Cohost Archive

Created . New page. This is written a bit hastily, so some formatting is a bit off. I'll clean it up... some time.

So Cohost is shutting down. (Cohost post from staff, but expect this to go down by the end of 2024.) Honestly it's a shame. I think it's a pretty cozy place to write stuff, kind of like Tumblr if it's not overly "corporation-ized".

Unfortunately, I'm not that active on Cohost in the first place. Fortunately, that means there's not too much to archive, so I'll use this page as a dumping ground.

Many posts are archived here. Some of them are very short and are deemed not worth preserving exactly; you'll see what I mean. These posts are organized by topic, not by time.

My Cohost posts include a mixture of my various identities. This includes puzzles and games, but also my furry side, including my NSFW side. Any such NSFW stuff is marked and hidden under a spoiler block, but be mindful of that.

MIT Mystery Hunt 2023 Recap

Summary: Recap of MIT Mystery Hunt 2023.

Note: Links are rotten. They led to the hunt website when it was still up (in 2023). I'll redirect them to the hunt archive some time.

Click to reveal

As usual, every time MIT Mystery Hunt (or a notable hunt, really) concludes, many people wrote some write-up. I also thought about writing one, but I couldn't think of where I should post it (my old blogs died). Well, until I remembered I had this account and I could write at length here.

It's been a month, so I've forgotten quite a few things, but I do have some fond memories. Here are what I can remember, whether that's working on a bunch of puzzles or memeing with other members of my team. In case you didn't know, I'm in ⛎ UNICODE EQUIVALENCE, a 32-member team that managed to finish the hunt anyway (although at the very last moment).

Spoilers for many parts of MITMH, including many puzzles.


Showcase: I remember looking at these and thinking why these puzzles are so ugly. Then we got the ICPC connection and found the problemset. Apparently it was called ICPC 2021 even though it ran in 2022; I ignored the set for several minutes before it was confirmed by someone else. At least I later made the immediate connection that these are input data to the problems, and solved a few problems by hand while others ran solutions published on the internet.

World's Smallest Logic Puzzles: The name contained "logic puzzles". I looked at this. I don't think I ended up contributing anything; these genres are absurd.

Direct Translation: I think I transcribed a few clues before leaving them for people. I later came back with the grid mostly filled, but with no idea how to extract, before the meta got unlocked and then solved. Many many hours later during Act 2, I went back here because I didn't have any puzzle I could help with. I pretty much solo-ed the final aha about reading the "golden" path along the sides and managed to get some sort of answer phrase, but I had to ask Deusovi to help reading it into the correct phrase that I could type.

Also I'm pretty sure I went to bed around 5-6am, approximately 4-5 hours since the hunt started. This made me miss Hall of Archaeology and I'm still low-key mad about that. That puzzle also sniped one of my puzzleideas.txt.

Think Fast: Perhaps my top contribution which really deserves a whole segment. See below.

Reflective Screen: The most logic puzzly thing in the whole hunt! At first I thought mirror cells would bounce all lines of sight, but that didn't make sense. Only then I realized it was simply diagonal mirrors. We got the aha, but the puzzles were pretty hard; those six puzzles (excluding the last) took us around two hours, even with four people working.

The Typesetting Machine: I and lovemathboy worked on this. Or really, it's mainly lovemathboy with me watching. I gave an aha with the final extraction picture, and suggested lovemathboy to throw the resulting cluephrase to a #help-wanted channel we have. Minutes later the answer arrived.

4D Geo: Funny thing about the geometry problems: most online resources tell you the formula of how to find the triangle center given the three triangle vertices, not the other way around. However, it sounded intuitive that the triangle center would be continuous: moving one vertex smoothly would also change that center smoothly. So to do the converse problems, I coded up a general-purpose thing. It took two vertices and the desired center, as well as a formula that gave the center given three vertices. (I believe all the formulas were barycentric coordinates; they were on the Encyclopedia of Triangle Centers.) What it did was, it sampled points uniformly on the plane, then zoomed in at the local minimum and sampled more densely around that point. Rinse and repeat until the epsilon was low enough. I'm still proud of this incredibly hacky solution.

