I can't really watch this in its entirety right now, but trickle down is generally the only kind of balancing that ever makes sense in a competitive multiplayer game. This is long and a bit design heavy so if you're not interested feel free to skip.
The act of game balance in a competitive game is from a design perspective essentially the act of managing unfair scenarios, because the "fun" of (competitive) games arises when players either win from a disadvantage or have the knowledge to create/recognize an advantage and act on it.
Tons of (generally less competitive) games will create this sense of unfairness through randomness and manage how unfair that randomness can be, but in a competitive game you need very little to no randomness because you place a bigger emphasis on mechanical & mental (gamesense) skill allowing unfair scenarios to more naturally arise than something like, idk, Hearthstone. Competitive TF2 itself has very little randomness remaining for instance - random fall damage and bullet spread on miniguns/pistols/revolvers/syringe guns being the heavy hitters.
Balancing in the sense that everything is fair would leave a game pretty homogenized/boring, but you can make a game "unfair" but give players all the tools to utilize. Your goal is just to make sure that nothing is so unfair that it dominates the meta and ruins all other options or to make nothing so useless it never sees play (although this is preferable compared to the former). Having niche or situational strategies is exceptionally rewarding for players, as it goes back into that mechanical/gamesense skill.
In 6s this is most obviously seen with offclassing, knowing when to offclass and how to utilize those offclasses rewards player knowledge, teamwork, and decision making. When you go Spy, backstab the enemy Medic, and that decision wins you the round it feels very rewarding and good. Having the gamesense to recognize that the other team sacced in for a Spy and beating that swap is conversely very rewarding and feels good. Spy is a situational class, but that's what makes him interesting and hype on casts when he shows up, if Spy was "fair" with the rest of the classes - all equally viable - he would be nowhere near as exciting because it's no longer a knowledge/gamesense choice combined with mechanical teamwork to attempt the stab, it's just "yeah we feel like running Spy coz we can all the time now".
But you can look at another popular game like Chess, which is a game draped in unfair scenarios even if the game overall is relatively equal since both players have access to the same pieces and the same amount of them. The pieces are all different power levels and how they interact through various board strategies creates a plethora of unfair scenarios that makes the game fun. Learning how to use those interactions, beat those interactions, or set up those interactions to gain an advantage or come back from a disadvantage is, at its heart, what makes Chess fun.
Trickle down from that design perspective thus makes the most sense because your best players are at any one moment the players that best recognize how to beat/create/recognize those unfair scenarios - this is what "getting better" at a competitive game entails. Getting better aim or positioning or gamesense is getting better at making unfair situations that benefit yourself.
The hard part of trickle down is that in a very high skill game you have to recognize that it is only applicable to the present though. Your current best players are still improving over time because in a skilled game you can't truly master it into perfection. I think a really good example of this in 6s would be how back in the very early years of 6s (2007-2009) Scout was formidable but not considered remotely as good or as feared as he is now. The top players back then - while great and all - couldn't utilize Scout as well as the top players of recent times now can. This is why careful balancing and taking your time on things can also be important because your best players can get better and shift opinions on stuff with little to no changes done to it.
New/low skill players are kind of interesting because you don't have to heavily appeal to them with balancing because they typically do not actually care about balancing until they really dig into the game. The main thing you target these players with via design is the actual design itself - you try to shoot for "simple to play but difficult to master". That's a side topic and you can do it in a variety of ways but one of the more obvious ways TF2 does it is by providing some easier classes that are more newbie friendly, can still do some work in the right scenario, and still have some skill growth that can tie into other classes, like the Heavy and Engineer. If you have a solid pathway to getting into the game & improving, by the time these players care about balancing they'll be at a healthy skill level where the balancing "makes sense".
Even if something is really dominant in low level play if it's bad at high levels you probably don't want to nerf it unless the counterplay required is simply too high (which generally only happens if the teamwork required to employ said thing is basically non-existent). A good example from DotA is stealth characters, they don't dumpster stealth characters even though they feed on super low level play because the counter is simple it's just the low level players have to learn it. You'd better fix this "problem" with UI elements informing players to buy anti-Stealth items when playing against Stealth characters for the first several times than you would actually rendering the characters worse at all other levels of play where people actively counter them.
If something does have too crazy of a counterplay requirement, is braindead, and is dominating low level play though you're pretty free to nerf/rework it because again it's preferable to have something be bad at the high level of play if it saves player retention. This is because your best players - while they care about balance - generally only care if something is overpowered at that level and don't mind bad/useless items as much. You're not usually going to hear Invite players complain that the Candy Cane is a bad Scout item for instance.
If something does arise that breaks high skill games (this would generally be something with extremely low risk, almost no skill required, and high reward) but isn't a big threat in low skill games you can also make adjustments with smart changes that is just overly beneficial to the game. Most notable one here in recent times for TF2 is the Razorback, which was a problem in basically Highlander only. Preventing overheal fixes the problem it has in Highlander without really affecting lower skill pubs - those Snipers weren't getting tanked by Medics nor were they playing around a combo in the first place.
The only reason lower end players tend to not be on board with trickle down is because they care more about feels rather than reals when it comes to the game. They care more about their feelings (I LOVE THIS ITEM DONT NERF IT, I LOVE THIS CLASS Y NERF IT) and don't talk about balance from an actual overarching perspective, these people are usually very transparent about it because they will outright state their feelings in their complaining too, usually not talking about the reality of the item/class/strategy in question at all or very minimally if they do. They care more about the way they actually play the game rather than how the game itself is actually played.