| Player 1 |
|
#player1 EL Ed Liebfried
#player2 JJB John J. Bulten
>EL: ADEILOT 8C IDOLATE +70 70
>EL: ADEILOT -- -70 0
#note 0:31 [24:29] Two high seeds in this round robin (plus king-of-hill) now square off, where EL could clinch first place with a win, but JJB is still only likely to get second with a win. (After the final round, these are indeed the standings.) Quackle uncharacteristically selects eidola/iodate 16 for EL's non-go rack; it rates the best leave as ADELT +16.0, but positionally adjusts the opening rating of oi 8h 4 to only 17.0 evaluation points, while eidola/iodate get 16+2.9 after adjustment. So does EL either score tiles or keep a partial bingo leave? Neither: he goes for the psychological value of idolate* 8c with hopes of getting the bingo points this turn or the next (12.7 behind).
>JJB: AABGKTT -ABGKT +0 0
#note 0:28 [24:31] JJB recalls the mnemonic: toad-lie, letter bank etiolated, (consonants) sofa never pink; so he challenges, but he cannot play the K (toadlike). (Apparently the clock was restarted late because the play took longer than 0:28.) Of course kab evaluates as 9.0 better than exchanging, and batt as 6.1 better, but given the first-turn advantage himself he is prepared to yield it back by hoping for his own improvements on EL's rack, well aware of the dilemma EL will face.
>EL: ADEILOT -O +0 0
#note ~0:24 [~24:05] EL now has the same choice against a presumably improved rack, but without the benefit of potential phony value. He decides reasonably to trade O rather than to give JJB six juicy lanes; this rates at +14.0 (4.9 behind eidola/iodate, 2.0 behind trading IO).
>JJB: AFNOORT 8H OF +10 10
#note 1:36 [22:56] Now JJB does not know what to block so goes for the 2-letter defense, 3.6 behind the best-rated afoot 8h 18. Leave is +7.6, but it's better to keep the points in the leave until EL shows his hand.
>EL: ADEILT J5 ADIT +15 15
#note ~1:50 [22:15] Based on what EL revealed about his rack, he probably drew A, E, or T (if he had drawn I, R, or V, as he did at some point in this sequence, he would have a bingo adding to detail). Apparently EL does not like the wide-open S double lane of adit 7g 16 but does not want to be too defensive or have too much turnover, so his play is at least 4.1 behind the lead.
>JJB: ANOPRTU 6H PU. +6 16
#note 4:12 [18:44] With no bingo on this tantalizing rack, JJB goes for more strong bingo defense rather than play Quackle's opener, puton k3 23, 11.9 ahead. He is still trusting the O, rather than supplementing it with the P for AOPRT, slightly better than ANORT. But un k4 8 and up 7g/9g 9 are also too open for his taste.
>EL: EELV K3 VEE +16 31
#note 0:16 [21:59] Vee 9g 19 is quite a blocker and with some possible racks is the best play (he probably did not draw elative, elevate, leavier/vealier, leveret, overlate (elevator), or relieve, which would mean his play has 3.0 to 15.7 diff). Certainly there is pressure to dump Es overwhelming his need for openness (the oe hook is reasonably sufficient), and this play is after all worth as much as either player scored in 3 turns.
>JJB: AINNORT H6 .R.NATION +62 78
#note 0:39 [18:05] JJB's defensiveness and 9 study pay off with the first-played bingo in the tight board. An exciting find, and a wild opener; also something JJB does a lot during the game (turning of the palms upward).
>EL: AEEILRT 11B LITERA.E +66 97
#note 4:18 [17:41] EL has drawn the hook pair (a)literate, (e)laterite, and knows it, and doesn't like either one here. He spends time rejecting the wide-open elaterin/entailer 70, but has nothing safer. Including relative 3e 63, the top 7 plays all open up easily hooked triple, double-double, or nonuple lanes.