World's Largest Logic Puzzle: Honestly I didn't really work on this puzzle. boboquack was several steps ahead of me, so while I was double-checking his work, he went on to the next step, all the way to the end. We ended up with the cluephrase "WSLP 10x10". After googling "WSLP" and not giving anything good, I looked at the puzzle title and realized WSLP referred to the previous puzzle. We were stuck at what to do with the 10x10 part. Well, the extraction grid there was 10x10, but what to do with that grid? lovemathboy suggested to just submit it.

⧟ boboquack — 01/15/2023 8:42 PM
  this thing is 10x10
  
✾ lovemathboy — 01/15/2023 8:42 PM
  submitting that ??
Ғ chaotic_iak — 01/15/2023 8:42 PM
  yes so i think we want to use that grid, but
  hmmm
  i did it
  it's correct
  lmb galaxy brain
☞ JUSTIN — 01/15/2023 8:43 PM
  Wait
  The answer is
  that grid????
⧟ boboquack — 01/15/2023 8:43 PM
  lmao

The Monday 5.30pm e-mail: This needs its own section too. See below.

I wanted this to mostly be a recap, so I didn't really want to say much else. Hunt was overly hard and long, but I believe it was just miscalibrated; it was still very fun with a lot of great puzzles. (That said, the amount of free answers at the end meant a lot of the late puzzles went underappreciated. I already spoiled myself reading solutions though, so.)


Think Fast

All of us worked on the various levels and difficulties until all the hards are done. And then the interaction prompt popped up at Saturday 2.36pm my time (1.36am Eastern). A few "uh oh"s came up. We weren't sure whether we should schedule the interaction or not, but I decided we should just go for it and see what would be waiting for us. I sent the request around 3pm, and the reply came back minutes later scheduling us for a 3.30pm timeslot.

Interaction at 3.30pm went... rather awkwardly. First off, the Zoom meeting was locked by a passcode, but the link they gave us had incorrect passcode, so we had to send a request asking the right code. During this time, lovemathboy tried to bruteforce the code, failed too many times, and got locked out from the meeting and so couldn't participate here. After asking one of us to screenshare, Tiralmo stepped up; I was only talking in text. We didn't have any trouble with the "easy" levels: levels 1 (numbers), 2 (positions), 3 (letters), 5 (alphabetical numbers). And then we tried level 4 (arrows). And tried. And tried. We went like 10-15 tries without success. The meeting went for like 30 minutes before HQ asked us to come back later once we're more prepared. We said yes, but we wanted to give level 6 (anagram) one try. Sure, they said. haha nope we failed immediately

Now knowing the structure of the expert difficulty, I wanted us to practice tackling this more. The problem is, most of the people that were in the meeting were on-site people, and it was way too late, so they wanted to retire to bed. Meanwhile, Asian timezone remote solvers were having several things to do, like getting lunch, getting dinner, and so on. I couldn't gather enough people to practice and we abandoned the puzzle for the time being. (I think I was also sleepy at the time.)

At 10pm, I woke up from a nap, saw the puzzle was still untouched, and decided to give it another go. I tried assembling people to practice and it was... disappointing, to be honest. I think there were lovemathboy and TheGreatEscaper. Most on-site people were still sleeping. No way we would get these with just three people.

Luckily, at 11.30pm (after I also posted in a #help-wanted channel about "getting as many bodies as possible"), on-site people began to wake up and come in to check this puzzle out. We began gathering in a Zoom call to practice. Since I was the most insistent, with various plans and strategies to do the levels, I ended up being de facto leader for this interaction. I realized I probably should have done that for the 3.30pm interaction too.

Either way, we practiced a whole lot for level 7 (hues), the level we were most scared of. I think our strategy ended up being, reroll until we get yellow-blue (or something fixed, I forgot exactly), then each person has been assigned to memorize a certain portion of the spectrum. About half hour of practice made us feel more confident. We also did a few tries for levels 4 and 6, with better coordination.

Before this practice session, at around 11.30pm, I sent an HQ request for the interaction, preferring a 12.30am timeslot (and giving us some time to practice). (Also, we went for virtual interaction because I was leading it and I was remote.) The HQ must have been swamped, since we didn't get a reply and our desired 12.30am timeslot was missed.

The reply only came at 12.57am. We were to do the interaction "RIGHT NOW", at 1am. We immediately filed in for that one. This time, I immediately took over as the one holding the controls and also assigning people to memorize things.

