I think about games a lot.
 
Bioshock Infinite is much more linear than any of its predecessors. The flow reminds me of Half Life 2 more than any prior *Shock game and I wonder if it might have been a better game if they'd gone all the way. But they didn't.

Later in the game there's a few more explorable hub areas, but even then there's a ghost-town feel if you go somewhere before the plot needs you to. It's a little jarring after being otherwise rewarded for exploring every nook and alley.

 
Anyway. There's a trick they use to disguise the linearity that's quite interesting. It's this layout:

When you leave an area, you'll often find that the corridor forks into two, both of which have a sharp corner that prevents you from seeing down them. But if you backtrack after following one to try the other, you'll find that both corridors go to the same area. 

It's a truism than non-linearity usually equals skipped content. By keeping these corridor sections short, they minimise wasted level design. But by knowing that there's a corridor you could have gone down but didn't, it makes you feel like progress is down to your own choices.

Except that The Wizard trains you to look behind the curtain. Bioshock Infinite consistently rewards you for exploring with tangible in-game rewards, and offers no benefit for pushing ahead quickly. The same trained behaviours that give you delicious voxaphones, unstable tonics and loot, also reveal the illusion of your choice. And it was a bit frustrating every time I fell for it, going back and expecting a new area to explore/pillage and finding nothing but a switchback.

And it's interesting to me how a clever technique to make the levels feel more open, clashes with the way the game encourages you to play. 

It's more successful when the areas look like this:

 
It's almost the same trick - whichever way you go you'll reach where you need to be - but on a much larger scale. The corridors are long enough to be interesting, and exploring the alternative route will have enough action to be rewarding. These areas are also large enough that the circular layouts don't feel like corridors. And when they're used as hubs, with different exits unlocking one by one, you might be approaching the action from a different direction each time, which wrings more mileage from the same scenery.

Another game is Darksiders 2. It has a lot of mini boss fights. There's a small cutscene where a giant monster makes a dramatic entrance and then you have to murder them. But it does something clever.

Sometimes you'll get the little cutscene and slay the giant undead scarab hulk or whatever, but then two more of the same monster jump down, sometimes with minions. Holy crap - the game just served you up one as if it were a boss, and now you need to take on *two*? And it makes you feel like a badass when you win. It does this a lot, but it always feels good. 

Of course they're not bosses, they're just a new monster type that then gets added to the normal rotation. But what a way to introduce them! Great showmanship. 

And clever too. The first fight is exciting because of the uncertainty - you don't know what this monster can do. It's also a teaching aid that allows you to learn its attack patterns in a simple one-on-one fight. The followup monsters require you use those skills in a more dangerous and complex situation.

Like the *Shocks, the Darksiders games also reward you for compulsively exploring every corner. It's the Zelda thing where there's collectibles everywhere if you look for them. And finding collectibles is fun, but as a completionist I find myself compulsively looking behind me after every doorway, and carefully inspecting the ceiling of every room. Which is slow and not enormously fun.

Here's a thing:
 
The game is more fun to play when I'm not actively looking for shinies. And when I *do* find a shiny, it's more exciting when I haven't been compulsively checking every square inch. Obviously that would be the most fun way for me to play, except for the gnawing anxiety about missing shinies. And that's interesting to me.

Compare it with Minecraft. I get that same thrill when I see an emerald or diamond block. The layout of caves usually prevents 100% exploration - there comes a point where it's more efficient to find a new cave to explore. But I don't get that horrible feeling that there's diamonds left unmined. I feel no urge to strip-mine the map chunk-by-chunk down to bedrock. Why is it different?

I suspect that it's because Darksiders/Zelda/Bioshock are finite. I know that if I miss a voice recording or health upgrade, that's something lost to me forever. The game will end and I won't have gotten full value out of it. In Minecraft I know there's always going to be more diamonds in the next cave, so there's no stress. It's impossible to complete a collection so I feel no urge to try. It feels so liberating.

Of course the problem is one that exists inside my head. I couldn't argue in good faith that one way is better than another (or even perfectly analogous). But it's fun to examine these things.

