Hah, joined to play top 1 ranked player and noticed "training mode"

[Caffe Latte]

Happily I could leave - thank you for no penalties un this mode. I did not notice earlier the mode, will be more careful. Ah, these elo keepers )

[nmego]

Well the new "competitive mode" should fix that issue to some extent, hopefully.

The thing with being "first" in a game, that even though you might feel obligated to keep your ranking and never let it drop. It is counter-intuitive because by not "having risks" you're not playing "at your best" so you're not "improving as much", and the second you turn off training mode you're going to have a higher possibility to lose than before. Not to mention that others can overtake you etc..

ELO in simple terms, is an indication of how "good" you are. but it only has value if it is always activated. It works best when you're playing with opponents at rating similar to yours (or higher) most of the time.

"Training mode" should only be for some special scenarios where you're like testing a new strategy or something etc..

Now imagine this, instead of turning on training mode at rank 1 and hiding in that shell, you decide to play anyway. not only does this open more room for growth. not only does this teach you how to deal with your rating dropping and not to get angry. but it also makes it so that, even if you lose, you get better and are more consistently able to gain ranking. It also helps you to learn how good you really are. Because some rating inflation happens, and you could sometimes get a rating higher than your skill level.

[Caffe Latte]

Thanks, nmego, for dedicated answer. I believe being ranked 1st comes with also pressure and mindset. On one hand I considered playing with him/her training mode to have a chance to play with 1st. On the other I felt I might not use this time for best since as you wrote this might be easy going.

Just happy there is option to leave without any penalty in training mode.

[souly4]

Now imagine this, instead of turning on training mode at rank 1 and hiding in that shell, you decide to play anyway. [...] but it also makes it so that, even if you lose, you get better and are more consistently able to gain ranking.

While I don't object any other part of your post, I think this is a little digressive. In luck based games, it's almost mandatory that you have an exceptional streak that surpasses your actual skill and it not easily replicable by others. While skill of course still plays a role, namely that you will attain such a (long) streak more frequently or hold a higher average rating long term. Otherwise we could just as well play Bingo or similar...

Look at these headliners in Lost Cities who reached over 600 (still hardly fathomable to me, maybe the old k-factor impacted it too) and who, after switching off ELO regularly, would have never remotely kept it as their true skill assessment.

Ruby1985
https://boardgamearena.com/gamestats?pl ... &game_id=9
The first info box...
Average points: 35.3
ELO points won: -56879

Keepy
https://boardgamearena.com/gamestats?pl ... &game_id=9
-27466

Or jedibr in Color Pop, a game with much less randomness.
https://boardgamearena.com/gamestats?pl ... game_id=27
-77 (only 368 games total so much less variance)

These huge deviances of course arise because you lose more and win less when being vastly over your true rating and dont let it settle down naturally.

Btw. it's not self-evident that you get better with more practice either. For example when it's hard to review whether you lost/won because of or despite certain moves, when that requires slightly higher abilities than within your scope. If you mix these two up, you will actually perform worse. Like changing only one variable in a formula to define its effect, to find out, you would have to replay the whole game with the exact same course, except a single one of your moves, which is not viable.

[ollyfish2002]

Simple : remove elo off option in the "play" part of the site. Create a training zone for those who want to learn or to teach. Reset elo each 1st of april.

[Yorgad]

Hey guys, why don't you just mind your own business ? Training mode is an option for the game maker. If you're not happy with, create your own table or join a tournament. When I play with my friends at home or with unknown people in a board game bar, there is no ELO. And we still have fun.

Cheers,
Y.

[nmego]

While I don't object any other part of your post, I think this is a little digressive. In luck based games, it's almost mandatory that you have an exceptional streak that surpasses your actual skill and it not easily replicable by others. While skill of course still plays a role, namely that you will attain such a (long) streak more frequently or hold a higher average rating long term. Otherwise we could just as well play Bingo or similar...

Look at these headliners in Lost Cities who reached over 600 (still hardly fathomable to me, maybe the old k-factor impacted it too) and who, after switching off ELO regularly, would have never remotely kept it as their true skill assessment.
*snip*
These huge deviances of course arise because you lose more and win less when being vastly over your true rating and dont let it settle down naturally.

Interesting, thanks for sharing this.
I think that those "huge deviances" are also influenced by turning on "training mode". So they are a little higher than the usual. However, they are still stupidly high. The color pop player seems authentic by checking their recent games. I don't think they will necessarily lose a lot of ranking by turning off training mode in that case. That happens when you bring a "more-skill-based game" into the equation though.

Btw. it's not self-evident that you get better with more practice either. For example when it's hard to review whether you lost/won because of or despite certain moves, when that requires slightly higher abilities than within your scope. If you mix these two up, you will actually perform worse. Like changing only one variable in a formula to define its effect, to find out, you would have to replay the whole game with the exact same course, except a single one of your moves, which is not viable.

Well, the amount of "skill increase" that you gain by practice decreases the more skilled you are.
For example, if you're starting a game for the first time. Practicing a lot gives you a high increase in ELO/skill. Because you're getting familiar with the game etc..
But, if you're at like rating 300-400. Then even practicing a lot might not be able to raise your ELO that high. You'll have to practice better

About the "changing only one variable" Instead of replaying the whole game to determine the effects of the changed move. you can just determine the effects of the move and how much advantages/disadvantages does it give you. And the magnitude of those advantages. Then study the move in the relative medium. This requires good knowledge of the game mechanics and concepts.