The first thing we heard was, "you guys just have to do 6 out of 7 levels". We instantly dropped level 7. That hue thing isn't going to be our nightmare, good. So we had to redo all levels, but levels 1, 2, 3, 5 fell down instantly. We knew level 4 would be a nightmare, so we went to level 6 first.

During the 3.30pm interaction, the only time we did level 6, we got the word RECOGNIZABLY. lovemathboy had trained for like an hour anagramming various common 12-letter words in his head, and we set him and ManyPinkHats to be our anagrammers while the rest of us memorized letters.

Then the letters came in. In the one second they were displayed, ManyPinkHats called out: RECOGNIZABLY. Well then.

Back to level 4, here we go. We rolled a few boards until the squares were spread out enough (lessening the chance of ambiguous arrows), then we went for it.

And got stuck.

It didn't make sense; various arrows were ambiguous, and even trying to piece together the chains in my head, I couldn't figure out how to disambiguate these. We decided to just take a shot, and if we failed, just roll another try. So I took a stab. And got the correct answer. That's 6 solved.

While others were celebrating, I looked at the resulting arrows. And realized that the grid did have multiple solutions! We did solve it by sheer dumb luck. (But our strategy should have been pretty sound that even if we failed, we would get it the next try.)

Anyway I'm just proud to have led a whole interaction.

(We would then be stuck in extraction hell for like 4 hours after that.)


The Monday 5.30pm e-mail

First, context. We realized that the Sunday 6pm Eastern (Monday 7am my time) deadline passed without any e-mail that the coin was found. That meant we resigned ourselves that we wouldn't finish (because we likely wouldn't win). So we just kept going, solving puzzles until we would get cut off.

Then at 5.30pm, we got an e-mail. "The coin will have been found" is a funny wording, but what alarmed us was the thing below it. "Other teams who wish to reach the end game must finish MATE's request by 8am." That's 9pm my time, that's in 3.5 hours. We were immediately on full panic mode.

I immediately pinged mbingo as team leader to wake up. Level 51 did so too. And then he remembered he had mbingo's phone number from last year.

🁕 Level 51 — 01/16/2023 5:44 PM
  guys i have maxs number from last year
  should i try it / anyone wanna try it
Ғ chaotic_iak — 01/16/2023 5:45 PM
  do you want to make an overseas call lmao
🁕 Level 51 — 01/16/2023 5:45 PM
  idm
Ғ chaotic_iak — 01/16/2023 5:45 PM
  go ahead
🁕 Level 51 — 01/16/2023 5:47 PM
  uhh how do i make overseas calls?
  LMAO
🁕 Level 51 — 01/16/2023 5:49 PM
  i got sent to voice mail 
  L

Well, that didn't work. Luckily Dukael managed to reach him a few minutes later and woke him up.

At this point, we were sitting on 24 "answerbergs" (free answers, but the name from GPH19 stuck). Given that we only had 3 hours left, we decided to be way more liberal at "berg-ing" things.

At this point, we had only seen 3 tiers of the Wyrmhole. We solved the meta, went up to the next tier, and saw... blank puzzles. This was something that we should probably be able to figure out, but we're running out of time and we berged two of the six blank puzzles. This revealed its meta to be Collage, we made the connection, and we managed to "properly backsolve" the other blank puzzles and thus saving up on the answerbergs. (That was not necessary, but just in case.)

While I and half the team (that were awake, at least) worked on Period of Wyrm, the other half were on The Tower of Eye. I didn't touch the latter, but I'm proud to have coded yet another thing for Period of Wyrm (specifically the tortoise-hare algorithm to find the period of the attracting cycles).

We finished the two metas at 8.46pm, 14 minutes before deadline. And then a final capstone puzzle appeared. Whoops, we're not going to make it. We sent a Hail Mary submission of THERE'S NO AI IN TEAMMATE three seconds before the deadline. (It didn't work.)

That said, we kept working on the puzzle anyway. We later got another e-mail where we would be allowed to get into the last runaround at 9.45pm (45 minutes behind deadline) if we were done by then. (We solved the capstone at 9.22pm.)

As a remote solver, I couldn't participate on the runaround, but someone on-site kindly streamed to us what's going on. And another Mystery Hunt ends.

