Dyo is a take on the two-player-puzzle genre that I haven't seen before: with one minotaur on the left half of the screen, and the other on the right, you can lock the viewports and walk into the other half. So, for example:

• A platform is too high to jump to. Nuts.
• But you can jump on the other character, and leapfrog up to it!
• Except now the other character is stranded below. Until...
• You lock the screen, and the lower character can walk over to where the higher character is.

It's a neat, mind-bending trick, and the game shows off additional twists and mechanics in each level -- like staggered platforms which require adjusting the screen-locks as you climb, or a pushable block which acts as both an obstacle and a platform.

The problem is... Dyo is terrible at teaching you how to play. There are no hints toward how its mechanics work (the pushable block can float!?), and there's no guidance on how you should approach a puzzle. A good puzzle-game design - really, a good design for any game - is gradually introducing new mechanics on top of old ones, so the player can learn incrementally; but every level in Dyo feels totally different, a whole new set of challenges that previous levels haven't prepared you for.

The result is that figuring out each level involves a lot of staring, and a lot of trial-and-error testing if your character can make a certain-length leap or fit through a certain-size gap. Learning the unique ins and outs of a level is tedious, and boring.

I'm impressed by Dyo's cleverness (and I love the Sands of Time-style rewind/undo feature), but actually playing the game isn't very fun.

Better than: Kalimba
Not as good as: Dyadic
What the fuck: is the point of this guide? "Press Ctrl + F2 to instantly win a level." Brilliant solution, genius. (Asshole.)

Progress: Got to the crossroads, and a few puzzles after that.

Murder by Numbers is a murder-mystery visual novel with investigation elements inspired by Phoenix Wright, and punctuated by nonogram puzzles. It even shares some of Ace Attorney's musical pedigree. This... should really be a slam dunk for me. And while I ultimately still liked this detective-story-and-puzzles game, I can't help but obsess over how I might have liked it even more.

Let's start with the good. Murder by Numbers has a slightly-absurd premise - protagonist Honor is an actress playing a detective, assistant SCOUT is a floating robot who really shouldn't exist in the game's 1990s setting - and yet, not unlike the Fey family's insane spiritual powers, strong characterization and scene-setting make it all totally believable. Detective-story tropes are both honored and subverted, Honor's friends have some genuinely interesting qualities, and characters react to each case's tragedies in poignant but believable and sympathetic dialog.

The case stories are engaging and well-structured, for the most part -- a notable exception being the final case, which feels a lot like it should have been two separate cases (especially since the Kino case really deserved a better resolution of its own). But otherwise, the written story of each case is plenty immersive and compelling.

And like some of the best Ace Attorney tales, Murder by Numbers does a good job of blending humor into its narrative.

Unfortunately, the way in which puzzles are blended in leaves something to be desired. What could have been a gameplay mechanic which supports narrative immersion is instead a distraction from the story, and ultimately detracts from it.

An example of how this combination "should" work is in one of the game's first puzzles. You're in a locked room and looking for a way to escape: you go into robot-assisted investigation mode (neat!); the flashing reticle and beeping noises draw you toward a ventilation grate (good so far!); and you complete a nonogram puzzle of the grate to uncover the exit. Pretty slick, right?

Well, this ludonarrative harmony is the exception, not the norm. From there onward, the investigation target is almost never related to the puzzle -- the reticle will converge on a bathroom mirror, and you'll pull out a fire extinguisher; or it'll focus on a featureless section of floor, and a gun will magically appear.

At first these inconsistencies between environment art and puzzle pictures feel like missed opportunities. But as the game goes on, and this discordance continues unabated, it hardens an immersion-breaking wall between story scenes and puzzles. Turning key items into low-resolution puzzle grids is one thing, but when that item doesn't appear in the investigation screen at all, ... what, exactly, is the point of investigating?

It doesn't help that these puzzles become increasingly tiresome in the final case. There are simply too many of them inbetween the later story scenes, and the puzzles themselves start to blend together, as almost all of them are 15x15 grids. (There are one or two 20x15 puzzles at the very end, but otherwise the game gets "stuck" at 15x15 for most of its running length.)

There is one other venue for puzzles in the game's story, in a few instances where SCOUT needs to "hack" something and you must complete a series of small, randomized 5x5 puzzles with a time limit. These segments are a welcome departure from the normal puzzles, and their hectic presentation, along with the time limit, makes completing them actually feel thrilling. But they don't occur very often.

