Although the level-select screen makes me appear within striking distance of Infinifactory's end (for real, this time), the levels themselves paint a different picture. Like the puzzles that closed the game's first half, these final levels are work.

I've still been enjoying my time working through them, but ... each one takes an evening, now, or even multiple evenings. These last few puzzles require enough task planning and complexity management to rival software challenges I've actually been paid to work on.

So it may seem weird that I'm taking a "break" by playing Shenzhen I/O, but relatively speaking, it really does feel quite relaxing.

Progress: The Homeward Fleet

Child of Light's fairytale narrative is diluted, significantly, by its insistence on rhyming dialog. The writing is routinely stretched and twisted just for the sake of a barely-rhyming word, sucking the personality out of any given character's lines.

(With the exception of the jester, Rubella, who adorably flubs the system by using a rhyming word's non-rhyming synonym instead.)

The game's story feels weak overall, as Aurora's goal is never terribly clear. That is to say: She's trying to get out of this dream world, but how? Something about a song? And a mirror? All I'm really doing is exploring my way through the map, and continually getting distracted by side-characters.

The gameplay, though, is uniquely fascinating and compelling. Both in the sidescroller-style overworld - which has enough light puzzle-solving, obstacle-avoidance, and maze-finding to add some substance to its aesthetic beauty - and in the real-time combat.

Probably the most defining characteristic of its battle system is the timeline, and how it can lead to interruptions. Each character has a speed stat, which defines how quickly it progresses through the "wait" part of the battle timeline; and each action you select has a speed, which is how long it spends in the "cast" part of the timeline. While a character is in the "cast" part, it can be interrupted by an enemy attack.

So while you're trying to time your attacks to interrupt foes, you're simultaneously trying to avoid being interrupted yourself -- using Defend to defer your action until later, or choosing a high-speed action to beat your enemies to the punch. The timeline pauses when you're making selections, giving you plenty of time to think your tactics over.

Combat also incorporates some real-time interactivity, via the whisp, Igniculus. He's moved with the right analog stick, and using his power will either slow down a specific enemy, or gradually heal a party member. The slowing, in particular, is pretty vital in working the timeline to your advantage.

There are plenty of other quirks to Child of Light's battles, like character-specific talent trees and the element-infused Oculi gems, but most of these components are never so prevalent that they make the game feel overly complex. For an RPG, it's relatively easy to wrap your head around.

It's worth noting that I'm playing on Expert difficulty (as opposed to the default, Casual setting), which I wouldn't say feels "expert" so much as it does "moderately challenging." There are some enemies who really frustrate me, but having fought through maybe half of the game so far, I've only had to use inventory items a handful of times.

Child of Light's storytelling is disappointingly thin; but the core gameplay is solid. More than seeing the plot through, I'm really looking forward to leveling up my party.

Progress: Chapter 6

At least so far, Shenzhen I/O feels a lot like a direct sequel to TIS-100. It's focused a little less on raw computation, and more on connectivity, in order to integrate with other components; but I'm still writing assembly-like code, blocking on inputs from other processors' outputs, and wrangling tiny amounts of runtime memory.

Time will tell, but at this point Shenzhen feels like a simpler game than its forebears, with puzzles that are more "cool" than puzzling.

Minor digression, but the solitaire minigame is total bullshit. The unstackable "dragon" cards really ruin it. I guess I'm supposed to fully analyze the card layout before making a hasty play; but ugh.

Progress: "Prototyping new ideas"

While the last few levels of Infinifactory's first half had begun to feel tiresome, its second half comes out swinging, with new blocks and new environmental quirks. The puzzles built on these new mechanics feel fresh, and have really revitalized my interest in pushing forward.

The story is heating up a little bit, too, although it's still not exactly thrilling stuff. More compelling than TIS-100, but only about par with SpaceChem.

The end of the game - the real end, this time, I'm ... pretty sure - is practically within my reach, now.

Progress: Atropos Station

Flying solo, Kalimba feels totally different. I think that's for two reasons, one being the single-player campaign's "unique" mechanics -- like color-swapping, ice platforms, and anti-gravity. I'm fake-quoting "unique" because, since I didn't make it very far in the co-op campaign, it's possible that these features would have popped up eventually.

But the other reason is that, like I complained about last time, co-op's fundamental four-totem stacking mechanic is simply broken. The timing and positioning requirements are just too precise.

