Monday, 6 February 2012

Backgammon and fragments of knowledge...

Teaching is not generally about telling people how to do stuff. It's more a process of helping them to develop their own understanding by breaking complex things into small bits that they can get their heads around.

As a teacher, I will not always just tell people how to do things. I'll ask them a series of leading questions, especially if the thing I want them to understand is counter-intuitive.

Here's an example.

Me: When should you change to a lower gear? When you want to go faster or when you want to go slower?

The question is phrased in a way that requires my pupil to give one of two answers. It's not an open question encouraging them to discuss an idea at length, although there are times when those type of questions are useful. I give the option of "when you want to go slower" last usually.

Pupil: When you want to go slower.

This is the answer I get most of the time. I've put a "." at the end of it. In practice, it should really be a "?" more often than not. It's the answer I generally want to get, because it's wrong in most circumstances. So it's then a starting point for more questions.

Me: What is the most powerful gear in this car?

Generally the response to this question is...

Pupil: "Gear 5".

In fact gear 1 is far more poweful. There's a trade off between speed and power. If you don't believe me, try riding a mountain bike up a steep hill in 21st gear, or trying to pedal down that same hill in first gear.

Me: When do you need power? When you want to go slower? Or when you want to go faster?

Pupil: When you want to go faster.

Me: So when do you change to a lower gear...?

Mostly they still don't really trust this answer at this point, but it makes sense to some degree. I'm not expecting them to really grasp it straight away. I'm planting seeds. I'm messing with their heads. And what I say is true. When you get to a junction or traffic light or whatever, you slow down or stop, and pretty much the last thing you do is you shove the car down into first or second gear. Then you accellerate.

Learners don't. They want to change gear first instead of last. The snippet of conversation I've highlighed above is a set piece of my job. The details vary from pupil to pupil, but the the whole follows more or less the same narrative structure in almost every case.

Anyway, since  this breaking down into simple unit elements is something I do as a teacher, it's also something I recognise as a learner.

And that brings me to backgammon.

I'm a reasonably competent backgammon player. Not world class, but good enough to beat a world class player perhaps one time in ten. It is a game of luck as well as skill.

I've started playing in my local Liverpool backgammon club one night a month, and I wish there was more around here. Mainly I play online at dailygammon.

Dailygammon offers unrated "friendly" matches. It also offers rated tournaments. In these, you don't have any say about who you get drawn against. At the end of a rated match, your rating changes. If you win your rating increases. If you lose it goes down. The amount by which it changes depends upon the length of the match you played (A match played to 9 points will affect your rating much more than a 1 point match) and the rating of your opponent realtive to your own rating (Beating a player rated 300 points better than you will improve your rating by much more than beating a player rated 300 points worse than you)

The baseline for the ratings is 1500. In practice the average rating of all active players at the site will be slightly higher for a variety of reasons. The very best players are rated around 2200. The very worst have a rating of perhaps 800.

This rating is visible on your profile page. Here's mine.

 DG   Top Page |Game Lounge |Discussion |Links |Log Out |Help  

personal high water mark? 1889.96 (12 March 2011)


I tend to oscillate between about 1820 and 1860. My current 1833 is pretty much representative of my abilities. But I do want to improve, and I've got to a point where I'm not going to get much better by the osmosis of just playing. I suppose I should read books on the subject, but I struggle with doing things that way, and I lack the self discipline to perservere. I have a golf book from my Dad languishing. I feel guilty for not reading it but not guilty enough to get my head stuck in it.

Looking at the rating of my opponent, I'd expect to see certain things.

If they're rated at 1200 or less, they don't have much of a clue about how to win the game. They may know how to play, but they don't really get the strategy beyond a very basic level.

Between 1200 and 1400 I'd expect them to make fewer obviously rash moves, but not give much thought to the game beyond the immediate roll they're dealing with.

Between 1400 and 1600, they're starting to play in a way that strives to obtain objectives, such as building walls, covering vulnerable points, etc. They may not always do so in an efficient way, but they have some eye for the bigger picture. They might still make some basic errors, such as not doubling post crawford.

Here for example, I'm playing an opponent who's rated in the high 1500's.

  _ _ b5 _ b1 b2   b2 b3 b1 b1 y1 _     lainie202
89 pips
score: 6
        6 3   1   9 Point Match
  y2 _ _ y2 y1 y5   _ y3 y1 _ _ _     helvellyn
92 pips
score: 8

I've just moved my 6-3 roll to the positions indicated by the pink arrows. It's now lainie's turn. It's a 9 point match, and the match score is 8-6 to me. I only need one point to win the match. Lainie needs three points. She should turn the doubling cube to 2. She stands to win two points if she wins this game, taking the match score to 8-8. If I win 2 points, I still just win the match. I don't get anything extra. But she probably won't. She'll just roll the dice instead.

Between 1700 and 1900 the player has a good knowledge of the game in just about all of it's aspects.

Above 1900, and certainly when you get into the 2000's, the player has a head full of probability tables and roll-outs. S/he can quickly weigh up all the options available and choose from the best of those options according to the situation. They take into account what their opponent has done in similar situations earlier in the game, or will even look back over their opponent's previous matches to guage their probable response to a gambit.

And some of all that complicated stuff that those top players do can also be broken down into small, easily digestible chunks. For example, if the match is tied, and the score is 2 away (for example the match score is 7-7 in a 9 point match) it's statistically advantageous to double at the first opportunity if you don't get the first roll. If you don't do so, it then becomes advantageous for your opponent to turn the doubling cube themselves on the next roll.

Today I've picked up another fragment.

It came from playing a real match with a real person in Liverpool. He expressed surprise at a move I made, and obviously thought I should have done things differently. So I asked about it on the discussion board at dailygammon.

And one of those top rated players told me that "With two checkers left have them as close to 2.8 pips apart as you can. If you plan on playing a lot it is worth just knowing how many good rolls each two checker position gives."

Another fragment for my collection of fragments. I used to have 1833 of them. Perhaps this added snippet will make it up to 1900 someday.

driving lessons in Wallasey?

1 comment:

Pete said...

You are probably right, up to a point, to simply play for enjoyment and thus keep you rate of improvement at a stoned snail's pace.

Here, I played a guy from Bebington who studies backgammon move probability tables. He beat me but was a dick.

There is my detailed psychological analysis