Many of my multiplayer gaming frustrations can probably be attributed to the excessive deployment of high variance tactics. Gamers take outrageous risks where they would normally not be warranted. If the risks pay off, they win big and feel awesome. If the risks do not pay off, they lose quickly, call something OP, then get another round to try to win big. After all, the downside of losing an online game is not that huge, especially if you down-weight the negative.
If you are the sort of person who plays Civilization on settings like “epic” and “marathon,” the idea of “win big or lose fast” is probably anathema. Whatever game you are playing, you are planning to settle in, focus on the fundamentals, operate efficiently and perhaps aggressively, and build to a satisfying climax. And then this twerp decides to throw absolutely everything at his first attempt, either failing miserably and quitting (smack talk on exit optional) or winning and declaring himself the best player ever (smack talk required).
This is where I place the distinction in an RTS between “rush” and “cheese.” Rushing is a perfectly legitimate tactic. It is early aggression that can be very successful, and a good player can rush while still developing a backup plan. Cheese usually involves overcommitting to one tactic that has a simple counter but is difficult to resist without that counter. When Cheese Fails is a series devoted to watching StarCraft II players fail at cheese. You will see people mis-execute their cheese, execute it successfully but have no idea how to follow through, get countered, and every other way your cheese can fail.
Playing Team Fortress II, you see players who stick with Spies and Snipers because they can get one-shot kills. You will see Snipers who get only headshot kills not because they are aimbotting but because they never dip the crosshairs below head level: miss or one-shot with no in-between. You will see other players using exactly one tactic for hours, such as placing an Engineer turret in the same location, charging with the caber, or sticky-trapping a particular corner. When it works, they get free kills; when not, they tend to die instantly. (You can tell who is actually good by their Plan B.)
You see these people in MMOs trying to clear dungeons as quickly as possible. They over-aggro, and if the AE works, they clear enemies in record time. If it does not work, they curse at the tank and/or healer, quit, and try again with a new group. Having a margin for error is for other people.
A key aspect of high variance tactics is total commitment to one path. If you do not hit the counter, you win. If you do, complain about the counter and try again. This also applies to tactics that usually work but still fail against an obvious counter. For example, your D&D mage is awesome but prepared no way to deal with golems except whining to the DM. For example, many games have some sort of stealth that allows you to ambush your targets at will but is completely ineffective around a detector. It is not just that counters to your favorite tactics exist; the particularly grating player is the one who knows almost nothing beyond his favorite tactic, is always all-in, and is morally committed to the notion that it is the counter that is cheap rather than his tactic.
As some of the examples suggest, this is not a strictly PvP problem. Having these people on your side is similarly annoying, especially when they expect you to support them to help their cheesy one-shot succeed, especially when they take credit for any success because the big numbers are by their names. Having one member of your team flame out within the first three minutes is enormously frustrating. It can mean that everyone loses and starts over because you cannot complete your objective without the full team, and this one person unilaterally decided (before even starting) to win now or leave.
When it works, few of us will refuse easy points, gold, whatever, but it definitely feels gimmicky and unsustainable, and the flash in the plan often flashes quickly or quite violently.
I have also placed some emphasis on attitude. With or against a mature player who is aware that s/he is taking a large risk, you may not get a complete and satisfying game experience, but you will at least have a gracious personal interaction. S/he is a fierce competitor who tried a risky gambit this time, likely calculated to be effective given the circumstances. You can respect that where another player might just be annoying for trying a gimmick. If the mature player recognizes a cunning or lucky counter to his/her gambit, s/he will respect that as the risk inherent in the tactic, rather than complaining about balance/teammates and quitting. Context and intent matter for your perception of meaning.
If you were playing American football against a team whose Plan A was an onside kick followed by a Hail Mary pass, win or lose you would wonder what was wrong with those idiots. Many gamers know little else.
In the original Nintendo Tecmo Bowl, the Minnesota Vikings (if I recall correctly) had a hand-off, reverse, and fake-reverse bomb. That fake-reverse bomb took forever but had a very high success rate for 50+ yard gains; failure meant a 2-second sack and try again. As a child, this was the greatest thing ever.