I think a lot of Twitch is not competitive games like Dota or Overwatch and are really casual games that anyone can understand. I never played Portal 2 cooperative but can easily understand what's going on if someone is streaming it. I think people that don't play the game but watch are watching games in that genre.
There are also games in the middle, like Fortnite, that are pretty easy to follow. "What do you do in this game?" "Kill anything that moves and run to the circle." Once you understand that you can probably figure out what the streamer is doing and why.
So far I've been proven wrong, but who knows, maybe it'll turn around.
Overwatch I got sucked into immediately. I knew the basic idea; kill the red team... and the rest of the details I picked up as I played. The matchmaking did a very good job of keeping the games winnable but not boring. The variety of heroes has something for everyone; from those that have been playing FPSes since they were 5 (Soldier, McCree, Widowmaker), those that picked up Overwatch as their first competitive FPS at 30 (Winston, Reinhardt, Moira, Mercy...), or those that want something completely different (Doomfist, Sombra, Symmetra). I have made many friends in that game, even met my (ex-) girlfriend via "stay as team" after a good match.
The biggest problem Overwatch has is that they made it appealing to the hard-core gamers as well, so you have to wade through all the people in the community that complain about "no aim, no brain" because they happened to play Aim Hero 13 hours a day and believe that's the only skill in a first-person perspective game that involves killing people. And the balance of the game is such that right now, they're kind of right... people with average aim in other games can absolutely destroy with Hanzo given the large hitbox of his one-shot-kill arrows. You don't need positioning, you don't need teamwork; just get those arrows in the immediate vicinity of the enemy's heads and you win, no matter what your team is doing. They'll probably tone him down a bit in the near future, though.
I clocked about 200 hours over 3 years before hitting that point, and since then I've done another 200 over 6 months.
Regardless, I think it's one of the few e-sports that's fun to watch regardless of how skilled you are yourself.
Also, RL has become my favorite esport by far. The problem is that unless you've tried the game you don't realize how crazily good the pro players are. Things that wows me don't impress my non-RL friends at all.
Not quite watchable on streams and videos currently because first person view on 2d screen is erratic and horribly flat - but I hope games like this is the future.
And commentators can make it even more accessible; their expertise is there to help provide context and guidance.
In football You don’t need to know, exactly, which way the plays are going to go and why, but you do need to know the rough constraints of the game and what is physically possible to perform.
Some of the things that happen in dota are utterly impossible to follow unless you have, at the very least, a knowledge of the (on average 3) active abilities of the ~110 heroes.
Games like MOBAs are opaque by comparison. I can see roughly what's going on but I couldn't tell you if a team is winning or losing to save my life. In football if the guy holding the ball is out in front of the defending team and running to the end zone I can guess he's going to score a TD, but a team can look like they're getting whooped in DotA but really they're tricking the opponents and someone is making a sneaky play in a different corner of the map or someone has an ability or gear that means the situation is the reverse of what it appears, etc.
All of that is really only apparent to someone very familiar with the game.
Watching streams is "about" gaming in the same sense that tailgating is "about" football.
You are absolutely correct though. Those first few weeks I had no idea what was happening aside from the occasional head-shot.
Games like DotA and LoL have too many mechanics built into them that are just debt from their origins as a Warcraft 3 mod. They're not really intuitive or sensical to the uninitiated.
Moreover, in Starcraft if lots of different things are happening at once, it's as technically challenging for the players to keep track of as it is for the viewers. With MOBAs the action is scattered across 3 to 6 areas of the map at the same time, so it's way more complicated to keep track of as a spectator.
The Blizzard MOBA, Heroes of the Storm, was kind of designed from the ground up to be a lot more watchable. They got rid of the item stuff and moved the complexity from being focused on the character builds to more strategic stuff that's visible on the map. But it doesn't seem to have taken off that much.