We're now reaching the tools that all of the previous tools were supposed to let you avoid, for the most part. However, it's important to realize that UI elements and other in-world helpers are not inherently evil (even though they can be abused and misused.) Many games, especially open-world ones, could not exist without UI markers or maps (and even in our day to day lives we tend to need some of those helpers), and a well-designed UI can actually add to the game universe rather than distract from it. However, they can also come at some costs, such as distracting, lowering immersion, creating a feeling that the game is about following the carrot, reducing the sense of accomplishment… It's an art, and the balance needs to be figured out based on the complexity of the game, its linearity, its target audience…
It is generally a good idea to design levels that can work without any extra UI elements, and to add them on top later if the game needs them. Otherwise, the UI elements can actually hide fundamental problems with the game or level design.
As repeated a few times already in this article, it is fine to tell the player what to do, but in most cases you should avoid telling him how to do it. This rule should also apply to UI elements. Using a radar, compass, pointer arrow, marker visible through walls… can do the job to show roughly where to go but not how to get there exactly, leaving some room for the player to figure it out on his own.
You can help the player form intentions by highlighting objects or characters with markers, silhouettes… That's the equivalent of making those naturally pop out of their surroundings via the contrast tricks mentioned previously. It's obviously less "elegant", but in complex games where a lot is going on at once, it can be necessary. Not having a marker or silhouette on a character you need to follow in an open-world game full of wandering NPCs would make it hard and potentially unfair.
Maps allow the player to, for the most part, not have to worry about mentally mapping the game world or level. They have the advantage over maps embed in the world that they can show both the player's location and that of all of its targets, allowing the player to figure out how to get there.
Usually when you introduce a map in a game, it has the effect of switching focus from finding one's way through the environment, to figuring out how to deal with challenges along the way. Games with parkour or driving get away with it better than others because the act of navigating can itself be challenging and rewarding, so finding one's way doesn't have to be deep.
What I like about maps though is that they are usually off by default and you must open a menu to access them. That makes them optional, and allows players to try to do it on their own, while knowing they have a solution in case they get stuck or lost. However, this only applies if the environment itself is also giving clues, if designers rely too much on the map, then players will be forced to use it too.
Highlighted paths are useful in the same circumstances as we need them in real-life, usually when trying to find a quick route to a destination in a vast open environment or a network of paths. However they do take some of the responsibility away from the player when it comes to navigation, and do not encourage the player to pay any real attention to the world he's traveling through, which in my opinion is a wasted opportunity (although I understand the need for it in some games.)
As with UI elements, it would be rushing to conclusions to say that cut-scenes should not be used. The thing is, many games are full of cut-scenes and still loved by millions, and without them those games couldn't tell some of those stories that people enjoy so much. Using cut-scenes does not make a bad game, although bad cut-scenes and bad use of the potential of cut-scenes can damage a decent game.
I am no cinematic expert, but as a player and designer here are some suggestions to make sure each cut-scene serves a purpose besides telling a part of the story or showing a cool thing happening. To take with a pinch of salt of course as they are mostly opinions.
Cut-scenes allow, amongst other things, to present the environment in ways unavailable to the player (e.g. bird's eye view, new perspectives…) and to showcase important elements without risk of the player looking away at a critical time.
Try to shape those mandatory cinematics to do some of the work of letting the player form intentions, flesh out his mental map, connect things in his mind… to lighten the load of things to communicate during gameplay.
Since as the name implies cut-scenes often cut, it's important to try and connect the player's current position or knowledge of the level to what's going on in the cinematic. If you cut to black and show a new area during the cinematic, players have little ways to connect that area to those they have mapped. If you start from the player's last position and move on to the new area, it's better. For instance, showing an ordinary door open in a cinematic with no reference point to the player or a known part of the environment will only confuse him.
Using transitions out of cinematics can also help greatly with that mental connection, for instance in the example above, if the cinematic ends with the camera moving back to the player's view, all the while facing that newly-opened door, the mental link is established without the cinematic having to be really changed.
It's not always so easy to communicate a goal clearly to players who may be screwing around and going after their own goals. I'm not advocating adding cut-scenes to set those goals, but that cut-scenes that need to be there anyway can be a great vector to do just that. Establishing a goal or threat during a cinematic can reduce the amount of tricks you have to use during gameplay to get the same point across, and they can also help preserve established goals or threats.
The longer you extend the interruption of the player's control, the more his immersion can drop, which results in shaking off mental mapping, tension, emotions like fear or anger, etc. It's often a good idea to keep any cut-scene that bridges two related gameplay sequences as short as possible, so the player's mental state is preserved, or you may have to re-build that mental state.
The opposite is true though, if you want the player to get out of a certain mindset, a long transition cut-scene can have that effect. In the recent Call of Duty games for instance, high-action sequences feature short and intense cut-scenes that maintain the player's excitement and the feeling of urgency (and also often follow the previous point about establishing threats and reinforcing goals). However, transition cinematics (loading screens) are much longer, more high-level and disconnected from the player's emotions, and thus cool down that urgency, so the next level could start with low-intensity pacing without being jarring for the player.
This is not really a tip on how to use cinematics to your advantage, but a general point I feel strongly about. I suspect that most of the bad reputation that cut-scenes have originates from not adhering to this principle. I don't mind a short and sweet cinematic to bridge a transition, communicate a goal or let me know a character better. However I do mind when I feel like it could have been something I had control of, that I was robbed of an opportunity to be challenged and have fun (e.g. it involves any actions that I regularly perform during gameplay.)
For instance, seeing my avatar crouch near a door and talk in his radio for a bit, then get up and step through the door… I could have done that myself, if the setup had been tweaked to discourage rushing in and so the timing of the dialogue worked. Or seeing my avatar pressing a button to call an elevator and entering it… Really, I could have maintained control a little longer, the extra camera angle and custom animation are not worth the loss of control. If that's because something happens when the elevator doors open, fine, but consider starting the cut-scene at that point then. Worse, seeing my character perform a cool action sequence like running from bullets and vaulting over stuff and jumping in the nick of time into a safe spot. Why couldn't I be the one doing those cool actions? An exception to the rule would be to use cut-scenes to skip tedious parts of gameplay, but I guess the question then becomes why have those tedious parts in the first place?
However, I understand things are not always so black & white, that cut-scenes are often used to hide loading screens and transitions, because going for a cinematic experience is hard without cinematics, or sometimes to cover a gap where a certain gameplay mechanic was supposed to go but was cut from the game, and I know it's not often easy for a level designer to have his say in the use of cinematics in his levels.