Game development is not an exact science. It is not possible to look at the design of a game on paper and determine whether the game will be successful or even fun. Experienced game designers will be right in their decisions more often than inexperienced ones, because it is possible to build an intuition around game design decisions. Still, the only way to verify a design decision is by building and testing it. This is why an important activity in the games industry is prototyping.
Prototyping means that you take an idea, and take the quickest path to implement it. Sometimes that means using a simple drag and drop editor, other times it means writing some quick code. The end goal is the same: to get the very minimal product that can be used to test your idea. Prototypes are important to verify game design ideas, because as explained before: it is impossible to always make the right decision based on intuition alone.
A special type of prototyping happens in game jams. Game jams are contests where contestants have a limited time to build a game. The time limit forces you to focus on a central game mechanic that is fun, and selling that. There is usually a theme the participating games need to be based on. These limitations help inspire your creativity, because sometimes you have to think out of the box to make something fit with the theme.
The reason for this long introduction to the subject of prototyping and game jams, is because game jams expose a very important core truth about games: if your core mechanic isn't fun, you're game isn't. You can make beautiful graphics or an awesome soundtrack, but if the game isn't fun, nobody will play it. If you participate in a game jam, you usually only have 24 or 48 hours to build something. This really forces you to focus on the "fun" aspect of games. Because the process of building a game is so condensed, it is easy to see how it comes to be fun. I have learned the hard way: if your game isn't fun to play after a few hours, you're unlikely to fix it. The one successful game jam I have made, I completely redesigned after the first 24 hours, and I finished early. It was very simple, but it was fun.
This core truth expands to all games. Think of any successful game, and try stripping it down to the very minimum. Mobile games work particularly well for this. In its very essence, Candy Crush is just a match-3 game. If you were to implement a very simple match-3 game, it would still be fun to play. If you don't believe that simple games can be fun to play: threes (better known as 2048) is almost as simple as it gets!
This strange interaction makes it so that the development of a full game can easily have 5% of the work work on the basic game mechanic, and 95% fluffing it up. That initial five percent is like the foundation of a building though: if you don't get it completely right, no matter how many pretty bits you add to the building, it will fall over or sink into the ground. Games are the same: if your core game loop isn't fun, no matter how pretty you make it, nobody will play it.
Does that mean that those other 95% don't matter? No, not quite. How long would you play Candy Crush if it were really just a match-3 game and nothing more? And while shooting people to score points is a fun mechanic on its own to many people, for every successful FPS, there are at least a dozen that fail. In games such as RPGs, the content and lore are usually as important as a good core game loop.
The core game mechanic is not always just about the exact mechanics either though. Racing games for example are very sensitive to feeling right. If you strip away the graphics, sound, and car customization, you're left with the core of the game: the racing. If the controls of that core don't feel super slick, there is no way to redeem the game with some nice dressing up. Gun play is another big one in shooters. You can get away with bad gun play by introducing interesting other mechanics (did somebody say Fortnite?), but if you pick a mechanic to be the core of your game, you better make sure it feels just right, and sometimes, that takes up a large portion of those 95%.
Many large game developers tend to stick with a proven game concept (yes, I am talking about you, Call of Duty) and focus on that fluffing and getting the feeling right. The last few years have seen a rise of smaller indie games. Indie games usually have less production quality than your average triple-A game (though there have been some notable exceptions on the triple-A front recently), but tend to be more out there with their core game mechanic. I admittedly don't have a complete overview of industry trends, but I have the feeling that indie games are gaining in popularity. This in turn has led to some larger developers taking bigger risks in their game mechanics.
All of this just confirms the first lesson of game design: a poor core game loop that has a beautiful soundtrack, tight controls, looks gorgeous, and has a banger of a story-line, still leads to a bad game. If you want to make a good game, getting that first 5% right is the hardest step. That is why the best game designers can let go of their ideas if they don't work. Fail fast, fail often, and one day you'll find the perfect idea that just clicks. When you do, it feels majestic, and all those rejected ideas will have been worth it.