Comment from @mbingo:

Lesson learned: turn off iPhone's sleep focus mode during Mystery Hunt. 🤦🏽‍♂️

Nice writeup! And great work/thank you for taking the lead and coordinating Think Fast for us over many hours. (Also oof on that puzzleideas.txt snipe… though maybe only one casualty is pretty good for MITMH?)

My reply:

I don't have many ideas in the first place, so one sniped is a lot. The snipe is so similar, too; interactive logic puzzle where a mistake costs you time. I'll just be glad that my idea was indeed worth exploring and being made into a puzzle.

pokemon breeding adventures

Summary: I tried breeding some regional forms (Kanto/Alola Vulpix, and Unova/Hisui Zorua) in Pokémon Violet, using Everstones. But I observed unusual counts on the resulting egg distribution. I didn't pursue a further explanation though.

Click to reveal

i had access to alolan vulpix and hisuian zorua, so i went to breed some to share through surprise trading. and i got some weird results.


assumptions about breeding:


i went to a picnic with these:

the shaymin shouldn't matter. the destiny knot passes down IVs but otherwise shouldn't matter for forms.

if my assumptions were correct, then there are 8 possible pairs to breed, which the game picks at random:

therefore i would expect the spread to be:

kanto vulpix 3 : alola vulpix 3 : unova zorua 1 : hisui zorua 1


my results? i got 35 eggs, and after i hatched them, the spread was:

alola vulpix : hisui zorua being 2 : 1 instead looks a bit fishy but i think my sample isn't large enough to say.

but the original forms having completely zero eggs is very weird. i expect an equal spread of kanto vulpix and alola vulpix. what's happening?

my hypothesis is: in case of regional forms, if any mon in your party holds an everstone, they will pass that form, even if not involved in the breeding pair. then even if the game picked ditto x kanto vulpix, because alola vulpix holds everstone, the offspring will be alola vulpix.

i have no idea if it's correct or not, and i'm not particularly enthused to do research here (the whole reason i bred was just to share some goodies through surprise trade), but if anyone wants to pick up and do more experiments, feel free

Comment by @OnionBoy:

I know that historically, pokemon in the daycare together will breed fastest if they are the same species with different OTs, and then same species with same OTs. It could be that with 2 vulpixes in your party, the breeding selector will never pick a ditto, leading to only the alolan form. And same with the zorua.

My reply:

the two vulpixes are both female though, so you can't breed them as a pair

recent games i've been playing

Dated: July 2023

Summary: Some games I played in the previous month and my quick reviews, and some recommendations of my own

Click to reveal

given that it's steam summer sale, i suppose i can write something. so here are some games i've played recently, followed by some random games i recommend. warning: heavy on puzzle games. i mean, i'm a huge fan of puzzle games.


👍👍 Bombe: minesweeper but you only solve each situation once. you write a deduction rule for your situation and the game automatically uses it. feels like programming, in a good way; kinda addicting to see your hard work clearing 16,000 levels at once

👍👍 Gabbuchi: puzzle platformer where you can switch colors and you eat blocks only of your color. some blocks you need to eat to be full, some blocks you need to leave out so you can stand on them. really fun game that is obscure even in my community of indie puzzle gamers; i've started speedrunning this

👎 Pertinence: twitchy action game that pretends to be puzzly. i mean, it is puzzly, but it's much more about execution. guide your little ball through various gauntlets; you need reflex, timing, and such. i got through the whole game in about two days. the concept looks interesting, especially exploring new mechanics you find, but i definitely didn't sign up for a strict timing-based game. ...that said, it's my subjective experience; if you know what to expect, you can very much enjoy it

👎 ibb & obb: i visited a friend's place last week and we pulled out co-op games; we ended up playing this for 3.5 hours and no other game. i came here pretty blind; i knew it's puzzle platformer with two characters, i knew there's gravity changing involved, but nothing else. after playing it for long enough, i decided it's not for us; it started getting really heavy on techniques and execution, and even when we know what to do, it's hard to execute it. so, similar to pertinence

