I was looking at the old emailing list about this longtime post of Phil Appleby. This was his Round 6 game vs. Michael Gongolo of Kenya at the World Scrabble Championship, Melbourne on Nov-1999, wherein he played just one tile E at 3i from his rack of AEELRST hoping to draw either E or I then he'd drawn the E luckily, targeting the T(E)ASELER/ IT/ GAE spot. The word source at that time was the SOWPODS (TLW98+OSW3).
Here was the post of Phil Appleby:
"I've since been analysing the position, to work out what Michael should ideally have played. Is there any way he can guarantee a win, even if I play a bonus? And if not, what gives him the best chance of winning? Here's the situation after (R)E, from his perspective:"
Phil's Score: 312
Michael's Rack: ABINNOT
Michael's Score: 359
Unseen: ACDEEGLRST (3 in the bag)
"Firstly, let's consider the potential for bonus plays, given the remaining tiles:
Play Score In bag
DECREASE at 14g 90 GLT
DEGREASE at 14g 88 CLT
TEASELED at 2g 78 CGR
TEASELER at 2g 76 CDG
EAGEREST at 2h 80 CDL
DECREASED at 5a 76 GLT
DEGREASED at 5a 74 CLT
DELEGATED at 5a 74 CRS
SCARLETED at 5a 74 DEG
RELEGATED at 5a 72 CDS"
"Firstly, let's consider the potential for bonus plays, given the remaining tiles:
Play Score In bag
DECREASE at 14g 90 GLT
DEGREASE at 14g 88 CLT
TEASELED at 2g 78 CGR
TEASELER at 2g 76 CDG
EAGEREST at 2h 80 CDL
DECREASED at 5a 76 GLT
DEGREASED at 5a 74 CLT
DELEGATED at 5a 74 CRS
SCARLETED at 5a 74 DEG
RELEGATED at 5a 72 CDS"
"So OBTAIN(ED) did block the most threatening bonus spot, but would lose against 5 possible racks (and it's reasonable to assume that I'm close to something good, after my single-tile play-off.)"
"Perhaps the 2g/2h bonus spot is the one to block, since there's an implication that I'm fishing for a bonus there? BOTANIC at n2 would be a very elegant play, but would obviously lose to any bonus."
"Looking at it another way, is there any way of playing off two tiles, leaving one in the bag, and scoring sufficient points to win even if I do play a bonus? Some sort of set-up play, perhaps? How about AB at n13 for 10? Now, if I play DECREASED at 5a for 76, and Michael picks GL, he can play TIGLON/ABO for 32, taking a lead of 401-388; now, my best play of JOT loses by 1. And if I play DELEGATED for 74 and Michael picks CS, SCION/ABO wins by 3."
"Is there any realistic chance of Michael setting up a winning bonus of his own? Four of my bonues involve leaving G in the bag; how about AN at g7 for 13, leaving BIONT, or AB for 17 leaving INNOT? I'm not sure this really helps. After DECREASE at 14g there's the possibility of BOLTING or BOTTING, but DECREASED at 5a kills them stone dead."
"Is there a way of playing off one tile, and leaving the potential for a winning play? Plus the possibility that I won't be able to play out with, say, CG on my rack?"
"At this point I gave up, but it certainly seems like an interesting position; perhaps even more so if Michael's lead had been slightly greater. My initial thought is that BOTANIC is probably the best play, given my previous move. Although I don't suppose computer analysis would see it that way."
"I'd certainly be interested to see further analysis..."
-Phil Appleby
Somebody replied from this post and answered back by Phil:
"Sam suggested that TOEA at f1 might be a good play at this point, blocking 70% of the playable bingos. Possibly, but bearing in mind that I'm likely to have all of the remaining vowels if I *do* have a bingo, it could potentially lose to a non-bingo endgame."
"For example, consider the actual situation, with me holding AEELRST. With CDG in the bag, if Michael picks DG, the best sequence of plays is:"
"Phil LETS 4a 39
Michael BING n12 20
Phil CEDAR 8a 27"
"This gives me a 1-point win."
"I suspect there are other scenarios where I can win the game without a bingo, but I'll leave that to the computer people!"
"So is TOEA the best play? Or BARONIC? Or something else??"
"For example, consider the actual situation, with me holding AEELRST. With CDG in the bag, if Michael picks DG, the best sequence of plays is:"
"Phil LETS 4a 39
Michael BING n12 20
Phil CEDAR 8a 27"
"This gives me a 1-point win."
"I suspect there are other scenarios where I can win the game without a bingo, but I'll leave that to the computer people!"
"So is TOEA the best play? Or BARONIC? Or something else??"
-Phil Appleby
Blogger's Analysis:
Running the Quackle Engine Simulation after 80,000 Iterations, here are the results:
The top choice would be ANE 2f (96.97% chance of winning) blocking one potential bingo lane along 2nd row, like T(E)ASELER, (E)ARLIEST in case of an E or I pickup based on unseen tile after the RE move, as planned by Phil, although this would leave the -ED spot opened for bingo in case Phil drew the letter C. Let us see:
>Michael: ABINNOT - AN(E) 2f +12 -> 371
>Phil: AELRST+C - SCARLET(ED) 5a +74 -> 386
Variation 1: If Michael drew DG
>Michael: BINOT+DG - OB(S)IGN d3 +36 -> 407 (BO(D)ING 8a also wins by fewer spread)
>Phil: E - (JO)E 10b +10 -> 396
>Phil: (DT) +3 -> 399
>Michael: (DT) -3 -> 404 wins
Variation 2: If Michael drew DE
>Michael: BINOT+DE - BI(D)DEN 8a +36 -> 407
>Phil: G - (JO)G 10b +11 -> 397
>Phil: (OT) +2 -> 399
>Michael: (OT) -2 -> 405 still wins
Variation 3: If Michael drew EG
>Michael: BITON+EG - BO(D)GIE 8a +36 -> 407 (OB(S)IGN d3 also wins)
>Phil: D - (A)D(S) 6i +8 -> 394
>Phil: (NT) +2 -> 396
>Michael (NT) -2 -> 405 still wins
In case of an E pickup by Phil as per the actual game, let us see further:
>Michael: ABINNOT - AN(E) 2f +12 -> 371
>Phil: AELRST+E - LETS 5a +39 -> 351
Variation 1: If Michael drew DG
Michael: BINOT+DG - BO(D)ING 8a +33 -> 404
Phil: CARE - CARE l1 +22 -> 373
Phil: (DT) +3 -> 376
Michael: (DT) -3 -> 401 wins
Variation 2: If Michael drew CD
Michael: BINOT+CD - B(A)NC 12l +37 -> 408
Phil: AEGR - RA(D)GE 8a +27 -> 378
Phil: (DIOT) +10 -> 388
Michael: (DIOT) -10 -> 398
Variation 3: If Michael drew CG
>Michael: BITON+CG - CO(D)ING +33 -> 404
>Phil: ADER - DEAR l12 +20 -> 371
>Phil: (BT) +8 -> 379
>Michael: (BT) -8 -> 396
The only way that Michael would lose based on unseen ACDEEGLRST if Phil had the following combination:
1) ACDEERS - D(E)CREASE 14g +90
2) ADEEGRS - D(E)GREASE 14g +88 even if C was the last tile, there are several out play like LIMBIC, COOP, CHUM.
By taking the mathematical combination C(n,r) of the possible racks that can be formed from ACDEEGLRST in 7 out of 10 tiles, this would give us 120 combinations. So 2 racks out of 120 rack combination or almost 2% chance that Michael would lose in the ANE play.
No comments:
Post a Comment