Play many games. Fast and slow.

Do not start writing longish entries in forums and wikis. (It is fun, but not a way to improve.)

Play some more games. (If possible against stronger opponents.)

It may help. Learning Go can often be frustrating and it's a bit of a long term project. Being an active part of a community, like here on L19, can strengthen the ties to the game and make the experience more "whole". It's good to "belong". Also, it's the closest thing to a dojo that we may have.

I do agree though that adult learners tend to over-theorize (e.g. read too much) and talk about stuff rather than actually doing it. Myself included. So yes, playing a lot is definitely the key. The magic key to the improvement shortcut probably doesn't really exist. (Just in general - there are people who improve speedily without much playing.)

I regularly go to Clare Buttery, sometimes to the university centre meet. Why? Do you ever come?

I used to before my time disappeared to family commitments - I sometimes go to Mill Road on Friday evenings (CB1), but it would be nice to pop into the Uni again. What time does it start these days?

A note about MC go programs: they play to have the highest chance of winning, not the highest score. If you play as well as they expect, they'll win by .5 every time. The fact that it won by 1.5 indicates that you made a 1 point error late in the game.

I also doubt that it's *really* 3d, but don't mistake the low score difference for an indication that it's not much better than you. There's a story floating around that you'll eventually hear about the mysterious player who beats people of two different levels by the same (low) amount... and when they realize that, they realize just how good he must have been.

_________________
That which can be destroyed by the truth should be.
--
My (sadly neglected, but not forgotten) project: http://dailyjoseki.com

I second that. Although I disagree about the "win by 0.5" a MCTS evaluator will play to optimize the chance of winning, not the margin. So consistently losing by a small score difference does not mean you are just a little weaker than the program. To judge your strenth relative to the program, keep increasing the handicap till you win say 1/3 of the games.

Also remember that the hardware matters. We can expect a program like MFOG to be about a stone weaker playing on a "standard machine*" than on the sort of workstation the bots playing on KGS are running on. And another stone weaker if running on a machine that is weak compared to "standard". Nor would this be consistent. The lower levels using just an AI evaluator might not need so much time and so not be as weakened by lack of "crunch power" as the MCTS levels.

* For MFOG 12 the "standard" amount of computing power is something like a 2 core 2 GHZ machine.

PS -- if trying to learn by playing against a program best to follow a few simple rules. Do not weaken the program till you can play it even. Under those conditions the differences between how a computer program plays and how a human plays will be great and you can learn some bad habits. Instead set the program to something like 3-5 stones stronger than you (19x19) and take the appropriate handicap or just a bit too little. Under these conditions the program will not be playing that much differently than a human opponent 3-5 stones stronger who is trying to teach you and when you make a mistake you'll get swiftly punished for it. Please note that I am not meaning to suggest not even better to have a human opponent 3-5 stones stronger willing to play lots of games with you to help you learn. But is this likely?

Wobble: Yes, hey there , good to see a familiar name!

Topazg: Ah, now I see why you messaged me! Hadn't been on this forum in a while. I replied to your PM .

Ah, alright, I see now, thanks for the replies! As a bit of an update if anyone was curious as to how I'm doing - I can beat the top MFoG level when it gives me 6 stones now, haven't tried on 5 or below. I gave the 9k MFoG computer 2 handicap stones and comfortably won, and I beat 9-10k players on KGS, so I guess my rating now is probably around 7-9k. Haven't played much at all over the past 3 weeks since revision has kicked in - hopefully I'll be able to improve more over summer! Spent £200 on go books to help me improve too , currently working through a book on endgame theory (Get Strong at Go Volume 7). Wanted to be a single digit kyu by the time I came back to Cambridge after my Easter break, and I think I've done that, so yay ^_^ . Still on course for my 'shodan challenge' goal , probably around 8 ranks to go in a bit over 10 months. Of course these are the harder ones now, but... onwards and upwards!

_________________
Sorin - 361points.com

I am fairly new to the game myself but what helped me improve quite alot was SL's beginner exercises/basic go books on fundamental principles. Right now I have discovered the utter brilliance of the'Graded go problems for beginners' series.
vol 1 (30-25k) vol 2 (25-20k) vol 3 (20-15k) vol 4 (15-10k). Working through these one by one will give you a very good foundation from which to improve quickly.

You can check out more recommended books here http://swamphome.tripod.com/baduk/studygo.html

Nothing has helped me more than studying books and taking my time after every move. I got several stones stronger when I didn't play instant-go like many beginners. Ask yourself ''what does my opponent want to do with his move'' and read out situations. If he plays here, then I play there..etc. I'm now 15k IGS and 11-12k KGS after about 4 weeks of study (nibbler on kgs, nyuubi on IGS). Feel free to ask me anytime and I'd be happy to help you improve