I think this is only really true if a beginner understands why they lost. Otherwise they just repeat the same mistakes, develop bad habits, and arrive at all manner of wrong conclusions about how to play the game properly. I would venture to say that most of the time, beginners quit the game all together long before they've racked up their first 50 "instructive" losses.

The two things are not mutually exclusive.
A baby learns to walk; she falls many times. Even though she falls, she is still improving.
She doesn't know why she is falling, but she is still improving.
She is improving from her experience. She can develop bad walking habits, but she is still improving.

A real beginner learns and improves from his experience. Yes, some (or maybe even most of us) will repeat the same mistakes,
develop bad habits, etc. (unless a good teacher points them out),
but a real beginner is still improving and learning, at the same time.

Aside from what Ed said, go is a long game. Besides winning, the players have subgoals. There are many opportunities for learning, and a player can lose the game yet succeed at a number of goals. True, success is generally better for learning than failure, but go offers many successes.

I agree that beginners should try not to form bad habits, and recommend that they play much stronger players. I lost almost every game I played until becoming an SDK.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Physical skills, and those involving a lot of eye-hand coordination, benefit from repeated drills with minimal coaching or expert intervention. And there isn't usually a whole lot of theory to embed in the learner either. Perhaps a better analogy is needed here?

I just think there is way too much faith around here in the process of learning by osmosis in something as strategically complex and tactically subtle as go. About the only thing a beginner is likely to learn just from losing a lot is how to avoid losing stones in atari needlessly. But I'm not convinced that beginners would even pick up in the importance of establishing two eyes without being told about it explicitly. There are many fundamentals that don't reveal themselves merely through the process of playing.

This boils down to whether go should be handled as a game or a sport. In the original poster's comment, it was described as a sport: pure competition with only perfunctory communication. I definitely think that game-ish aspects can make go more fun without diminishing any of the 'hardcore' player's ability to study, dig, and compete. For example, options that have been discussed on the forum before:

1) Achievements, brought up by Araban originally
2) Separate ratings to measure different things. For example, the level system described here would be another viewable rating, and you would be able to choose which you could see.
3) Unlockables such as custom boards/stones.
4) Guilds/etc.

As far as some of the reservations people have, this stuff would all be optional and none of it would give a player a better chance of winning, so a player who was not interested could ignore all of it. As far as having multiple accounts, there could always be an option to link accounts so that the actual ELO/serious rating was separate but nothing else was. That way your automatching would not be affected but you would still have access to unlockables and such.

The downside to all of this is that it would be work on behalf of go server makers. In the meantime however, it is fun to talk about stuff like cool achievements. All of these sorts of things are kind of like getting to choose your username: some people really don't care if they have the options and it doesn't make you better at go, but it could make the game experience more fun for a decent number of people.

_________________
Someday I want to be strong enough to earn KGS[-].

Not sure why, but if Go ratings were more like this rather than this I'd find it cooler.

This one, at least, has been implemented.

http://senseis.xmp.net/?KGSClans

I'm not sure if they're still doing it. Haven't heard anything about it in ages, though, so I'd assume it's died out...

I'm actually just graduating from my bachelor's in Cognitive Science this year, and as to the matter of how learning works, here's some insight I remember, but am too lazy to link the references to, so you probably shouldn't take it at face value,

Basically, some things, such as breathing, walking, and yes, even language, are known to be innate. Our brains are equip with innate mechanisms to learn and do these things spontaneously at certain ages (called "critical periods"). Other things, such as learning to ride a bike, are not meant to be learned intuitively from birth, however, simple trial and error is sufficient to accomplish basic cycling skills with little to no special training. Other tasks, such as learning to perform surgery... probably needs some teaching to learn, and there is little or no intuitive counterpart our brains are equip to learn from. Feedback is key in trial and error - you have to know when you are failing. On a bike, this is easy, you fall off, you fail. Surgery does have feedback, but obviously going straight to trial and error on patients isn't optimal, so much textbook learning is desired.

The contentious part in learning Go is... how is your feedback? Do you know if a stone laid was good or not? in the case of losing an atari'ed group, probably, but for more advanced moves, maybe not. This is probably why teachers are so highly valued! They can give proper feedback on your play, as do tsumego.