Basics. A floorplan is more than a layout to place things in a map. It has to be about strategy and how the players are going to take advantage of it to win. Hourences compares it to chess. From my personal experience the natural path that everyone follows is to first see floorplans without any thinking on strategy. It's all about deciding that there should be a bridge here, a tunnel there, a window on that wall, pick ups here and there, without really putting in much thought about how it's going to play. Beginners shouldn't be expected to have a much deeper insight on the strategic component.
Corridors. Hourences says that long corridors lead to bad gameplay because as one player is running straight, the other has an easy target because there is no way to escape. One tip that he gives is to use shorter corridors to avoid players being trapped. I'd add that corridors have two dimensions to keep in mind: width and length. Width, if it becomes too wide, transforms a corridor into an open area and I really don't think that there is an exact measure such as 10 meters or 5 meters to set a threshold between a corridor and an open area. Because we can populate the corridor with boxes and other objects and they may make the corridor feel wider or narrower because of context. For the purposes of a gametype that requires fast movements, placing boxes and other objects should be avoided because they block the player's movements.
Dead ends and interconnection. Dead ends lead to bad gameplay because they trap the player or because they can mislead the player, causing confusion. Unless a dead end contains some reward, there is no reason to have it.
The other aspect is having more than one way in and out of areas. To allow the player to have more options it's a good practice. Unfortunately I don't have an answer for how many ways in and out there should be. There is certainly a threshold between too few and too many. In other words, there is a fine balance between having too many choices and too few. This applies not just to level design, but game design as a whole. Mark Rosewater in his talk about lessons from making magic tells that one way to approach this issue to think about what can be removed without affecting the core gameplay.
Open areas. They sort of have the same issue from long corridors. They force the player into a position where there is no escape route. They should be left for situations in which the player can traverse is at high speeds, such as with vehicles. From a technical standpoint Hourences comments on some often overlooked issue: open areas are more prone to lower framerates due to the lack of occlusion. Sometimes artists are unfortunately burdened with adding more and more objects and details to large open areas, which are harder to optimize and put more stress on the processor. Again, I unfortunately don't have an answer for this. There isn't some golden rule that I know about on how to tackle this issue. In some cases is comes down to style and personal preferences.
To combine both large open areas and enclosed spaces is possible, but difficult to balance. Players may be very well more interested in one area than the other because it plays better, in which case the underplayed part should probably be removed. From a personal standpoint I'd say that's better to focus on one alone. Can an area be both open and narrow at the same time? Metaphorically yes. There certainly are ways to imagine how that would be accomplished. But for all practical terms, we may end up trying to be original while disregarding other more important things.
Items and power-ups. The concept that is behind item placement in maps is risk vs reward. The most basic form is by placing highly valuable items in dangerous spots, such as narrow corridors, long corridors or inside traps. Unreal Tournament used this concept extensively. This is one of the core gameplay aspects of fast paced FPS games. There is another aspect that Hourences discusses that requires a deeper understanding of levels. It's about where to place the items. When I was a beginner I would place them in a way that spreads out action, trying to achieve a balance by having as much distance as possible between clusters of weapons and ammo. The action is naturally going to be "hotter" near the "hotspots". Conversely, there is less action near the "coldspots". How do you know how to spread out things? One criteria is about how powerful a weapon is. For example: don't place two strong weapons in the same area of the map. Or don't concentrate all health in one area of the map.
Going deeper and there is a matter of functionality behind where to place items. That's something I don't have experience with, admittedly. Picking up items can be tied to items emitting sounds, which in a multiplayer environment is a way of locating players. In here I'd comment that this depends on technology itself, because true positional audio is not something that every game is going to have. In Unreal Tournament picking up an armor would emmit a characteristic sound which could be heard from anywhere, regardless of the game not having true spatial audio. Physically speaking this is completely wrong because the game is not obeying to the real physics, but the game has its excuse for that. In this case to know that a player has picked up an armour is more important than the correct simulation of the sound waves propagating in space.
Another detail that is often missed is that the placement of items shouldn't be random or based on criteria such as distance alone. In here Hourences says that forcing the player to run longer distances for low rewards is wrong. Imagine a square room with one weapon and two ammo pick ups. Don't place the weapon in a corner, while the ammo is on the opposite corner. His argument is that players should spend more time on the action and less time on running around because the item placement forces them so. I agree here.
Traps and physics. I don't have much to comment because I lack the experience to do so. What I could say is that physics is tricky because with physics we have some degree of unpredictability and this may harm the gameplay because the players may take actions that weren't intended. If I were to think on some example I'd think on walls. Imagine that players can build walls to block the path. Any player may very well build walls that block everyone from getting in or out, thereby ruining the game for everyone else. This issue would be more about concept design than related to level design itself.
About traps, they should add depth to the gameplay. Because if they don't, they are unnecessary. They are then obstacles that are nothing more than hindrances or annoyances for the players.
Cover and snipers. As a general rule of thumb, avoid encouraging snipers when the gameplay isn't focused on that. That's what Hourences says. A good example is Facing Worlds from Unreal Tournament. Both sides have spots for snipers and the players are easy targets when running on the wide open area between the bases. The key point is that both bases face each other, meaning that we have snipers to counter-balance snipers. Geometry plays the most important role here because there can be unwanted lines of sight that encourage sniping and causes many "feels bad" moments for the players.
Aural fixation. The concept that Hourences describes is about having an unique identity for many aspects of the map. The most obvious is by using textures. The not so obvious technique is by giving different sounds for each piece of the map. So doors, wooden surfaces, metallic surfaces, lifts, water, all having unique sounds. Think about a blind person. How would a blind person locate himself or herself in the game's world? They have to rely on audio and by having different sounds for different lifts and doors we can help them locate themselves and to distinguish different places from one and another. For a player in a multiplayer game aural fixation means that they have one more piece of information that adds to the gameplay's complexity at their benefit.
Spawn points. I think Hourences missed this one. Never place spawn points that make the player spawn facing a wall or in the middle of the action. This is annoying for the players.