League of Legends General Discussion Thread

[Krill]

I don't think all ELO Hell arguments I've heard have centred on that issue, but I'd agree that that is most of them. I've heard ELO Hell referred to as losing 10 games in a row because you can't win as 14/x/0 Annie etc though.

I think the main point is that ELO is affected by things other than skill, and it should be accepted as such instead of the raw metric of player skill which is why the misinterpretation causes so many complaints about ELO Hell. The duo queue thing for example is one such point.

[v8mark]

It should perhaps be accepted as a rough guide rather than an absolute cast-iron judgement on a player's skill - I'd agree with that.

[kayapo]

ELO is affected by things other than skill, and it should be accepted as such instead of the raw metric of player skill

One could argue that on an infinite number sample, only skill matter, all other factors aproaching zero interference.

In practice though, i agree.

It`s like voter`s opinion. The larger your sample the less margin for error there is. That is a mathematical fact.

That said, in practice there are many things that can spoil your sample and these outside factors play a big part on analyzing your prediction.

On the Elo hell argument... you deserve your Elo, assuming you've reached a critical mass of games (usually about 100).

If I remember my statistics, with a sample size of 100 games I can say with 90% (or 95, can`t remember) certain that your elo is within ~10% of your "true" elo.

If you do 350+ games its around 5%.

Anyone that played 1k+ ranked games is within 1% or so of their "true" rank.

[Sir Bruce]

If I remember my statistics, with a sample size of 100 games I can say with 90% (or 95, can`t remember) certain that your elo is within ~10% of your "true" elo.

If you do 350+ games its around 5%.

Anyone that played 1k+ ranked games is within 1% or so of their "true" rank.

That is not how Elo works. It is a moving average of a fixed number of previous games, more-or-less, not the average of all games you have ever played.

[Atlas]

There’s something about League of Legends that makes it very difficult for people to judge player skill. I think it’s simply because there are relatively few meaningful inputs from the player; there’s a relatively low mechanical skill cap. Once you’ve reached a certain level mechanically, 75% of being good at LoL is strategic, I think. Poor players are often poor players precisely *because* they don’t understand what is good play in this regard, which makes them less likely to recognise their own poor play.

The answer to this is simple (Speaker pointed it out more than a year ago) so i will bring it back into the discussion


The Dunning–Kruger effect is a cognitive bias in which unskilled individuals suffer from illusory superiority, mistakenly rating their ability much higher than average. This bias is attributed to a metacognitive inability of the unskilled to recognize their mistakes.... http://en.wikipedia.org/wiki/Dunning%E2%...ger_effect

[v8mark]

That is not how Elo works. It is a moving average of a fixed number of previous games, more-or-less, not the average of all games you have ever played.

This is true -- but even if what Platot said were the case, a system like that doesn't account for player improvement over time.

The answer to this is simple (Speaker pointed it out more than a year ago) so i will bring it back into the discussion


The Dunning***8211;Kruger effect is a cognitive bias in which unskilled individuals suffer from illusory superiority, mistakenly rating their ability much higher than average. This bias is attributed to a metacognitive inability of the unskilled to recognize their mistakes.... http://en.wikipedia.org/wiki/Dunning...3Kruger_effect

This is also absolutely right, but my point was that because major mistakes in LoL tend to be strategic rather than mechanical, it's especially difficult for people to realise their mistakes because they don't understand what they're doing wrong. So the Dunning-Kruger effect applies more in the case of LoL than it does in, say, CoD or FIFA, because there mistakes tend to be more mechanical (most people can recognise and understand what the pros are doing... they just can't do it!)

[Sir Bruce]

Argh! The Dunning-Kruger effect is BS! Or rather, the famous study can't distinguish between that hypothesis and simple alternative explanations of the data like measurement error, regression to the mean, or that everyone thinks they're better than they truly are. It could be true, but the best evidence points to 'probably not.' People should stop using it to justify their belief that not only they are better than others but that those others are too dumb to realize it! I guess at this point it's just a meme for people to justify their arrogance based around their own confirmation bias on the internet and this war has been lost, but still.

Ok, social science rant over, back to LoL talk:

How about those big line-up changes? Aphromoo left the sinking ship of TSM.Evo and is laning with MuffinQT for MTW. That bot lane was the best part of v8 by far and given how strong MTW has looked recently I think MTW is 3rd or 4th in NA now.

[sunrise089]

What's the path towards the end of Season 2?

[Tyrmith]

Pain and tears.

[Speaker]

Argh! The Dunning-Kruger effect is BS!

As a Cornell grad, I resemble this comment!