The first thing I'd like to point out is that those statements, regardless of variation, are all wrong. LoL as a game changes wildly from patch to patch and season to season, and while the existing Summoner's Rift solo queue metagame is more or less fixed in terms of roles (duo bot, one jungler, a "support" who does not CS in favor of giving it to an ally), none of that applies to premades of three or more players or other game modes. Even between the roles, there's a wide variation in expectations of what those roles should do. Stop blaming the game; think outside the box and start seeing the game for what it really is.
Inside the real metagame
In practice, there are about eight or nine champions per role that are played or banned in the lion's share of games. There are more in some roles and fewer in others, but overall it comes out to about 40 to 45 common picks. Right away I want to point out that is a big number, that it represents a third of all playable characters, and that these picks are not static and change from week to week even without patches.
In addition to these 40 or so characters, there are at least another 40 that are regularly played, just not as much. For instance, it is pretty rare to see anyone say that Anivia is bad, but she is not commonly seen right now. Karthus is in a similar situation, as is Jayce and Singed. They are not the top picks, but they happen from time to time. Now we're looking at more than two-thirds of the cast seen in high elo solo queue.
Yes, there are some picks that are not seen at all, like Janna, who is currently viewed as worthless. There are some characters who fit in only a non-standard role such as solo top Soraka or support Annie. These picks are relatively few in number. No game is perfect, and for a game with as many usable options as LoL, I think it's pretty admirable that our worst legitimate complaint is that the supports released before Season 2 are kind of bad right now. Riot knows, so even that's going to change. Hopefully the devs embrace solo top Soraka, as I think she fits very naturally in a solo lane.
Now just because you think that you're a rebel does not mean you should go play a ridiculous off-build that has no actual viability. There are established item builds for a reason, and we should embrace them because they work.
For non-standard picks like solo top Soraka, we can extrapolate. She probably wants boots, a Grail, and tank items after that. Spirit Visage naturally synergizes with her ability to heal herself. Sunfire Cape lets her deal AoE magic damage that benefits from her MR shred while making her very tanky. Rylai's Crystal Scepter lets her drop inescapable slows that help her peel for her entire team at once. All at once, we have a build that works with the things a character has rather than ignores them. Her AP ratios are poor, for instance, so building traditional AP on her is a bad idea. Her abilities are fairly close range, so she needs durability. She relies on sustained spellcasting to contribute, so CDR is effective.
Builds like AP Miss Fortune ignore the traits that a character naturally has in favor of very weak ratios that rarely match those of a standard build. If you want to build a mage that uses auto-attacks, play Orianna or Ziggs or Lux and don't try to fit a square peg into a round hole.
One thing that is fairly universal is tank builds. Most characters in the game are decent when built tanky even if other builds are better. This is because LoL's defensive mechanics are fairly universal. AP varies widely in effectiveness between characters, and some characters do not have the stats or synergies to be built as AD. However, while some characters benefit a lot from building durability, there is virtually no character penalized for picking up some armor or bonus HP. Tanky Zed is a lot more useful to his team than AP Tryndamere.
Matchups are the real metagame
Every role in the metagame is influenced by other picks in the metagame. This includes picks for other roles (Jungle Vi makes squishy mids with no escapes risky, for instance) and counterpicks for your own role. This is something that is very useful to know.
If you're planning on picking an unorthodox or underutilized character, you can benefit a lot by knowing the opposing matchups. Not many people know the vs.-Swain matchup, but almost everyone knows the vs.-Riven matchup. If you end up as Swain vs. Riven, she has a really strong lane against him, but only if she knows the matchup better than he does. If you know what she wants to do and have some reasonable counters for it, you will be likely to outplay her.
Some characters just beat other characters, though. It is very hard for a lot of characters to play against Jax. If your character has a terrible matchup against Jax, and you see the enemy team pick Jax, it might be best to change your strategy. On the other hand, if you're playing blind normals and the only matchup you don't know is Jax, it might be worthwhile to request a lane swap. If you can play the 1v2 or mid lane better than you can against Jax, it might be a good plan.
In general, your choice of character in ranked should be tempered by knowledge of what that character can beat, and your choice of character in normals should be based on experimentation to find out what you can make your character do. If you know what your character can do and your opponent does not, you will have an advantage in most cases even if your opponent wins on paper.
We understand what it's like to climb the skill ladder in League of Legends. The Summoner's Guidebook teaches you the tools you need to get a competitive edge. Whether you're climbing the ranked ladder, playing Draft Dominion, or getting crushed by intermediate bots, every enemy has a weakness. And every Thursday, Patrick Mackey shows how you can improve improve on yours.