There are some really good ways to get better at high rating of a game. Once you gain high rating you'll have to starting working smarter as well. Not only harder.

So to summarize:
1. Practicing hard does make you better at low-skill to average-skill. I think it could get you into a good-skill level as well.
2. But beyond that, you'll have to Practice smarter, not only harder.

However, the context here is about really high skill levels of play. so obviously as you said, it can get really difficult then. Though, I think someone who have reached this far would be theoretically able to "make their practice" slightly better to increase their ranking bit by bit.

As for luck-based games. This all applies for the "average" rating of a person. It stands that once the average rating increases, so do the "exceptional streaks".

Hey guys, why don't you just mind your own business ? Training mode is an option for the game maker. If you're not happy with, create your own table or join a tournament. When I play with my friends at home or with unknown people in a board game bar, there is no ELO. And we still have fun.

Cheers,
Y.

"Training mode" is an option to just play without risks. Test new strategies. Just play with fun. etc... It is a good thing to have.
And everyone is free to do as they want. It shouldn't stop the discussion of the topic though?

[souly4]

Interesting, thanks for sharing this.
I think that those "huge deviances" are also influenced by turning on "training mode". So they are a little higher than the usual. However, they are still stupidly high. The color pop player seems authentic by checking their recent games. I don't think they will necessarily lose a lot of ranking by turning off training mode in that case. That happens when you bring a "more-skill-based game" into the equation though.

Of course, that's the same point I meant, turning on training mode permanently when you feel you reached your highest rating in a cycle. If we suppose the starting point is the same (ELO), someone who wins 70% of the time in a game at average might put training mode on after 12 consecutive wins if he seeks to maintain his high score, while one with 90% winning probability could opt for 20 wins or more (according to a brief calculation chance of such a 20-streak would be over 12% after all). That reminds me a bit of Can't Stop with meta-game, who is most adept in artificially maximizing his ELO

Well, the amount of "skill increase" that you gain by practice decreases the more skilled you are.
For example, if you're starting a game for the first time. Practicing a lot gives you a high increase in ELO/skill. Because you're getting familiar with the game etc..
But, if you're at like rating 300-400. Then even practicing a lot might not be able to raise your ELO that high. You'll have to practice better

Sounds reasonable to me, though I would view it a bit more nuanced and include the 'actual skill' or 'potential skill' of someone in that game. If it's 700 ELO, he will under normal circumstances reach 400 a lot more quickly than if it's 400, because if you imagine such a curve, learning effect on the y-axis (approaching zero), time passed on x-axis, or better in this case ELO gain on the y-axis (approaching 700/400), if you assume the same time duration, the gradient must automatically increase for 700.

About the "changing only one variable" Instead of replaying the whole game to determine the effects of the changed move. you can just determine the effects of the move and how much advantages/disadvantages does it give you. And the magnitude of those advantages. Then study the move in the relative medium. This requires good knowledge of the game mechanics and concepts.

The problem is very often you can't cleary determine the effect of that move on the outcome. I can't think of a game where it wouldn't apply, but I take Stone Age again. My biggest struggle would be mid-to-end-game here, when you are required to assess if it's better to take a valuable card or the hut in the middle, or wood mostly. Because the middle huts (tools, food, meeple) lose in importance with passing time, and the cards/points huts get more worthwile conversely (and so the resources spots themselves because you might need some reserve for being more flexible later), all gets somewhat mashed up and one would have to be pretty competent to evaluate the best long term move.
That would by the way support your claim that for example the gap between 400 and 600 ELO players would be smaller than 400 and 200 ELO. The more obvious a move (larger interval to 2nd best, 3rd best...) the more likely even lower players could identify them. In order not to be boring or 'no-brainers', most games though would consist largely of close to call moves. Here it's more assumable one would fail (especially the lower players), but the upside is that they don't wreak so much havoc on your score, well because they are close. So if we start from the 'easiest' move, perhaps 90% would succeed, with heightened difficulty this number would shrink, but increasingly slower. The hardest move of all game would be so subtle that it would suffice to constitute the difference of a 780- and 800-master, just to state some arbitrary numbers in this hypothetical scenario.
A general principle is that you can only note the true quality of moves up to the peak of your skill, not beyond. Similar to the saying 'the advantage of smart people is they can pretend to be dumb, but not reverse'; and this has even an established name in psychology, Dunning-Kruger-Effect. If there was a more optimal move somewhere, you will not genuinely spot it unless you become more proficient so it's within your range. Because 'the same abilities that are required to come to a solution are necessary to recognize that solution as correct', as these scientists once described. Or with other words, if you make a mistake (= not optimal move), you have no clue that it's one, because then you would avoid it naturally.

Actually there is much consensus between our lines, so this is only meant as an addition what I considered sensible or relevant.

[Yorgad]

It shouldn't stop the discussion of the topic though?

I was reacting to sentences such as :
"Training mode" should only be for some special scenarios
"Remove elo off option in the "play" part of the site."

Discussing on the usefulness of ELO, ranking, competition on how getting better totally fine, of course, that wasn't my point

[asd123321]

In a number of games I can't find top players any more. Letting lower players in then gives a huge risk if I lose,
and not much if I win. It would drag my ELO down playing new excellent players who still have a low ELO. This is because the ELO system is so oriented toward the last 20 games or so.