|
Further Comments from Game 1
|
the vote is for DOLINE (B5a) , here is MAVEN's move 8, GRIFT (L3d) .
New tiles : E L L O R T W
Uk Scrabble Maven
1. BLOC F8a 16 16 (Change 5) 0 0
2. CLARTING G7d 66 82 SHEENIER B12a 74 74
3. DINKUM J5d 36 116 DISTAVES B5d 72 146
4. QIS A6d 55 171 FEY K6d 33 179
5. MED C13a 31 202 EURO i10d 9 188
6. ZIT E11a 52 254 SENATOr i14a 71 259
7. HArO O12d 30 284 TAXI J15a 42 301
8. DOLINE B5a 14 298 GRIFT L3d 30 331
unseen tiles : A4, B, D , EE, G, I, J, N, OO, PP, RR, UU, V, W, X, Y, ? (24 tiles)
No vote yet, but possibilities include:
OWLER i2d for 25 leaving LT
LOWER i2d for 22 leaving LT
RELAX L11d for 24 leaving WOLT
GLOW at L3a for 16 leaving LERT
The latter gives the strongest leave, but our chances of a bonus are looking slim now unless we can pick up the blank or get very lucky indeed. However, I don't fancy our chances of winning a dogfight endgame against Maven when we're already 33 points behind. OWLER looks reasonable but could be hammered if Maven has B, J, Y or ?. LOWER has similar problems with B, G, P and ? Both OWLER and LOWER block the H1d TWS spot, but that is only vulnerable to the ? so should we worry about it at all? We may need it ourselves. However, if we don't take the R-IFF hook spot ourselves, Maven could score well with it.RELAX, despite the weakness of the leave, is a nice 'wait and see what Maven does' move - and the W could come in handy for scoring. Don't know yet, but we need to get lucky whatever we play here.
REWAX L11d is the highest score - 30, but I hate those two L's, we've got to play at least one of them. some talk about 'wait and see', with just 12-14 letters in the bag my instinct is that RELAX is too passive, if you'll pardon the pun. However, as endgames are the weakest part of my game , I'm looking forward to learning where I might be going wrong.If we're going for a bonus, another couple of moves to consider are WILL e4d for just 7, but a good leave, opens up row 3 for the T-WILL hook and still leaves the G as a floater (not that G is a terribly strong floater)Also OWL f4a - 14 points leaving -TLER, opens up row 3 to a lesser extent than WILL but scores 7 more. I prefer this to GLOW which uses the same letters for 2 more points. LOWER and OWLER were my first thoughts, they are definitely worth considering, especially for turnover. Is someone going to sim the last 2 moves ? There were interesting choices on both.
There's also REWAX for a useful 30, but the leave of TOLL is dire.
I've just seen OWLET too - which only takes an H front hook and leaves LR. I think that's my favourite so far.
Another dilemma, don't like OWLET or indeed OWLER done from 02,seems just too precarious due to the proximity of the premium squares, with the wrong letters. REWAX scores well, or how about RELAX dumping one L.
Going back to playing OWLET, I think I would prefer the anagram, TOWEL played down from M7 for 18, the E next to the Triple Letter square shouldn't be too harmful as the D tile is the highest tile that can be played horizontally and vertically How about WORTLE down from D1 scoring 26pts, I know the W looks bad and the rack leave of LL isn't inspiring, but is the W that much of a threat where it is, a response of VIEWY for 42pts, looks about the worst scenario. LOWT across from 4E for 17pts sorts out some of the dross and leaves quite an appealing rack leave of REL. Like Philip I have considered fishing with the play of WILL leaving the T hook.Not voting yet, awaiting comments.
A vote for LOWT 4e across, decided the REL leave is too good, and if it comes to it we can always fall back on REL(AX) next go (that's of course, if MAVEN hasn't sneaked in there already).
Here are some Maven simulation results, based on 5000 2-ply iterations:
Best AEOLIAN b9a
- 2.16 DOLINE b5a
- 2.68 OOZED e9d
- 3.40 DOOLIE b5a
- 5.04 DOOLE b5a
- 5.94 ELOINED d5a
- 6.43 OLEO c2d
- 8.21 LOOIE c1d
- 9.91 OLEO n9d
-12.91 OLEINE n7d
So DOLINE and OOZED are pretty good, but AEOLIAN is best. 5000 iterations isn't a huge number, but I don't suppose the results would be significantly different with a larger number of iterations.
