The most dangerous ideas are the ones so compelling, nobody wants to admit they're bad. Also the atom bomb was pretty nasty, but that's a bit out of a weekly RPG column. Instead, let's pick one of the chocolate teapots that people keep mistaking for the Holy Grail - the idea that RPGs can hope to offer anything close to a classic DM experience. It's a terrible idea. It's not going to work. Stop wasting everybody's time.
Now, I'm not talking about dedicated tools like Roll20 or more specific ones like JParanoia here - tools whose job is primarily to connect people and handle the fiddly stuff like character sheets. I mean RPGs that want to offer both a game and a DM experience, which always fall far, far short of the dream or basic sell. Many have tried, including Vampire: The Masquerade: Redemption (AKA Vampire: The Masquerade: The Crap One), Neverwinter Nights, and more recently, Sword Coast Legends.
Between them, the various approaches have brought along just about every raw feature needed to pull it off, and time and again players have hit the same issues - the biggest one being perception of scope. What I mean by that is that, well, look at just about any procedurally generated game. As much as developers like to claim that their scale is some crazily large thing, like Elite Dangerous having a hundred bazillion star systems or No Man's Sky rendering twelve universes without breaking a sweat, in practice their scale is limited to the point where you as the player feel like you've seen the edges of what it can do. At that point, the magic of the thing is immediately lost and all that remains is the hope that the core gameplay loop can hold players' attention. Cracking skulls in Diablo 3 for instance. The quest for credits in Elite Dangerous.
If the DM has one job here, it's to try and disguise it with hand-crafted content and a human eye for the rules and systems. The catch is that even using something as powerful as the Neverwinter Nights editor to create a complex module full of wonder and whimsy, once the game starts that player just becomes another puppet of the game engine - albeit one with the power to see and occasionally twang everyone else's strings. They're not in charge of the rules, because that's the game logic, and even the simplest of engines makes it a pain to create content on the fly. Sure, you can make a module or a map in advance, but then you're still stuck with the problem of not being able to react with the speed of thought to game limitations and player freedom. Which as ever, will usually manifest in variously murderous killing sprees.
What does it usually boil down to letting you do? Add monsters on the fly and twiddling. That's about the extent of what a modern game can allow, with the result that what should be the mode that expands a game and makes it seems full of possibilities actually ends up just revealing the limitations all the stronger - how the traps work for instance, how progression is kept out of the GM's hands with loot tables and equipment controlled by someone else, how it's not possible to have a secret door unless someone has specifically coded that in advance. It's creation within a straitjacket, further tied down by interface design and balance. You're not working hand in hand with the source material to create something great, you're working for it. You're not the master of the game, you're its unpaid middle-management.
For the concept to have any chance of living up to the promise, design itself really needs to change - for players to accept a lower fidelity world, and for games to be built primarily around the DM experience instead of the individual players. That sounds backwards, but it's the level of trust required for it to have any chance; to design the user interface and game logic and expansion potential around player creativity instead of just serving up a standard CRPG where one player has a few extra tricks. It's also why it never happens. That's far too big a risk for any modern game, especially as the number of players who will have the technical skill to handle that and the willingness to fully engage with the process is far too limited to bet the farm on. Now take that and factor in that at the moment, most RPGs just don't have much of a creative community in the first place, never mind one willing to get that hands-on. Skyrim is obviously huge, but it's the exception to the rule. Divinity: Original Sin for instance has made no waves whatsoever with its tools, and however many players are currently logged into Neverwinter, it's not enough for the rest of the industry to have followed its lead.
So what happens instead? Typically, either DM mode is a glorified level editor which doesn't allow the player to actually create much except arenas, or a challenge mode, as in the upcoming Fable Legends or simple bashy-smashy games like Dungeonland. If any upcoming game had a chance of pulling it off, it was Sword Coast Legends, and... well, it doesn't. At all. Ironically, part of the issue with it is that while a tabletop RPG can make combat with a single enemy exciting and full of drama, in games it's always going to boil down to a hack-slash-hack-slash type loop, where victory is expected and interest predicated as much on grind as adventure. If DMing was going to work as a mechanic, it'd probably be by completely stepping away from how RPGs currently look and heading back to the genre's strategy roots - a focus on more intimate encounters of the kind we see in other games that borrow from the style. Card Hunter for instance is arguably a better starting point for making it work, despite lacking the glitz. Turn-based, not too complicated, focused on a few key mechanics that the computer can help with but offer flexibility for the DM and with scope for adding flavour and taking control of players... it's the only realistic way to go, if there is one.
Of course, it's always a bad idea to use the word 'never', unless you're talking about Neverwinter Nights, because otherwise people would just look at you and wonder what you were talking about. There's usually at least one game in development hoping to crack the formula, with Divinity: Original Sin 2 the next biggie planning to try its luck. One day, one of them may create something as intimate as a table of friends and as open as human imagination. It just doesn't seem likely any time soon, to the point of feeling like a complete waste of energy that would be better spent on finding more ways to advance what computers can actually bring to the table in terms of things like world simulation and depth and creating experiences that neither require a human DM, nor make players long for one to give our adventures life.