Outside of the story, there's also the Memories section, where - without narrative accompaniment - you can complete puzzles to "repair SCOUT's database." These puzzles are unlocked based on your progress through the story, which works fine for the most part, with one irritating exception: some of the third case's puzzles are missable, consequently limiting the number of Memory unlocks.

The backstory rewards you get for completing these Memory puzzles are also thoroughly underwhelming, but, hey -- more puzzles.

Finally, I'd be remiss - given my recent complaints about Puppy Cross - if I didn't mention that Murder by Numbers has somewhat mediocre input handling of its own. It's nothing so bad as having to wait for irrelevant animations, but click-and-drag doesn't lock to a row or column at all; and there's no undo button, so it's awkwardly easy to drift into marks you've already made and overwrite them by accident.

I don't want to make it sound like I merely tolerated the game's shortcomings. I mean... I kinda did, in the Memory puzzles. But it isn't as if the game's story mode was "bad," hell, the storytelling was actually pretty good! The art is great, the music is catchy, and the puzzles are as functional as they need to be.

The whole package could have been even better, though, with just a bit more attention paid to the narrative-puzzle bridge.

Better than: Professor Layton and the Curious Village, Puppy Cross
Not as good as: Nonogram - The Greatest Painter, Professor Layton vs. Phoenix Wright: Ace Attorney
Not as good as most Ace Attorney games: except for that Edgeworth spin-off and those godawful "Asinine Attorney" DLCs.

Progress: Finished all cases and all memory puzzles.

Mugsters has some slick and action-packed trailers (as seen on its Steam page). Sadly, the game itself is ... fairly different.

There's no introduction, cinematic or otherwise. There's no in-game storytelling except what's implied by the existence of teleporters and loud machines. There isn't even any background music!

Although the game doesn't tutorialize its controls at all, there's not much to learn: jump, run, pick-up, punch/throw. The lack of available player actions is apparent from the very first level -- a human is trapped in a glass tube, so, how do you free it? There are no interactive elements, and punches don't work... guess you'll need to pick up and throw something at it.

Its physics sandbox is pretty much all that Mugsters has going for it, and gameplay leans hard on this: throw explosive barrels at this thing, jump a car over a ramp, et cetera. And yet, there isn't much to do in the game's sandbox, other than lower barriers and destroy walls until you reach the end of the level.

As nice as the tilt-shift perspective and "clean" un-textured art style look, these become liabilities during actual gameplay: obstacles frequently obscure your immediate surroundings (prepare for lots of camera rotation) and ground elevation changes are difficult to read.

And although Mugsters can be played co-operatively, the appeal of a multiplayer sandbox is brought low by the auto-zooming shared camera. Your player avatar is already tiny when flying solo; a second player just zooms out even further, and can make the level quite inscrutable.

If you really, really want to play around in a fairly-sparse physics sandbox with vehicles, then Mugsters offers several levels' worth of that. But the game is otherwise outstandingly barren.

Better than: Cubots: The Origins
Not as good as: Zoo Rampage
For a better vehicle sandbox, consider revisiting: Blast Corps

Progress: got to island 7

Puppy Cross has a cute premise - completing nonograms of dogs (and cats!) to let them frolic around a garden - and a hefty amount of puzzles backing it up. Even if the customization aspect of the garden is kinda shallow, finishing all the puzzles feels like its own reward.

It even has one of the best quality-of-life improvements I've seen in a nonogram game: once you've marked all the right cells in a row or column, the "X" cells are filled in automatically. This is a pretty great time-saving convenience that streamlines the flow to the next row or column.

Unfortunately, Puppy Cross's strengths are dulled by some unpolished fundamentals. Like disorienting puzzle progression: you need to complete a "world" before moving on to the next, but each world's first puzzles are an order of magnitude easier than the preceding world's last puzzles!

Even though there's barely any text in the game, it suffers from a clear and frequent localization/grammar error. Every completed puzzle unlocks a new item, but they accidentally a word in this messaging:

... and then there are the input problems.

• Click-and-drag can select multiple cells in a row or column, but if you drag past the edge of the puzzle, the input is lost. So filling to the end of a row/column requires more precision than it should.
• Meanwhile, click-and-drag can "X" cells outside of the current row and column, but can only un-"X" one cell at a time. It's easy to "X" more cells than you meant to, and hard to remove those marks afterward.
• Maybe worst of all, mouse-clicks are ignored until animations have stopped playing. Every time you return to the puzzle menu, you can't select a new puzzle - or even get any visual feedback on your input - until the go-to-garden button has finished fading-in, several annoying seconds later.

