going to wrap back around to the special rank of Rank 21: Cheater if I do it.

2017-Dec-09: TG-16 History and speedrunning rambling

I speedrun a lot of strange things. Among those things is a pair of TG-16 games. TG-16/PCE speedrunning is a pretty niche thing with only a handful of people actively doing it, and most of them on emulator (myself included). As such, information is kind of short, which is why it's neat when I learn something new.

In this case the education came in the form of discovering a new glitch in Dragon Egg. Specifically, properly jumping and starting a hold of turbo button 1 can allow you to hover infinitely, letting you fly through an autoscroller and skip several tricky jumps. I figured I'd discovered something that'd cut a ton of time off the run, since after all turbo is allowed in TG-16 speedrunning since turbo was a feature of pack-in controllers.

Fortunately I know someone who has Dragon Egg on native hardware and they attempted to replicate my findings without success. Unsure if I'd discovered an emulator only glitch or what, we did some testing across multiple emulators, but the solution was much simpler than anything like that. After scratching our heads one of my viewers pointed out something I should have considered sooner: maybe it comes down to turbo timing and implementation?

Until then I'd just assumed turbo controllers universally just slammed the input as fast as it can. Usually that's 30hz. Why 30? Because most consoles poll input at 60hz, and you have to have one poll "on" and one poll "off" each press of the button, so 30 presses per second. Not so with the TG-16. The pack-in controller does indeed poll at 60hz, but the turbo utilizes one of two frequencies, controlled by a switch: 15hz (1 on, 3 off) or 7.5hz (4 on, 4 off). Since the reference implementation has a cap of 15hz, most other pads just stuck with that, so games really aren't designed to handle 30hz like that.

Emulators, of course, don't care about that. Bizhawk by default uses 30hz but fortunately you can specify frequencies, so once I turned my turbo down to 15hz, I too stopped being able to infinite hover.

Using the "illegal" infinite hover could cut about 45-60 seconds off the run, but it'd be unfair since console players couldn't replicate it, so what do? I ended up updating the rules on the Dragon Egg leaderboard to note "Turbo on emulator is discouraged". Outright banning it isn't fair either, though I know that if someone posts a run utilizing infinite hover, their turbo settings are wrong and the run isn't valid for comparing to console runs.

I doubt anyone's going to do that. To date there's been four runners of the game in 2+ years. I just found the whole thing interesting.

2017-Oct-08: Waxing nostalgic

Dating myself here, but my after-school life through middle to high school consisted of mostly four things: AIM, Quake, Graal, and writing.

This week AIM announced it is shutting down. I haven't used or even looked at AIM for about 5 years but it's still one of those sobering reminders that a lot of the mainstays that formed my developmental life are either gone or changed. I peeked in at AIM and my buddy list was empty; I presume everyone on it hasn't logged in in about as many years as I have not. It's also... sobering... how many people there are that moved on with their lives and lost contact and I'll probably never see them again.

While kicking around news articles about the AIM shutdown, I found one interesting description of the service: "A secular confession box for 90s and 00s teens" and... yeah, definitely. In a way this is where the mentality of freedom from consequences online started. You could hit someone up and talk to them, say anything, and once you logged off you were nobody to them. You'd talk to strangers about your life, your day, who you loved, who you hated. You'd give advice, get advice; these people would change how you felt and how you lived but at the end of the day beyond the screenname you rarely knew who they were. The generation of adults today were raised on the lack of permanence presence then.

In the waning years of AIM, the MO of messaging and chat changed a bit. Instead of 1 on 1 conversations leading to group chats, we moved toward broadcasting sometimes leading into 1 on 1, but often not. Twitter, Facebook, Slack, Telegram... they all kind of lead more to just blatting your thoughts out into the ether and seeing who replies; but more often than not no one does. I think it's kind of sad; but then again I was one of those people who, in the absence of Twitter, just pinned someone down on AIM and said what I woulda tweeted anyway.

I don't know, maybe I'm just being an old coot and sitting here in nostalgia but really: so much of my personal development happened in red-on-black in an AIM window. I don't know for sure if we're better or worse off for that method of conversation being gone, but I feel an order of abstraction more disconnected from current society in its departure.

As for Graal, that's still around but not in the way it was when I was there. I should talk about that sometime.