My take on this is that RTSs are generally unpopular because of their long feedback loop. In an FPS game, or even a modern MOBA, the decisions you take result in an almost immediate death (yours or the opponent's). If you're doing something wrong, you'll know pretty much immediately, rarely will you have to wait more than a minute between doing something wrong and getting punished for it. This also means that you tend to know what exactly you did wrong, maybe you shot too soon, maybe you aimed wrong, maybe you didn't jump enough, etc. In most game genres you go through encounters quickly: spawn, die, repeat until you get good.

RTSs are almost unique among game genres in that you're forced to make decisions quickly but the results of those decisions don't become apparent for anywhere between 10 minutes to an hour. A single "encounter" takes a good chunk of your time and your reward for that time can often be that you get to watch for 10 to 15 minutes as your best laid plans are ripped apart and everything you've built up is methodically destroyed, and at the end of it all, you can't even be sure what is it that you did wrong because you made so many decisions without seeing their immediate result. Add to that the fact that most RTSs are pretty unbalanced at any level above absolute beginner. So in the end, you're left wondering if the other guy won because you built too many kbot factories too early, or because you didn't scout enough, or because he rushed you with Flashes and those are overpowered? Did the map favor him? How do you counter a Flash rush? All those questions crop up and are really hard to answer. This makes for a frustrating experience for new and medium-skill players alike.

That said, I don't think RTSs got less popular with time in an absolute sense, I think they only got relatively less popular because gaming itself has become more mainstream. It is my belief that it takes a certain kind of personality to enjoy RTS games, which represents a certain percentage of the total human population, which is much less than the percentage of the population that can enjoy video games in general. As gaming became mainstream, the number of people enjoying RTSs stayed roughly constant, but as a percentage of all games it became smaller and smaller, thus it made no sense to keep developing RTSs. Multiplayer RTS has always been a niche genre that few people enjoyed, but when gaming was also niche it still made sense to make RTS games.

Now, when it comes to the question posed in the article, why did Blizzard's RTSs, Brood War and later WC3, got popular and stayed popular? Some people will say is that it's because it's balanced, quick and varied, but Starcraft didn't start out balanced, and it didn't get balanced until map makers learned to make balanced maps. In a sense, Starcraft is still not really balanced because you can easily make a map where one race will dominate (just try to find an Island game anywhere on ladder and you'll see what I mean, for example).

My theory for why Starcraft (and later WC3) succeeded where the others failed comes down to two things: Use Map Settings and Battle.net. I said earlier that few people enjoy multiplayer RTSs and I stand-by that, and if you spent any time on Battle.net in the early 2000s, you would see that most people didn't play Starcraft the way it was intended. UMS always had a lot more games going than regular Melee, and a large percentage of regular Melee was 7v1 computer or Faster Map Ever, No Rush 20 minutes or other things that changed the base mechanics of the game to make it a lot more forgiving. The amount of people playing actual, competitive 1v1s or 2v2s was tiny in comparison. So why didn't all those people who didn't really like RTS stuck around in an RTS game? Well because their friends were there, and all the other UMS maps were fun, if you got stomped in a bunch of 1v1s and got frustrated, you could go play some Bounds, or a Cannon Defense. Maybe an RPG or an Open RPG map, maybe The Thing, maybe Aeon of Strife for that early MOBA experience. Everything was available within that same game, and the best part was that all your friends were just an /f m away, later on even if they were playing Diablo 2 or WC3. It was pretty seamless, and other gaming services couldn't really compare until things like Steam and Xbox Live came about. You could stay in the game without ever touching the RTS portion of it, and I know a lot of people who did, and you could meet people and become friends without ever needing another method of communication besides Battle.net, all from within Starcraft (unlike most other gaming services of the time). Somewhat ironically, what made Starcraft (and later WC3) so popular are all the parts that weren't an RTS, it was a do-everything game that had a little something for everyone. All the balancing and the e-sports came later, after Starcraft had already gotten big enough and gained enough of an audience for people to learn to make balanced maps. Brood War was released in 1998, and the patches where they finally nailed the current balance came somewhere around 2001 (1.08), and by that time it was already pretty popular. I'm not sure Starcraft would have ever gotten to 1.08 without UMS and Battle.net.

This got a little long-winded, because I've spent considerable time thinking about this very topic, but my main point is that RTS (particularly the competitive multiplayer kind) as a genre has always been niche and will remain a niche genre, and the wild success of Starcraft and its successors is specifically due to the fact that it had a lot more to offer than just RTS.