Stu,
The question from my point of view is why other systems have had the E so
very high. On a typical board, the difference if you sim exchanging seven
tiles or exchanging six to keep an E will almost always be very slight. I
don't know all of the details of how Watkins, Sheppard, Gordon, and others
created their leave values, but I can see how if a program has only a very
basic concept of synergy (or none at all), "overvaluing" the E might
actually cause it to play better. An E makes a world of difference when
added to already promising leaves, but much less for average (or worse)
leaves. AERST is 31.19 in Quackle's leaves vs 18.37 for ARST, while ABCDE
3.22 and ABCD -2.16 (most of the gain for ABCDE is vowel/consonant balance,
not the E). If in implementing a very simple system for leave values we
boosted the E everywhere, overvaluing things like ABCDE would hurt us very
little, because there are lower probabilities of such leaves even appearing
on a rack, and the tiles score well enough that it's likely that even if
they're hugely overrated, they'll still be broken up, and the program
thinking that a play keeping ABCDE is the 100th best play when it should
really be 300th has zero consequences.
John
On 5/10/07, [EMAIL PROTECTED] <[EMAIL PROTECTED]>
wrote:
John and all--
Thanx, John, for posting these Quackle leave values. But I'm
tremendously surprised by the low value attributed to the E, much lower than
the values I've seen gibven by other systems. Can this be explained?
Stu Goldman
----- Original Message -----
*From:* John O'Laughlin <[EMAIL PROTECTED]>
*To:* [email protected]
*Sent:* Thursday, May 10, 2007 12:00 PM
*Subject:* Re: [quackle] Equity and Leave Files
Until recently, Quackle's leave values were loaded from a human
readable file in the data directory. We changed the format so that it
would load faster and take a bit less space. Below are Quackle's
values for one tile leaves. Bear in mind that these are the values of
the actual leaves, not the tiles. To get values of the "tiles", by
which I mean values would make better approximations of longer leaves
when you add them together, subtract about 0.5 points from each of the
consonants, and add about 0.7 to each of the vowels.
I can make the longer leave values available on the web soon.
? 25.57
S 8.04
Z 5.12
X 3.31
R 1.10
H 1.09
C 0.85
M 0.58
D 0.45
E 0.35
N 0.22
T -0.10
L -0.17
P -0.46
K -0.54
Y -0.63
A -0.63
J -1.47
B -2.00
I -2.07
F -2.21
O -2.50
G -2.85
W -3.82
U -5.10
V -5.55
Q -6.79