👍 UFO: Unidentified Falling Objects: it's not out yet, it's out in two weeks. i played the demo. it's very similar to Super Puzzle Platformer, an old-ish game (2013). you're on a platform of colorful blocks, and new blocks constantly fall in, plus there are some hazards like spiky blocks. shoot at a group of blocks of the same color to pop them, survive and score as high as possible. score attack game, basically. i liked the gameplay loop of SPP so i enjoyed UFO too. the improvements over the basic formula were great; i played SPP and felt it was missing those things i found in UFO. can't wait for this game


and here are some random recommendations of games that i happen to remember (not that i played them recently, just i remembered they were good)

👍👍👍 N Step Steve (Part 1, Part 2): this is, bar none, my favorite puzzle game, and it is in the running for my favorite game ever. also, it's free! the idea is a sokoban-like, with you as a player character moving on a grid. you can only survive moving for N steps (usually 5, but really varies) before you die; step on a flag to restore your step count. throw in several different mechanics. part 1 introduced these mechanics individually, and then part 2 gave devious concoctions of these mechanics interacting. it's incredibly deep and i was delighted with its various surprises. that said, it's also very hard, and if you're not interested in thinking at depth, you won't enjoy it. but if you're a puzzle game fan, really go play this

👍👍 Linelith: a short puzzle game you can complete in one sitting (around 1-2 hours), but somehow it embodies what i love from puzzle games, constantly surprising me in delightful ways. given that it's so short, i won't say much about it and let you try it for yourself; it's pretty cheap (around $2) but it's well worth it. the game itself is part of a bundle, CosmOS 9, consisting of 9 original short puzzle games, but i found out only Linelith was really worth it

👍 Autonauts: think Minecraft or Factorio, but now you have robots that you can program to do tasks for you. need a bunch of wood planks for construction? automate a forestry with robots that plant new trees, cut the trees, pick up the logs, and process logs into planks. the idea of the game is great and i do enjoy it. i think i just end up getting distracted by many other games in my backlog, and whenever i go back here, i just reset my progress because i no longer understand what my robots are doing. but if you can be attached to this game more, i'm sure you'll enjoy it plenty

more recent games i've been playing

Dated: August 2023

Summary: More games I played, more reviews, more recommendations. This is posted after my Cerebral Puzzle Showcase post, so it includes several games I picked up there.

Click to reveal

after what i did a month ago of course there are more. also do note there are some more games from my post about cerebral puzzle showcase.


👍👍 Candle Prick: nice and chill puzzle game. okay, "chill" might not be the right word, but i got through the levels reasonably well. there aren't too many mechanics, but they are utilized very well, with nice surprise moments around. i 100%ed the game in around 3 hours, but it was time well spent. my issues are graphical: the screen is too busy (takes me quite a while to learn what's foreground and what's background), and some of the palettes are atrocious (i ended up just using the default palette all the way through). but the game is still a lot of fun.

👎👎 Dr Fetus' Mean Meat Machine: classic puyo puyo game with gimmicks! sounds incredible. nope, it's all but that. you have to dodge hazards on board, but they often fill the screen a lot leaving little space to do anything. which means you can't actually build up long chains whatsoever. that's not puyo puyo, that's just some match-4 game. if this game took more departure from puyo puyo -

  • say, making you match 3 in a row instead, or at the very least avoiding the allusion to another puyo puyo game -- maybe i could give it more credit. but, nope, this game is not worth it.

  • 👍 Magicube: absolutely no filler whatsoever. right out from level 1 you're forced to think hard. i think i only finished 6 out of 50 levels and i was already struggling. i'll get back to this game at some point (and likely rate it higher), it's just that it takes a lot of time and brain power for the game.

    👍👍👍 Pâquerette Down the Bunburrows: i think this is my favorite puzzle game of the year, and is in serious contention to compete against N Step Steve (mentioned in my previous post). the mechanics are simple (okay, it's not very simple, but mostly reasonable) but very deep, with some incredible puzzles -

  • and more importantly, incredible moments. the first "holy shit what just happened" moment for me came in area 3 of the game. aren't there only 5 areas, isn't that halfway in? wrong, that's like 5% into the game. and then i keep getting more of these moments, breaking into laughter several times. this game is similar with NSS in that it explores the implications of its mechanics very thoroughly, packaging all sorts of surprising results as moments for the player to find.