Much as I appreciate the game's amount of content and its thorough hinting system, the input issues are fairly critical. Puppy Cross is certainly playable, but hard to recommend over more basically-functional nonogram games.

Better than: Pokémon Picross
Not as good as: Nonogram - The Greatest Painter
Dog and cat puns not as good as: Cat Quest

Progress: 100%

Wuppo isn't a "bad" game, exactly, but it's weird in ways that I found more off-putting than charming.

Not just because you jump with the left-bumper, or because you spend currency by equipping it like an item, or because the map is forced to share screen real estate with a pause menu. More because the game's story is told with cartoonish nonsense - it introduces the world through a history of races with names like "Blusser" and "Splenkhakker" - and because, although the game clearly displays your objective, what that objective means is often elusive.

Like, my mission for most of the game thus far was to get back to the "Wumhouse" that I started in. But how? Well, I guess by scouring the accessible map and looking for un-followed routes, or using items with interactive elements until something new opens up. Wuppo doesn't just obscure the path forward with a lack of bright-and-shining objective markers, but also with shades of adventure game (il)logic.

And my problem with Wuppo isn't necessarily that it asks me to understand and execute on its weirdness; it's that I don't know why I should. Mechanically, it's a platform-y exploration game that's pretty unremarkable outside of those weird design choices. And narratively, the premise is that you were kicked out of a hotel for making a mess. What, uh ... what is the point of this game?

I feel like I could "get" Wuppo if I kept trying, but it's not giving me a very good reason to do so.

Better than: Forma.8
Not as good as: Headlander
Very distinct from, but ultimately as uncompelling as: Strider (2014)

Progress: Got back to the hotel, then got lost and gave up.

Ibb & Obb is an adorable co-op puzzle-platformer with cute little characters who will be taunting you in your nightmares before you reach the end.

Its primary puzzle mechanic is a, uh... gravity line? Imagine that the ground is a mostly-horizontal line, and a character above it will be drawn downward, while a character below it will be drawn upward. Portals on the surface (or in walls) can be used to swap from one side of the line to the other, and you and your co-op partner will need to manage "who's on which side" to defeat enemies and overcome hurdles as you travel from left to right.

Gradually but surely, the game adds elements and complications which make for more brain-melting puzzles -- like character-specific portals (only one player can pass between them) and trampoline pads (jumping on them transfers your momentum to a character on the other side). Gravity management remains the core "thing" of the game, but additional foibles keep the overall design feeling fresh.

And then there are the twitchy, time-sensitive challenges, including: tandem-jumping over moving enemies, using gravity-inverting bubbles which can float out of your reach, and - what ultimately made my partner and I give up - using the other character as a platform while they're mid-jump.

Ibb & Obb does a great job of building up its puzzles' strategic complexity, and figuring those puzzles out was great motivation for most of the game's duration. Sitting and staring at the screen for minutes at a time, trading solution theories with your compatriot, is surprisingly satisfying.

The late game just cranked up its bar for precision platforming and reaction times too much, testing my appetite for masochism until it finally broke.

Good fun up until that point, though. Sincerely: this is the best co-op puzzler we've tackled in a while, despite its eventual, unreasonable difficulty.

Better than: Dyadic, Kalimba
Not as good as: Pitfall Planet
Unlike some other "relationship ruiner" games: here, you can commiserate against a shared enemy (the game).

Progress: Got part-way through level 13 of 15.

Unlike previous Bat-DLCs, (deep breath) Batman: Arkham Knight - Season of Infamy: Most Wanted Expansion doesn't just throw you into a series of arenas and ignore everything you've done in the main game. Instead, it adds optional missions to Arkham Knight's campaign.

What a thought, right? Actually "expanding" the existing game? Brilliant.

Although there are clear signs that this content was an afterthought - there's literally a separate room in the GCPD building for these villains' cells and evidence lockers - these new missions nevertheless feel like cozy companions to the game's original Most Wanted. And they employ a rich variety of Arkham Knight's mechanics, including exploration and investigation, which is a refreshing change of pace from just fighting shit.

Do I wish that there were more than four new missions? Yup. Do I wish that the Mad Hatter's obligatory hallucination sequence was longer and more fanciful? Yup. Do I wish that Mr. Freeze and Ra's al Ghul had more presence in their missions? ... A little.

