Matchmaking feels terrible, but I don't think it's entirely a disparity between team skill so much as Riot has (intentionally) made the game MUCH more snowbally and so most games are effectively "over" much faster than before. People like pointing to "League of Graphs" to show that games are basically the same as they've always been, here's stats proving the core difference in what the game was vs what it is now. The following are comparisons between Playoffs (Quarters thru Finals) for Summer 2015 (when League arguably "peaked" in popularity) vs. Spring 2019 (most recent playoffs)
Average Gametime: 39:12 mins
Longest game: 55:09
Shortest game: 27:18
Median game: 39:00
Percentage of games where both teams took at least 4 towers (broke beyond the front line): **41 %**
Percentage of games where one team took 1 or less towers: **6 %**
Average Gametime: 34:05
Longest game: 45:20
Shortest game: 24:03
Median game: 32:52
Percentage of games where both teams took at least 4 towers (broke beyond the front line): **23 %**
Percentage of games where one team took 1 or less towers: **27 %**
In Summer 2015 playoffs, approx 17% of games went longer than the longest Spring 2019 playoffs game.
In Spring 2019 playoffs, approx 23% of games were SHORTER than the fastest Summer 2015 playoffs game.
Summer 2015 playoffs had ZERO games where one team didn't take any opposing towers, Spring 2019 had 3 such games.
Basically, based on pro-stats (which I admit does not fully translate) you'd have maybe **1 in 20** games where your team got so smashed you could barely take a single turret, now that is **1 in 4**.
Even at the highest level, your average game is 5 mins faster and the disparity between the winner and loser is much wider. And the average game would probably be even faster if pro players ever forfeited LCS games, as happens all the time outside pro-play.