Icon style guide

I've made a lot of icons, and I've tried to keep them to a particular style. That's a little pretentious - my style is a crutch to hide a lack of ability. Still, it's hard to draw over 1600 little pictures without getting better at making them, and so I've suffered quality-creep. The ones I make now are often more illustrative and less symbolic. I include details that, back in the first hundred, I might have abstracted.
  Every good icon I make makes its predecessors look worse, and as I get better I'm making more better icons. 

Still, there's some consistency. From the start I developed a set of unwritten rules about how my icons should look. There's a few exceptions, but they're almost all due to oversight or moral weakness.

White on black
Everything is white shapes on a black background. If I want to do something in black, outline it in white, or put it over a white shape. When actually using the icons for something I can invert them if necessary, but for consistency when browsing the base icons, they're all white on black.

Sharp black and white only
It seemed silly to bother with colour permutations (red heart on black background, black heart on yellow background…) for the icons when that's so easily accomplished in a paint program. I can colour, paint over, use as masks on a gradient or anything else. The simplest possible base icons enables the greatest freedom when using them.
This is why I'm always a little annoyed to see them used as-is; just white shapes on black squares. They were never meant as ready-to-use icons. They're stencils.

Border
Nothing touches the edge of the frame. There's an invisible 1pt no-go zone. Again, this is just for consistency amongst the base icons and ensures I have some bleed when I cut out the icon to use elsewhere.

Fill the frame
The icons are always as large as I can get them within the frame. They'd be much less useful as a set if they were all drawn to different scales. If I need a small heart (for example) I can just scale down a large one. If I really wanted a small heart icon I'd give it some kind of framing detail.

Left to right
Consistency. If an icon has a direction, it's going left to right. Up vs down is more flexible, but when it's isn't common sense I tend towards top-to-bottom.

No lexical symbols
These are meant to be illustrative. No letters, numbers, currency symbols or overly specific symbols. If I needed a letter or a number, I could raster it from a typeface. And if I do one digit I should probably do all of them. Likewise letters. Hash marks are illustrative enough that they feel ok. There's an omega too.
 
Not too specific
I don't want twenty different flavours of axe. If I need a specific icon for a specific variety of thing, I can make it to suit. One or two axe icons will fit most needs. And the icons are supposed to be symbolicaly illustrative rather than perfectly specific anyway.
I've bent this one a lot, but I ration myself. There's a lot of different swords and botles for example, but I only do them every now and then so they don't take up too great a fraction of the total. And I try and make sure they're all sufficiently different from each other to be worthwhile. I'm not interested in doing ten different helmets with only slight differences.
I suppose this one includes other people's trademarks and logos too. No Samus helmet, no buster sword and so on. For several obvious reasons.

Not too abstract
Counterpart to the previous rule. If I need a circle I'll draw a circle. Likewise an arrow, a diamond or other simple geometric shapes. The icons are meant to be illustrative and there's a point at which a shape is too simple to illustrate anything. I've done a few miscellaneous weird shapes, which could be glyphs, thingamabobs or jewellery. The problem is that they're so easy to do I could easily churn out a hundred of them. But even if they're pretty, they're meaningless. There's not much difference between having a half dozen available and having a hundred. They're not challenging to make and usually a waste of time, so I don't let myself do many.

Definition and resolution
This is the most important one and the biggest pain in the arse. There's a minimum line width of 1pt. And that includes negative space - two separate white shapes need to have a 1 pt separation between them. This is my minimum "resolution" for detail. No bumps smaller than 1pt, no shapes smaller than that.
There's necessarily a little leeway where two shapes come together to a point, but I stick to this rule hard as I can. It puts a hard limit on the level of detail I can put in and keeps things consistent.

Line standards
Lines get squared-off ends. If the icon really needs it they might get a taper. But usually not. And no taking the piss either - If there's a taper then it should be a short taper. Sometimes there's a drawn element that is only the width of a line, and then I might round off or taper the ends but, again, usually not.

By the way 
The coloured icons in this post are just sloppy five-minute jobs to make a point.