First off, bwahaha, I just milled a priest. They got 2 clerics up at the same time against me spamming weak shaman tokens and minions. Because I'd end up with the healing totem, I could injure multiple minions, and then mill 2-4 cards out of their deck.
Once their hand was full, they lost Shadowreaper Anduin, a Shadow Word Death, an Eternal Servitude, and possibly both Holy Novas. Then they surrendered at around 28 HP. =P
Now, let's address the 50% thing. I've said this previously and I'm not sure anyone argues against it:
The 50% thing is just any elo system ever though. If you lose a lot, you will get paired up with worse players because your elo has gone down. If you win a lot, you will be paired with better players because you deserve a higher skill bracket.
If you always have a high win rate after like 100 games, that isn't fair for the people playing against you. Why should they be punished by being matched against someone who is expected to beat them on average?
The thing is that this works over a long period of time but can be frustrating in the short term for people without a lot of spare time.
Also the Overwatch example isn't great because players can have good stats but still be a poor team player or have a poor attitude -- stats are an easy way to feel like one's teammates are to blame, and I know that from experience.
If Hearthstone has a hidden elo, which I believe is a common setup, it will try to get you to your skill level. What does your skill level mean? You have even matchups. If you win most of your games, doesn't that mean you're in the wrong skill bracket? You're stomping people, which means you move up and they move down.
Why would it be rigged against you specifically again?
I'd argue not that it's rigged against Karsticles specifically, but that what deck you play (maybe the class?) is somehow factored into the matchmaking algorithm. I've done similar results where I did 7 games with a class, got a certain opponent breakdown (a ton of enemy class X for example). I then did 7 more games with the same deck, almost exactly the same opponent breakdown.
I then did 7 MORE games after switching to a different deck and immediately requeueing, and I abruptly got a different opponent breakdown. Same time of day, same rank etc.
Is that too small of a test to be statistically significant? Sure. But I think it's at least reasonable to suspect that Blizzard considers what deck and/or class you're playing when they try to find a fair opponent. You could be an amazing Warrior player but really mediocre at Priest, so if you switch off Priest you might get harder matchups.
Is Blizzard's algorithm advanced enough to say "That's a Pirate Warrior deck rather than a Taunt Warrior deck and therefore we will pick certain archetypes to counter it"? I would be more surprised by that, but there was at least one streamer who thought including a low-winrate card (wolfrider?) affected matchmaking and gave him easier games.