One brain, using one controller, can maybe manage to coordinate two totems well enough for Kalimba's levels. Hell, a one-player four-totem quadruple-jump would be hard enough. But adding a layer of human communication to it, and still expecting timing with 0.1-second precision, is just cruel sabotage.

The single-player game was just interesting enough to see me through its first third; the regular introduction of new mechanics was neat, but the demanding platforming wore me down.

As for the co-op, I can only summarize it as unplayable.

Progress: Solo, 33% (beat the first world/boss); co-op, gave up in the second level.

At about five hours, and with more collectibles to go if we wanted (but we didn't), Pitfall Planet has a fairly substantial amount of content. The game's simple controls are exercised really well throughout its run, and the stranded-little-robots premise never feels worn out.

Other than the fact that the mouse cursor stays on-screen - there's got to be a "hide cursor" setting they just missed - the whole affair is very well-polished. Even occasional physics quirks, like needing to fiddle with angles to get up some ledges, feel appropriate within the game world.

Pitfall Planet is satisfying and fun, and while it may be a bit on the "mild" side, it's hard to pick out any real flaws with it. (As long as you have exactly two local players.)

Better than: Dyadic, Trine Enchanted Edition
Not as good as: Captain Toad: Treasure Tracker (not co-op, but otherwise thematically similar)
And the soundtrack is cool: stay a while, and listen.

Progress: fixed the spaceship, 89% item collection, got the cat-ears hat

Pitfall Planet is the endearing story of two simple robots trying to repair their spaceship, by harvesting ore from co-operative puzzle-platform levels.

As in any successful co-op game, you can grab and throw your fellow robot, over an obstacle -- or off of a cliff.

The majority of the levels aren't hard on their own, but revolve around basic coordination tactics. You throw me over this pit, I grab this item and throw it to you, you use the item to hit a switch, I go up this platform to hit another switch... and so on.

That said, there have been a couple levels that stumped me and my companion for quite some time: like a giant Rube Goldberg device, or a falling floor-tile maze over a pit of lava. And although most of the game's difficulty is in figuring out what to do, there are occasional challenges of timing as well (like running away from tiny drill robots).

Pitfall Planet really excels at encouraging teamwork in solving puzzles, helped by an engaging soundtrack and a slightly-mysterious world design. (I hope, before the game ends, we discover what happened to this planet's previous inhabitants.)

Progress: 72%, 28 ore, 389 gems

In the first few moments of Kalimba's co-op, one player has one totem piece, and the other player has another piece. They need to be moved in tandem, stacked in a particular order, and jumped with coordinated timing. Simple, cool, fun.

It isn't long, though, before the game gives two totem pieces to each player. And then - given that either player's pieces usually move at the same time - both players need to coordinate stacking all four totems in specific arrangements.

I feel like there must be some aspect of the controls I'm missing, because this four-part stacking just seems ridiculously difficult to me. Maybe I'll try the single-player mode and see if it's more instructive.

Barring that, Kalimba is hypothetically interesting, but the fun is blocked by how hard these basic stacking operations are.

Progress: Gave up in the second co-op level.

The combat-oriented demo of Nadia Was Here managed to both impress me with its unique gameplay, and scare me away from the full version.

Nadia's battle system is organized into lanes, sort-of like Mega Man Battle Network or Radiant Historia. They're narrower, i.e. only one party member and one enemy can fit in a lane, but the interesting part is that - while those other games are all turn-based - Nadia is real-time, with a continually-ticking attack timer on each character.

The result of this combination of mechanics is that you can move a party member into a lane to strategically attack an enemy, and then move out of an enemy's lane to avoid being attacked. And since enemies can also move, there is an element of evasive strategy to staying ahead of them.

The practical consequence, then, is that you have to keep track of all the concurrent, slightly-offset attack timers; and continue moving around just to survive. It's a startling amount of micro-management for an otherwise simple premise.

I admire what this system was able to pull off, but have no interest in subjecting myself to more of it.

Progress: Didn't finish the demo.

Eight years ago, I pointed out a critical flaw in Trine's co-op experience: the game isn't really designed for it -- and most obstacles don't accommodate multiple players.

While Trine Enchanted Edition may be prettier than the original, it doesn't solve this problem.

Trine is still fun to stumble through with a friend's help - or alternatively, fun to make your friend stumble through by accident - but isn't quite "enchanted" enough to motivate a full run. Plus, as I once lamented on Trine 2, the "floaty platforming and shoddy hit-detection" aren't as easy to live with as they used to be.

Progress: Gave up at Fangle Forest