According to JF rollouts, as well as evaluations, making the 3 pt is
clearly better. Human experts discovered this too, before programs were
strong enough to be trusted. The second best play is 13/8 24/21, btw.
- Fredrik Dahl
>In Paul Magriel's book he recommend that an opening roll of 5-3 is
>played 13-8 and 13-10, because 8-3 6-3 leaves a gap of 2, but
>JellyFish always suggest 8-3 6-3 even on level 7, so is 8-3 6-3 better
Yes, it is.
Expert opinion on how to play an opening 53 has shifted several times
since the 1930s, and at least once since Paul Magriel's "Backgammon"
was published in the 1970s.
8/3 6/3 is recommended in several books from the 1930s. For at least
one author of that time, the second choice was not 13/8 13/10, but
13/5. In Bruce Becker's hyperagressive Backgammon for Blood (1974),
13/5 was his first choice, 13/8 13/10 was second, and 8/3 6/3 a poor
third, for the same reasons Magriel gave in his book. In Tim Holland's
Beginning Backgammon (1973), 8/3 6/3 is called the "conservative 53,"
and 13/8 13/10 the "aggressive 53." Goren's Backgammon Complete
(1974, with "Technical Consultant" Chuck Papazian) also recommends
13/8 13/10 but calls 8/3 6/3 only "slightly inferior." Interestingly,
Goren says that 8/3 6/3 had been the "automatic" play until recently.
Magriel's recommendation of 13/8 13/10 reflects the body of expert
opinion in the mid-1970s. Rejecting the three point as "too deep,"
experts aimed for the 5, 4 and bar points. The idea behind 13/8 13/10
is to diversify builders with only a small risk in order to build a
potentially powerful prime.
Today, however, experts agree that making the 3 point is better even
though it leaves a gap on the 4 and 5 points. There are several
First, a point is a point. Perhaps because modern play rewards both
splitting/slotting *and* hitting those blots aggressively, any point
in the home board, even the 3 or 2 point, can be valuable in an early
exchange of blot hitting. And of course the 3 point becomes a much
more effective blocking point as soon as the 5 or 4 is made also.
Second, while 13/8 13/10 aims for flexibility in making valuable
points, it can quickly lead to an inflexible position, because it
weakens the midpoint by stripping it of all but one builder.
Third, putting a fourth checker on the 8 is not helpful at all. The
builder on the 10 point *potentially* makes 5/3, 5/1, 6/3, 6/2 and 6/4
play well, but not if the blot gets hit first, and not necessarily
much better than how these numbers would play after 8/3 6/3 and the
opponent's next roll.
As Goren put it, in the 1970s, experts played 53 to try to make
"potentially useful points rather than settle for the scrawny
bird-in-hand." Since then, the experts have realized that the bird
isn't so scrawny.
Daniel Murphy San Francisco, California rac...@cityraccoon.com
Opening rolls, as with many other things in life, are subject to trends
and fashion. At the time of Magriel's book, a 53 was often played 13-8,
13-10. "Backgammon for Profit" by Joe Dwek which came out a few years
before the Magriel book also recommends the same play for a 53.
These days most strong players simply make the three point and go from
With regards to your last comment, it is probably always right to try to
play like Jellyfish!
Tony (Topaz on FIBS)
Oh, by the way, this prime-the-opponent-at-all-costs style of play is
known in the backgammon world as "pure" play.
It is definitely better to make the 3 point. In fact, you might be
surprised to learn that making the 2 point with a 6-4 is competitive
with 24/14 and 24/18 13/9.
I used to routinely play 13/8 13/10, as recommended by Magriel and
other books from the '70s. Then I got a book by Phillip Martyn, which
stated something along the lines of "Most experts say the 3 point
is not worth having early on. They are wrong. The more one plays
this game the more one realizes the value of *any* inner-board point."
