What's an MMO?

What the devil does 'massive' mean for an online game? These days it's no more than a marketing gimmick, something anyone slaps onto any product to give it that appeal, you know, for the kids?
The easy answer that I've seen trotted out by any player who doesn't actually know what to say is 'lots of players on at once'. My response to this: you're suckers, and being treated as such by game companies. Congratulations, you're part of the problem. Ten chess players playing on five boards are playing five different games even if they're in the same room and a thousand players playing fifty separate instances are playing fifty different games. They are not playing the same 'massive' game.
When the first of these products came out, Ultima Online and Asheron's Call and the like, they were indeed trying to offer something new. The internet was full of multiplayer games. We could laser each other to death on 20-player FPS servers, blow up each others' bases in 6-player Starcraft games or team up with a couple of other players in cooperative PvE games like Diablo. Calling a game 'massive' was meant to symbolize something more. These games were meant to be the all-encompassing escapist fantasy of cyberpunk scifi, an entire brave new world into which we could plug ourselves. It was meant to be the matrix.

The key element here is interaction, and I'll stick with my chess player comparison for a bit. The ten chess players are not playing one 'massive' chess game because they have no potential interaction with each other. If I take your pawn, this action does not affect the other four boards. Even in a tournament, every match starts fresh.
In order for a computer game to offer interaction among thousands of players, it required more permanence, a stable, persistent game world which the players simply enter and exit, leaving their mark upon each other. Every action you take would, in this scenario, have the potential to ripple out through a butterfly effect through the entire game world. Say you're at the market trying to buy a new piece of leather armour because your clan is about to pick a fight with another one. I steal your coin purse leaving you bare-serked. Because of this you get killed in the fight, losing a crucial strategic location, thereby allowing the enemy to invade your castle, and your entire fifty-person clan is now homeless. Way to go, hotshot. It doesn't stop there, though. Your enemies, emboldened and enriched by their victory and plunder, continue to grind the entire world under the iron heel of... well, so forth.

Interaction in a virtual world just as in the real one can take many forms, but they must all have repercussions. Direct confrontation, trading, resource-gathering, everything should impact other players either by providing or denying them access to something, taking something they have or setting them back in some way. In the currently popular marketing scheme, nothing has any effect. Players going into separate instances are just like the chess players in my example: they don't affect each other. You go in, do your thing, and when you're done the instance closes behind you and nothing you've done has done anything. This brings me back to the reason I called you (us?) all suckers at the beginning, and that'd be the monthly fees.

A true persistent world is more costly to maintain. I know nothing about computer science, but I can only assume the constant interaction between hundreds of players all in one virtual space requires better hardware than separate twenty-player servers. With every action having repercussions, the game would require constant arbitration, which means the company must keep game masters on staff. Balance becomes more and more important as players can affect each other. The slightest advantage will become a flavor of the month, and this means the company must be constantly making the appropriate changes, so it cannot simply release the game and leave it with a skeleton crew of GMs. All of this high maintenance justified the new subscription business model. Suddenly companies saw the chance to sell more expensive products, but their next logical step was to cut costs. This meant removing precisely the player interaction which required all that maintenance. Removing PvP was one solution, as this limited the effect of player actions. Trivializing everything removed the need for arbitration. Death no longer costs you anything, and being the first on site to kill a monster and loot it doesn't matter because the monster pops right back up a minute later. The biggest change, however, was instancing, which brought the number of players involved in anything back down to pre-MMO levels and pre-MMO costs. The monthly fees stayed, though, as did the 'MMO' catchphrase plastered on every game ad down to three-button browser games.

So now, we're paying MMO costs for what is basically Counterstrike or Diablo.

______________________________________________________________
 edit 2017/04/16
(mostly commas and capitalizations, but also the following)

At the time I wrote this, the long-term cash flow of a subscription model was gradually giving way to the immediate cash grab of digital rights management as the principal motivation for putting every possible game online. The MMO label was a reliable way to make customers swallow the otherwise perennially unpopular DRM scrutiny, but it has led to a reliable observation that any title which consists of 95% single-player level and gear-grinding has no business being online. As keeping customers under constant surveillance also facilitated datamining and market manipulation, MMOs and MOBAs joined in the brave new frontier of completely hollow dress-up games.
More recently, some developers have been marketing mixed online / offline RPGs like Shroud of the Avatar, cutting costs further by reducing server use while still demanding regular log-ins.
The MMO label, however, seems to have finally been poisoned for the masses, the sheer tedium of WoW-clone gameplay sinking in at long last after fifteen years of the same crap. Funcom de-branded its big-budget flop The Secret World from an MMO to a "shared-world action RPG" when it was relaunched as Secret World Legends.

From the other angle, many smaller developers are promising to bring back the 90s by reinstating the logical MMO expectations of a fully interconnected player-driven world, like City State Entertainment with Camelot Unchained. Or Dual Universe, Star Citizen and a slew of smaller titles. It remains to be seen whether anything comes of this, as similar attempts over the past decades have yielded vaporware (Dawntide) or degeneration into rampant legitimized cheating, zerg guild pandering and other exploitation (EVE, Darkfall.)

For now, two things seem to have come of the industry's attempts to force DRM by MMO-ing every possible game.
1) CD Projeckt has been doing quite well for itself with its DRM-free GOG.com distribution service, despite most of the industry refusing to deal with them.
2) The most popular online games after World of Warcraft's hey-day have been either FPSes or small-team PvP games like MOBAs, TF2 or Overwatch, a bastard of the two concepts. These basically embody the best features of WoW-clone class-based team gameplay with the hollow pretense of "massiveness" stripped away.