I vote for REWAX, L11d, for 30 points. I think the score is too good to forgo.
The leave of TOLL doesn't fill me with horror, unlike most other players, it seems. We can't pick up another T or L, and the O goes well with all the highest scoring letters left. ) These don't include the X, by the way, that shouldn't be on the list of unseen tiles).
I agree. There are easy points on that X, and it might as well be us that takes them - looking at the remaining tiles, BANJAX is developing into a genuine danger if we don't. I don't see TOWL as a significantly better keep than TOLL, and certainly not six points better. Add my vote to REWAX, please.
It IS very tempting. As mentioned earlier, we can't pick up any more Ts or Ls - however there are two more Os to come. :( Maven could also play DEWAX, PANAX, BORAX and maybe others I haven't seen too. Our chances of a bonus on this board are slim. Although I said I didn't fancy a dogfight endgame against Maven, REWAX would take us to within 3 points before Maven plays. Staying close (and getting a stroke of luck - like picking up the J and the ?) might be our best chance of winning. I really don't like TOLL as a leave, but .... My other favoured move is OWLET, leaving LR. Is the leave 5 points better than TOLL? The maths says OWLET is two points better overall. Do I believe this? No. Bird in hand principle again for me please. It's close between the two moves so I'll take the points. Another vote for REWAX. I'd definitely like to see a sim for this one too.
Many thanks for that Pete, although, I must confess, after looking at AEOLIAN again, and considering the fact that it left a floating O rather than a consonant when we were somewhat overloaded with vowels, I really went off the idea. The sim was quite a surprise!
1 vote for LOWT at 4ea please. I really think we need a good rack leave to try for a bonus on either the '3' row, or the N column perhaps, before Maven shuts us out for the endgame.
I would like to add my VOTE to REWAX (L11d). The score is very good given the board position and our tiles (and as stated no more Ts or Ls to come).
To get a bonus *this* late in *this* game we will probably need to pick up the second blank. Given that there are still 4As left unseen from *22* tiles (I'm assuming Maven would have played GRAFT instead of GRIFT on its last move had it had an A at that point - Does Maven assess remaining tiles/vowels in this way, I presume it must?) then I going to 'imagine' that we will pick up an A as well.
Therefore a pickup of A + ? + either one of the two Es or N would give us a bonus around the floating G (assuming that Maven goes somewhere more fruitful [parallel with DOLINE perhaps]).....Is that wishful thinking??!! Not great odds on that coming off anyway - You know when a game is running away from you when you start selecting the letters you want in advance in your head....!! - But as commented earlier, we are going to have get lucky to win from here....
I'd like to join the REWAX bandwagon, for much the same reasons as Paul. TOLL isn't all that bad, and we should take the -AX points first... that's a vote.
the vote is for REWAX (L11d) , here is MAVEN's move 9, WAP (C4a) .
New tiles : A E L L O T ?
Uk Scrabble Maven
1. BLOC F8a 16 16 (Change 5) 0 0
2. CLARTING G7d 66 82 SHEENIER B12a 74 74
3. DINKUM J5d 36 116 DISTAVES B5d 72 146
4. QIS A6d 55 171 FEY K6d 33 179
5. MED C13a 31 202 EURO i10d 9 188
6. ZIT E11a 52 254 SENATOr i14a 71 259
7. HArO O12d 30 284 TAXI J15a 42 301
8. DOLINE B5a 14 298 GRIFT L3d 30 331
9. REWAX L11d 30 328 WAP C4a 29 360
unseen tiles : AA, B, D , E, G, I, J, N, O, RR, UU, V, Y, (16 tiles)
I vote cOLLATE/EH .Who on earth said TOLL was a bad leave? :) We have cOLLATE or TOLLAgE in column N, or TOLLAGEs or TOLLGAtE in row 3. cOLLATE looks best defensively to me.
Yes, I think so. Not the highest score (I reckon that's probably TOLLAGEs), but certainly the move that will give Maven most problems in mounting a comeback.
By my reckoning there are 18 unseen tiles: AABDEGIJNOOPRRUUVY
Thank you , the unseen tiles are indeed as you say, after Maven's 9th move.
Despite the fact that my vision in 20934 came true I think COLLATE is clearly right here. Less points gained on this move, but the benefits gained from blocking the easiest bonus spot, putting an L next to the middle-right TWS and a C to the south of the TWS at O1 will obviously reap rewards in the endgame. Get COLLATE down (that's a VOTE) and bring on the J......*and* at least two vowels.......*and* hope that Maven doesn't get an 8-letter word around the floating G.