>JJB: AIJOOSU 15H SOJU +69 147
#note 0:28 [17:37] Look for a better one! Always. JJB improves jus to ajis to soja to soju then hurriedly plays the S hook, scoring more than his setup bingo, but without spotting sajou 75 (9.9 ahead). Of course, EL's response would be much more deadly with sajou.
>EL: ELNOSUZ L9 ZONULES +97 194
#note 1:07 [16:34] The power of random draw. But sajous/zonules would have been 115 if playable.
>JJB: AFIOSWY D8 WIF.Y +36 183
#note 2:07 [15:30] Good, but flawy b10 44 is 8.2 ahead. Intriguing lesser options include oilway b9 28 and ofay m7 29.
>EL: ACW A10 CAW +21 215
#note 0:56 [15:40] Cashing his own hook to block bingos falls 5 behind caw m8 26.
>JJB: ADHNOSU 8A DHO. +33 216
#note 1:41 [13:49]
>EL: CHIY 3K .ICHY +32 247
#note 3:13 [12:27]
>JJB: ABKNOSU O1 BA.OU +33 249
#note 1:44 [12:05] Grabbing the obvious spot costs the opportunity to play the bohemian bohunk n1 50 (16.6 ahead).
>EL: ADM L1 MA.D +26 273
#note 3:07 [9:20]
>JJB: AIKNSTX C8 .X +38 287
#note 0:48 [11:17] Fully 10.8 better than ixia 41 in the same place. JJB overscores by 4.
>EL: EEGRT F8 REG.ET +11 284
#note 2:46 [6:34] This block must be important and/or this rack must be very repetitive (EL trades next). Retem 21 might be much better.
>JJB: ?AIKNST M6 INTAKeS +83 370
#note 0:44 [10:33] Always look for a better! JJB stops after this great overlap, but has just studied four higher bingos that he does not even search for: katsinam (1e 89), and antiskid, inkstand, stinkard (all a1 89).
>EL: Q -5 +0 284
#note 1:19 [5:15] Rack unknown.
>JJB: AEELNOO K11 ONE +15 385
#note 4:01 [6:32] Better: leone b11 15, only by .5.
>EL: MV 6L V.M +14 298
#note 0:16 [4:59] Possibly better: vox c7 17.
>JJB: AAEEILO 7A AAL +23 408
#note 1:24 [5:08] Missing aah n1 20, up by 2.7, but better than trading all unless blank hunting is the goal.
>EL: EIRRT 1G RETRI. +24 322
#note 0:35 [4:24]
>JJB: ?EEGIIO A10 ...InG +33 441
#note 3:43 [1:25] JJB is mindful that EL has 6+ consonants, and does not go for such recommendations as Quackle's tie 8m 9 (rated deceptively 26.6 ahead), an opening which will only work better for EL than for him (in fact, EL has balanced up to the one bingo possible for him there, reblends; so Quackle's top pick makes it 413-384 with 2/3 chance of Q draw and very close game, and many similar non-emptying openings actually lose some simulations). Quackle admits that JJB's lead-padding is the best immediate blank usage, even though it's an automatic Q stick except for the 4.6875% proposition that EL has seen no S since his bingo. But why not save the blank for the Q and/or the bingo without opening for EL? Suddenly high-vowel leaves look flexible with this bag; but Quackle's 3-tile plays will be easily blocked by opponent, and 2-tile plays may not yield openings for the opponent's known possibilities. By this standard the highest sim-value play seems to be oe 2f 6, ahead of the cluster lei b11 8, log b11 9, neg 7m 8, oi g13 9. All the same oi seems to better accord with experience; since cawing empties the bag and uses the blank, its value comes out about 40 less despite its padding.