That coment by Martyn got me thinking. I decided to do a systematic
review of the rolls after both pointing (8/3 6/3) and building (13/8
1-1 edge to pointing, since we might hit
2-1 edge to pointing, since we might hit
4-1 edge to pointing, since we might hit
5-1 edge to pointing, since we might hit
3-2 edge to pointing, since we might hit
5-2 edge to pointing, since we might hit
6-2 edge to pointing, since we might hit
4-3 edge to pointing, since we might hit
5-3 equal (played 8/3, 6/3, of course!)
6-3 big edge to 8/3 6/3
5-4 big edge to 8/3 6/3
1-1 edge to pointing (extra point, no shots)
2-1 edge to pointing (3 point versus 10 point)
3-1 edge to pointing (3 point made versus unmade)
4-1 edge to pointing (3 point versus 9 point)
5-1 edge to building (5 point versus 3 point)
6-1 edge to pointing
2-2 edge to pointing (3 point made versus unmade)
3-2 edge to pointing
4-2 edge to pointing
5-2 edge to pointing
6-2 edge to building
3-3 edge to pointing
4-3 edge to pointing
5-3 edge to building
6-3 edge to building (7 point versus 3 point)
4-4 edge to pointing
5-4 edge to pointing
6-4 edge to building (4 and 3 points versus 4 and 2 points)
5-5 edge to building
6-5 edge to pointing
6-6 edge to pointing
When you total it all up, making the 3 point puts you a tempo
ahead of the building play. Sure, building comes out better if you
roll 5-1, 5-3, 5-5, 6-2, 6-3 or 6-4. But the chance of that is just
30%, and even that is diluted by the fact that the opponent could roll
5-4, 6-3 or a small doublet and take away the benefit.
Now, in defense of the building play I must note that the gain from
covering the 5-point or 4-point is very large, whereas the gain from
owning the 3-point is small. In fact, the 3-point has little priming
value (because of the gaps), so its value is limited to the extra
3/36 of the time that the opponent will stay on the bar when you hit
Adding up the rolls of both sides, it seems to me that you will be
happy you made the 3-point over 75% of the time. And my judgment
is that the infrequent advantage of the 4- or 5-points over the 3-
point is insufficient compensation.
JF gives 8/3 6/3 a .046 evaluation, and 13/8 13/10 a .014 evaluation.
If accurate, these numbers give 8/3 6/3 about a 3% edge.
> [much interesting stuff about how 5-3 is best played by making the 3 point]
Do these arguments apply to an opening roll of 6-4? How do you play it,
and is it dependent on the match scoreline?
-------------------==== Posted via Deja News ====-----------------------
http://www.dejanews.com/ Search, Read, Post to Usenet
> Do these arguments apply to an opening roll of 6-4? How do you play it,
> and is it dependent on the match scoreline?
There is greater choice in the opening 6-4. There are three reasonable
plays. I have seen strong computer programs choose each of the three.
You can run with 24/14. Tries to escape a back checker and build the
outer table. Most of the time you get away with it and have a small
plus, but when you are hit you are suddenly a substantial underdog.
These conditions indicate that running is a good play if you need to
have a high overall chance of winning, and do not care about winning
or losing a gammon. Say at double match point.
You can split with 24/18 13/9. This tries to build, to make an advanced
anchor, and to escape a man. The man on the 18-point looks exposed, but
if it is hit there with a blot, then there are 16 return shots. In fact,
the opponent can actually expect to lose ground in the race after he
hits with a blot! This move creates a lot of tricky counterplay. It is
an excellent move if you are the stronger player. (I like to think
I am a stronger player, so this is my move!)
You can point with 8/2 6/2. This solid move leaves no shots, and it
has tactical benefits. For instance: the opponent cannot split his
back men with an ace, so he has to play aces with 19/20, leaving a
shot when you have made an extra inner-board point. The problem
with pointing is that you lose a lot of equity in the priming game,
since it is hard to build a prime that includes the deuce point.
On the other hand, you strengthen your blitzing game. It follows
that this move is excellent when winning a gammon is important, say
if you are down by a lot in the match.