Another VOTE for cOLLATE, please, can't see any value in the extra 9 points for TOLLAGEs. I suspect the sim will, but I don't care.
voting so far,
COLLATE (N6d) 7 votes
TOLLAGE (N6d) 0 votes
TOLLAGES (G3a) 0 votes
TOLLGATE (H3a) 0 votes
One more vote for cOLLATE, n6d, for 63 points. (As if it were needed). I can see a danger in JIRGA, i3a, but we'll have to meet that if it happens. And JUDGE, JAGGED, JAGGER etc. We definitely haven't won yet, but at least we're back in with a good chance.
I'm amazed that there is such overwhelming support for cOLLATE. It *may* be the best move, but it only takes us into a 31-point lead, so after Maven's next move the game is likely to be pretty much level - worse after something like JUDGE. The arguments against TOLLAGES or TOLLGATE seem to be based on the possibility of a bonus comeback, since they take off the most dangerous J-spot. How likely is this? I don't know the answer, but I'd like to spend some time thinking about it! I might conclude that cOLLATE is the best move. If it took us 50 points ahead I'd *definitely* conclude it was the best move. can you give me at least tonight to decide??
In my opinion the remaining letters contain too many bonuses that would be playable after TOLLAGES or TOLLGATE. There are 18 beginning with A alone. (Thank you Lexpert.) I couldn't be bothered to count the rest. The J is one tile. Maven has seven tiles out of the remaining 18 - so it's more likely that Maven hasn't got the J than that it has. If we play TOLLAGES or TOLLGATE, there are two seven letter bonus lines plus the possibility of eight letter words through the final E or S. By playing COLLATE, we restrict Maven's danger moves to an eight letter bonus through the G (there are only 30 playable in total) or a high scoring J move. Yes, we may be in for a tight endgame after COLLATE, but unless Maven bonuses or gets a big J move, I fancy our chances of winning it.
Maven played 3 tiles on its last move. I think we can deduce that he did not have the J at that point, as he would have played JAP instead of WAP. That means that there is a 3/21(14%)chance that Maven has the J now whereas after we play our bonus we have a 7/18 (39%) chance of drawing it. Assuming that Maven plays 4 letters on his next move that gives him a 61:39 chance of getting the J at some point. Therefore playing TOLLGATE certainly looks too risky.
I'm not so sure about TOLLAGES being a bad move, I don't think a bonus is a substantial possibility on row 2.But on the whole COLLATE seems a lot safer and with 4 letters left in the bag I fancy our chances of winning the endgame, providing I don't have to play it on my own. Sacrificing those 7 points does seem a little foolhardy but overall it's COLLATE for me as well.
> Maven played 3 tiles on its last move. I think we can deduce that he
> did not have the J at that point, as he would have played JAP instead
> of WAP.
Good point.
> That means that there is a 3/21(14%)chance that Maven has the
> J now whereas after we play our bonus we have a 7/18 (39%) chance of
> drawing it. Assuming that Maven plays 4 letters on his next move that
> gives him a 61:39 chance of getting the J at some point.
I don't think that sounds right. Maven drew three from a bag of fourteen, we're drawing seven from a bag of eleven. The seven unknown tiles on each others' racks don't come into play. I believe that we'd have an exactly 50:50 chance of getting the J (assuming Maven takes the last four, and there's a strong chance it would leave one in the bag), but I can't quite come up with a satisfactory demonstration of that. Just this once, I think a statistical problem is going to elude me! :-)
you're right that the tiles on the racks shouldn't come into play. You're also right that the odds are exactly 50/50. Here's how: Maven drew 3 from a bag of 14. That's a 3/14 chance of picking the J last turn for Maven. We're drawing 7 from a bag of 11. That's a 7/11 chance, **conditional on Maven not having drawn the J earlier**. The actual odds are thus (11/14) * (7/11), where the first fraction is the probability that Maven does not draw the J, and the second is the probability that we do. This works out to 1/2. And just to make sure everything adds up, the odds that Maven picks up the J on its next draw (assuming it plays 4 tiles) are (11/14) * (4/11) * (4/4) = 4/14. As you can see, the 3 probabilities (3/14, 1/2 and 4/14) add up to 1. Which is a good thing. An easier way of reaching the same conclusion is to say, after Maven's last move, there were 14 letters in the bag, and one of them is the J. We're drawing 7, and Maven's drawing 7 (the order is irrelevant). Hence the odds are 50/50.