>EL: BDELNRS B6 B.. +14 336
#note 1:32 [2:52] How long should one play against a self-stuck Q? Seven turns? An optimal set without intervention by either player is ed/id 6, nae/nu 5 (the pair then improved to dyne or end 13), bah 14, el 2 (the pair then improved to lah 8, with unblockable blah 9), oke or fe/ne 9, ern 4 (then improved to the unblockable fere 9), bayous/vims 20, total of 68 in 6 turns. Review of other groupings does not suggest any that can be consistently applied without opponent interest (e.g., en/in 4 might be met with ag/gen 7 to prevent end/od 11). (1) A draft of interaction where both seek all they can is first lah 8 (the weakest point, and blocking gad), gnus 5 (blocking end), dyne 13 (blocking fog/oe and lo/wo), pe/zap/oke 29, fe/ne 9, then (order varying) me/ne 6, blah 9, too 4, fere 9, gi 3, bayous/vims 20+20, net of 41. Does JJB lose anything to block bayous? Lah 8, nog 8 (blocks bayous), zas 12 (blocks pe), fe/ne 9 (claims E), re 4 (claims E), er 3, dyne/end 13, pretrim 11, fere 9, gi 9, blah 9+20, net of 35. (Nog takes -1 in tile score to gain 8 in block; zas returns -9 in block; claiming Es early gains 2 overall; but gi gains 6 in the setup.) (2) If EL claims bayous first, it appears gad 5 does too much damage to B and L plays, even if EL blocks pe with tel 9. Without the existence of exact endgame programming it's hard to be certain, but lah b6 8 appears the correct setup, net of 35. For instance, EL may have the opportunity to keep a tile-group outplay that nets better by catching JJB without using up all his non-Q tiles, but JJB may also be able to prepare for such a threat by grouping his own tiles correctly, and so this contingency is not being analyzed herein. (3) The actual endgame start was close to this thought: simply swap lah 8 ... blah 9 with bah 14 ... li 2, and the play appears only 1 behind best, so it is fitting to credit EL with missing only 1 point here in the (b)lah setup.
>JJB: EG 8M .OP +15 456
#note 0:47 [0:38] Although stopping the clock after 14 seconds to verify score and notes, JJB must split between verifying the Q is stuck and seeking outs. The previous analysis sets the block nog 7m 8 as netting -27. Based on it, if JJB plays top 8m 15, his optimal G plays are blockable with end i11 13, requiring a retreat to neg 14a 6 (or similar), fe/ne 9, oke 9, bayous 20, pi 4, trod 6, pass, el 2+20, net of -36. Call it a deficiency of 9.
>EL: DELNRS 6L ...S +20 356
#note 0:24 [2:28] So the block end i11 13 will net 51 to EL. Bayous 20 allows gnus 5, dyne 13, fe/ne 9, oke 9, re 4, fere 9, pi 4, el 2+20, net of 51, and nothing is lost now by not blocking.
>JJB: EEGIQ 2E GEE +8 464
#note ~0:26 [~0:12] So the G alone (gnus 12j 5) nets -31 to JJB. Instead he throws his chips in, gee 2e 8, suggesting dyne 13, pi 4, oke 9, pass, fere 9, pass, el 2+20, net of -41, deficiency of 10.
>EL: DELNR 3F LEND +14 370
#note ~0:38 [~1:50] EL takes the bait instead of squeezing out additional style points. Dyne would net him 49. His play makes the next actual responses optimal: lend 14, pi 4, ern 4+20, net of 34, deficiency of 15.
>JJB: IQ O8 .I +4 468
#note ~0:07 [0:05] No more style points need be lost now.
>EL: R 9F .R. +4 374
#note ~0:41 [1:09]
>EL: (Q) +20 394
#note Reported 468-394, corrected 464-394. A competing-bingo opening, a variety of flashy scores in midgame, and JJB drawing the Q on the last rack to an impossible board, where EL has the option of an out in six, optimal if everything so far is correct. Readers are encouraged to point out improvements in the calculations of the complex armchair mathematical puzzle included, which has a little bit to do with the actual blitz-mode game-life puzzles chronicled. If no changes are made, the following figures are correct. Points improvable based on tiles played: JJB 28, EL 29. Points improvable based on values: JJB 114.0.
|
