There is a big resurgence in adventure games, especially form indie developers, many emulating classic tech, with big pixels. It's amazing, and a boon for adventure lovers like me, who grew up with text adventures. But what enabled this? Easy indie development for one, and the technology for nicer palettes and voice acting, which the originals couldn't do. But I think the biggest reason is difficulty.
Adventure games went away, a strong argument goes, because they became too arcane; too difficult. The joke goes that the final straw for the adventure-game-playing public came in Roberta Williams' Gabrial Knight 3. In it, you had to place small pieces of Scotch tape on a counter, scare a cat into running past the tape -- thus depositing fur on it and enabling you to make a disguise mustache, that somehow works to trick someone. But there were many adventure game sins, committed in their quest to challenge experienced players, or eke more play hours out of the same content. The worst offender, for instance, was pixel hunting for invisible objects. Many other faults were less severe, but in combination created brutal experiences, were too little could be figured out, and too much had to be done by tedious trial and error -- combining all actions, with all inventory items, all dialog options, and so on.
Playing Technobablyon recently, among other Wadjet Eye games, the Deponias, etc. has shown what some of these new games are doing. On one hand, they can appeal broadly because they're simply easier. They're aimed at less specialized, hard-core gamers, or even at children. This has also meant a big increase in story-telling, and that can make games that are easy but still fun for a more experienced gamer, likes Chains of Satinav.
But there have also been some new games with good puzzles to solve. I don't mean the most difficult of puzzles; I mean interesting ones, which are not obvious, so provide some satisfaction, and get you involved in the story. I want to break down specifically what these games are doing to make the puzzle-going more pleasant without simple making non-puzzles. As an example I'll pick Technobabylon, since it's a very well made game of the 8-bit variety.
This means simply fewer objects in inventory, fewer locations available, and fewer stationary objects in locations. This is pretty straightforward, and reduces the combinations of interactions available to try. But the core of the puzzles can remain strong, if they're designed well.
Technobabylon does a lot of this with its episodic structure: in each one, there are just three or fewer areas to visit, each with just a few locations (screens). By the end of the game, you've visisted lots of places, but few were accessible at the same time.
The designers have also taken some pains to limit the need to shuttle back and forth between areas. Puzzles don't span them, per se: each is largely self-contained in one area; and when complete, may give you something to use elsewhere. This means there's less travel time, just for the sake of dragging things out. All they really had to do was group related objects in one area.
The number of objects in your inventory is also limited. This is achieved partly by giving some items multiple uses, in different puzzles, and also by the removal of items when they're no longer needed.
Similarly, there are few if any useless objects: everything that can be interacted with is useful, except some scenery, for color. There are no red-herrings in the sense of completeley wasted things. The only exceptions come from scenes with many indistinguishable objects, where the puzzle is to determine which is important (e.g. there are a dozen plants to take DNA samples from -- which is the one you need?).
This comes in a soft and a hard form. The hard form is to really offer hints per se when the player may be stuck, and asks for one. A common device for this now is a side-kick or partner, who you can enter into dialog with, and ask about various current obstacles. The Blackwell games tried this, though not as successfully as Technobabylon, in my opinion. A key is for the hints to be context dependent.
The soft form is more classic, and a hallmark of good text adventures: lots of feedback for player actions. This means that whenever you fail at something, you have the possibility of learning something; you know why you fail. This is perennially a weak spot for games with voice acting, as it's hard to record all the possibilities necessary; but most graphical games have been weaker at it than text ones. But when they make the effort, it's a natural way for the player to learn about the environment.
This is easy to detect in retrospect: when a puzzle is finally solved, and you're ready to move on, did what happened actually make sense? In a lot of games, the answer is no, or a shoulder shrug. Because some objects were combined, but why they produced the desired result isn't obvious. Or, almost as bad, it's not obvious why some slightly different combination of objects couldn't have done the same thing. If a rags stopped up the drain, why not the old newspapers?
One note here is consistency; but it's really consistency about the physical nature of the world, and the physics or other dynamics involved. Objects should be helpful at solving puzzles because of their properies, not because of their names.
Related to this is the ability of the player to infer what those physical properties, and abilities, are. Some uses of an object are obivous. But the obvious doesn't make a great puzzle. Good puzzles are one step removed: often you use the object in an unusual context, or you use one of its less obvious properties. The properties have to be inherent to the kind of object, or be described when you examine them, or try to use them elsewhere. Otherwise it's just guessing.
The current crop of games are better about requiring just one mental leap to repurpose an object. The exception is in longer multi-part problems, but the good games still break these into parts, and give positive feedback along the way. So in part A, you can discover that the knife makes a good crowbar, and in part B combine your "crowbar" with something else. This means you're not trying to be MacGuyver, creating elaborate and improbable inventions in one fell swoop.
There should be a place for hard games, with hard puzzles. But I personally welcome these more approachable, story-driven games. Their puzzles are easier, yes, but they also make more sense; they fit better within a narrative; they depend less on tricks and annoyances. This is especially welcome if it makes adventure gaming popular again, adventure games profitable again, and thus enables more adventure games to be made.