> In my opinion the remaining letters contain too many bonuses that
> would be playable after TOLLAGES or TOLLGATE. There are 18 beginning
> with A alone. (Thank you Lexpert.) I couldn't be bothered to count
> the rest.
There are 18 racks that form a playable bonus beginning with A out of many thousand possible racks with 18 tiles unseen. (Maybe someone could work out exactly how many different 7-letter combinations there are - my maths isn't up to it.)
> The J is one tile. Maven has seven tiles out of the remaining 18 - so
> it's more likely that Maven hasn't got the J than that it has.
Yes, but only four tiles are needed to get a 40-point play involving the J. I suspect there are hundreds (if not thousands?) of possible racks that include the J and three tiles that would play at i3a.
> Maven played 3 tiles on its last move. I think we can deduce that he
> did not have the J at that point, as he would have played JAP instead
> of WAP. That means that there is a 3/21(14%)chance that Maven has the
> J.
Excellent point. Deducing what your opponent has or hasn't got based on their previous play is a much-neglected area of the game. I'm still not totally convinced, but I'll join those voting for cOLLATE.
I haven't been involved in the game but somewhere in all the statistical analysis it seems you have forgotten Etheridge's First Law of Scrabble: If there is a letter that you do not want your opponent to be holding, there is exactly a 100% chance that he/she is holding it! This dovetails with the Second Law: The lower the probability, the higher the certainty!
I did a sim for our move 9, this as I though it was another position that didn't inspire my style of play. I ran it at 4 ply depth for each player and 27000 iterations and added a few extra words to the sim window for interest sake. The result were:-
WORTLE D1 +3.22
LOWT 4E + 1.82
REWAX L11 - 0.25
OWLET I2 - 0.29
OWLER I2 - 2.02
ROWTH 3J - 2.05
REGLOW 3J -2.33
WORTH 12K - 2.63
TOWEL 4E - 3.58
TOWEL M7 - 4.28
GLOW 3L - 4.85
OWL 4F --5.55
LOWER I2 - 7.90
GOWL 3L - 9.29
LOW 4E - 9.96
I hope you find this useful
As WORTLE won that sim, can I ask one of the devotees of the leave-value calculating system to explain how it could possibly do so? The score was 26, the leave value was *7, giving a total of 19. When, for example, the overall value of OWLET was 26, how could WORTLE possibly come top of the pile? We are told that differences of 2 are significant. OWLET was 7points better. Vowel/consonant adjustments don't affect this as LL is the same as LR in this respect. Are we to believe that the synergy of LL is more than 7 points better than LR?S
urely this can't be so. I would genuinely like to understand, but I don't * and sim results like these do nothing to change my lack of belief in the leave-value system. Or possibly in the value of sims. Or both.
I started playing the game & then it seemed to speed up enormously. Just caught up with it again - some very interesting decisions in previous moves. If it's not too late another vote for cOLLATE n6 (63).
the vote is for C O L L A T E (N6d) , here is MAVEN's move 10, YO (O7d) .
New tiles : A B E G N R U
Uk Scrabble Maven
1. BLOC F8a 16 16 (Change 5) 0 0
2. CLARTING G7d 66 82 SHEENIER B12a 74 74
3. DINKUM J5d 36 116 DISTAVES B5d 72 146
4. QIS A6d 55 171 FEY K6d 33 179
5. MED C13a 31 202 EURO i10d 9 188
6. ZIT E11a 52 254 SENATOr i14a 71 259
7. HArO O12d 30 284 TAXI J15a 42 301
8. DOLINE B5a 14 298 GRIFT L3d 30 331
9. REWAX L11d 30 328 WAP C4a 29 360
10. cOLLATE N6d 63 391 YO O7d 26 386
unseen tiles : A D I J O P R U V (9 tiles)
No J. Bugger?
First instinct is to play BUGGER B3a for 26, and keep AN for flexibility with the last pickup. I'm slightly dubious however, as it makes a slot for PRIVADO. How big is that risk? In theory 1 in 72, but I don't know if WAP and YO give any indication whether it's more or less dangerous than that. Also I'm not sure if any other lines would beat BUGGER. No vote yet by any means.
How about GANGER? It doesn't leave the PRIVADO spot, but the leave of BU isn't very flexible and the TLS overlap could cause us problems. Definitely not a vote, just thinking aloud.......No, probably not. There's the possibility of JURA too.