  • 👍 Pikmin 4: i've never played pikmin before so i tried the pikmin 4 demo in june. it was so fun that i preordered it immediately. after actually playing the game, it started getting repetitive and stale, but also it ended just about at the right time. i do like the puzzliness of the dandori challenges (one day i'll platinum them all), not so sure about the battles. i liked my time here, although i don't think i'll catch up with the past games, at least not yet, just because as i said it gets kinda stale

    👎 shapez: factory game. except that i quickly felt the need of moving and rearranging my factories around... but the copy function is locked until level 12 or so. i was level 10ish when i decided the game wasn't for me. i ended up reinstalling and playing Autonauts after.

    👍 UFO: Unidentified Falling Objects: wait chaotic didn't you mention this game before? well i did, but this time it has been released and i played it quite a lot. it should be something i get bored of for being repetitive, but i 100%ed the game before that point. (well, "almost" 100%ed. there's one objective that seems broken for some reason. also, i might be missing some hidden objectives.) if you like action puzzlers like tetris, lumines, etc, this is worth playing.


    and as always, a few other recs:

    👍👍 A Dance of Fire and Ice: rhythm game, but only one button needed. (well, that's a lie, you'll need two or even more buttons when you get to the faster charts.) it's incredible how you can simply change the usual presentation of rhythm games and get something really novel. normally, in rhythm games, you have notes falling to you, but change it into following a track using angles as timing and it becomes fresh. i've done a lot of the levels (although not speed trials); some of the levels i disagree with, especially if you get to the custom charts that are usually just unreadable speedy mess, but overall the game is still worth playing.

    👍 Normal Diner: serve sandwiches to people :^) while this game is just an arcade with no progression, it has some promising ideas, and it's always good when a game makes me laugh out "what the fuck" (see also Paquerette above). and apparently it has a sequel Galactic Diner in development?

    cerebral puzzle showcase

    Dated: 2023

    Summary: My recommendations from Cerebral Puzzle Showcase 2023, whether it's games I'm looking forward to (most of them have released by this writing in Sep 2024), games I've loved in the past and were also on sale, etc.

    Click to reveal

    the biggest puzzle game event is now live! lots of high quality titles, including some new ones i'm ecstatic in trying


    new releases i'm looking to play:


    upcoming games i'm still eagerly awaiting, many of these have demos:


    and games that have been released before that i generally like:

    big names

    i think these games are widely known to be good, at least widely known among the indie puzzle games community. i'm skipping the blurbs but i can say these are good.


    these games aren't part of the showcase but i've been playing them anyway:

    Hillhold

    Kink/fetish-based. Strictly speaking, not actually "lewd", but definitely not something innocent.

    Click to reveal

    Summary: An idea I got about a sport in my kink universe where people like being turned inanimate.


    Idea for a sport in Stillen: hillholding. (Name may change.)

    Two teams play in an arena filled with obstacles (like parkour / world chase tag) and one or more pedestals. The competitive standard has teams of five, but the arena may be freely designed; adapting to a brand-new arena is considered part of the game. Variants may involve smaller teams.

    Each team has a score, initially set at a high number; the goal is to be the first team to bring this down to 0. The game also ends after a time limit; the team with a lower score (if tied, the first team to hit this score) wins.

    If a player is able to stand on a pedestal, after a short countdown, they freeze completely immobile, although they are still aware, and can see and hear (although not speak). As long as they are frozen, their team's score counts down. Different pedestals may have different rates, although usually the score ticks down at 1 point per second. If the arena has multiple pedestals, they are independent; a single team claiming multiple pedestals will have their score tick down faster.

    The opposing team may tackle a player off the pedestal, in which they no longer help bringing their team's score down. However, they remain frozen in place until tagged by another teammate.

    (With poor planning, a team might be entirely rendered immobile. They may still win if they are close to victory in the first place, but generally this leads to a loss since nobody can contest the opposing team claiming all pedestals.)

    There may be additional game-adjacent mechanics that I may come up with.

    By tradition, at the end of the game, there's a little celebratory ceremony. The winning team are all frozen as trophies, while the losing team has to bear a little humiliation of lifting and carrying them up during the ceremony.

    The name comes from "king of the hill", a children's game with similar properties of claiming the top spot and holding it for some time. (Might also be familiar as the Splat Zone game mode in Splatoon, if Inkipedia serves me correctly.)

    Other posts that got their own pages on my site

    Little posts that I deem not worth preserving exactly