But I'm happy with the gameplay and narrative quality of these diversions, I'm happy with how well they integrate into Arkham Knight's story, and hell, I'm even happy to see Killer Croc again.

Better than: Batman: Arkham Origins - Cold, Cold Heart and all the other Arkham DLCs.
Not as good as: The Witcher 3: Wild Hunt - Hearts of Stone
The "right" way to play this: is probably by having it from the beginning of Arkham Knight.

My understanding of the Arkham Knight DLC situation had been that, while most of its "story" packs were in the same neighborhood as the Harley Quinn pre-order bonus - which I previously described as a "pretty baffling disappointment" - the Batgirl-centered A Matter of Family was much more substantial and worthwhile.

That turns out to be a half-truth. The "more" part is true; not the "much" part.

A Matter of Family is set some time prior to The Killing Joke, and starts with a flimsy excuse to set Batgirl and Robin up against the Joker sans Batman. (Joker claims that he'll kill Jim Gordon if Batman shows up. Since when would that stop the Bat?) The map is open, but miniature -- Seagate Amusement Park is an artificial island isolated from the rest of Gotham, and its size feels comparable to one section of Arkham Asylum's overworld.

After the DLC's intro, you're tasked with rescuing three groups of hostages. While the map's openness means that you can go after these groups in any order -- after the first and second rescues, there is a non-optional objective to defuse a bomb at a specific location; then after the third rescue, a boss fight. So aside from picking which hostage to rescue first, the mission is intensely linear. And the whole affair is like, one or two hours long, maybe another if you seek out the optional collectibles (which I didn't).

Batgirl's moves and gadgets are all the same as Batman's, but fewer. The only "new" ability here is that some hackable objectives and traps don't require a password, that is, using the hacking gadget just finishes the hack automatically. Yes, that is a vanishingly minor distinction from Batman's kit.

A Matter of Family "works" in the sense that it delivers some of the same great stealth and combat fundamentals as Arkham Knight. Otherwise, this DLC pack is shallow, un-innovative, and utterly uncompelling. It's got more meat to it than the other story packs, but ... that's a pretty low bar to clear.

Better than: Batman: Arkham Knight - Harley Quinn Story Pack
Not as good as: Batman: Arkham City - Harley Quinn's Revenge
I kinda can't believe: that Batman: Arkham Origins - Cold, Cold Heart ended up being the best standalone Batman episode.

Midnight Protocol appears to sit somewhere between Hacknet and NITE Team 4, in terms of both realism and aesthetics. Its hacking mechanisms are less artificial than Hacknet's, though still abstracted from reality to be more "game-y." And while it revolves around a flashy node-based network visualization (similar to hacking in the newer Deus Exes), you still control everything with a command-line interface.

Unlike most hacking games, Midnight Protocol is turn-based: the "trace" countdown only ticks when you take an action. So, you can take your time planning how to optimize the CPU usage of exploit and stealth programs. There's an Uplink-style economy, where your ill-gotten money can be used to purchase upgrades -- the demo didn't show it, but there is a "hardware" tab, so I assume that your rig will scale up (as will program requirements) over time.

And the story teased by the demo is promising: you've got a dark past, there's some underground conspiracy, a hack gets abruptly interrupted by a counter-hack with ominous messaging... all typical stuff, but the quality of the writing "sells" it fairly well, typos notwithstanding.

Basically, everything I saw in this demo was good. But I'm trying not to get my hopes up; hacking games have a habit of letting me down with some combination of clunky interfaces, under-baked game design, and boring narratives. I'll want to see some post-release opinions on whether the rest of Midnight Protocol can measure up to my admittedly-high standards.

Progress: Finished the demo.

My initial concern in Patrick's Parabox was that the mechanical distinction between "push this block" and "enter the maze inside this block" is too ambiguous. I.e., you can't enter the side of a block until its opposite side is against a wall, and pushing an un-pushable block may have unexpected consequences if you're not paying attention. As more blocks and more pushing come into play, this gotcha could become frustrating.

But the demo levels didn't grow horizontal complexity like I expected -- they grew depth complexity. The last level I played had, like, four nested layers of block to traverse through. Now my concern is forgetting which layer I'm currently in.

That's just the kind of mind-twisting puzzle mechanic that I'm interested in learning more about. (And Parabox is more immediately accessible and rewarding than my peeks into Recursed have been, so far.)

I'm not "sold" on this one yet, but certainly ... intrigued. I wonder how the full game's later puzzles will turn out.

Progress: 14 puzzles solved in the demo.