Difficult isn't it? BUGGER leaves the danger of PRIVADO and also leaves the possibility of JIAO and JUDO from F2a. However, our leave of AN would allow a play on the H1 triple if either of these was played. GUNGE doesn't leave any bonus possibility, but JURA is still a danger. Our leave of BAR allows a play on the HI triple if this is played. If it is not played, we may be able to play GRAB from i3d for 24. However, GUNGE scores 8 points less than BUGGER. How likely are we to win if we are only 23 ahead after this move? My instincts don't like leaving the PRIVADO option, but if Maven hasn't got the bonus, those extra 8 points could be vital. I'm no good with sums (or sims) - somebody tell me what we should play here. I suspect BUGGER is the right play, despite the PRIVADO danger. Another very interesting position.
All the more so as I've changed my mind and now think it's only a 1 in 36 chance, not 1 in 72. I used to be good at statistics as well... If I knew for sure that BUGGER would win against anything other than PRIVADO, I'd probably still vote for it. But I'm not, and it's a lot of grunt work to check all the lines. I'm wondering about GRAB I3d for 24. I think I'd feel pretty confident if Maven's reply were anything other than JO J2, leaving us 1 point ahead with NEU## against the remainder of ADIPRUV. Even then, we'd be well in contention. Anyone see any big dangers that I've not spotted on the first pass?
GANGER was mentioned as food for thought. Is this at i3a? If so, PRIVADO can still be played on top of GAN...., can't it? I'm trying to think of something more constructive to say, I'll be back.
No, that's helpful - GANGER can't possibly be a better option than BUGGER. Removing one of the alternatives makes life a little simpler. I already dislike my earlier suggestion of GRAB as well. On reflection, I'd be amazed if none of the possibilities after JO result in us losing.
> All the more so as I've changed my mind and
> now think it's only a 1 in 36 chance, not 1
> in 72. I used to be good at statistics as well...
1 in 36 is correct.
> I'm wondering about GRAB I3d for 24. I think
> I'd feel pretty confident if Maven's reply were
> anything other than JO J2.
JIAO h1.
Bye bye GRAB. :-)
Yes it can. Didn't realise at the time. I think the main alternative now is GUNGE, but I just don't know enough about statistics for this. In fact, I don't know anything about statistics. :)
2 thoughts: 1) I like GUNGE from the defensive point of view, but I do wonder what we would do if we picked up the J. We could get rid of it, say, by JIFF, but not much else perhaps. Well, there would be RAJ at m2a, perhaps, if Maven hasn't got there first. 2) If the general opinion swung back to BUGGER, taking the risk of PRIVADO, I think it would be worth considering BUGGAN instead, because that takes an E hook. It would also make RAJ at m2a impossible, which or might not be an advantage. As you can tell, I am still in the initial stages of my thinking about this very intriguing move.
OK, GANGER is no good as pointed out as it DOES leave the PRIVADO spot. Then thought of GRAB i3 (24) leaving JURA 2f as a danger but as Abraham points out JIAO h1 is a killer so GRAB is definitely no good.
BUGGAN is a great find and a word (along with BUGGANE) which I certainly didn't know. In any case, we'd be taking the 1/35 odds and it would be Maven who needed to get lucky on number 11, black odd...But are the odds really 35/1 against PRIVADO? We "know" Maven didn't have the J two turns ago, because of WAP, which does affect the odds slightly. Are there similar, more important, inferences we could draw from the last turn and Maven's failure to play JIRGA I3a (42) or even RAJ I4d (30)? However, the same analysis can be applied to other tiles and to my mind, Maven's failure to play VLY M8a (27) stands out as evidence that it didn't have the V. In which case, there's a 50-50 chance the V is in the bag, and only a 55/1 chance that Maven has PRIVADO (trust me, I'm a bookie...) - notwithstanding the "J" analysis above. Instinctively, I think BUGGAN must be the best play, not least because of the score and possible future score using BUGGANE, but I'm still not voting yet!
I just cannot decide between GUNGE and BUGGAN. One minute I feel cautious and plump for GUNGE, the next minute I say to myself "I'm not frit!"*, and go for BUGGAN. Can any one help me make up my mind? * This dialect expression is much used round here, that is,Grantham, whose most famous/infamous daughter used it in an important speech asserting her determination to press on regardless. I had been trying to convince myself that GUNGE would win, and I came to the realisation that JUD/GUNGED could leave us in big trouble - especially if Maven was also holding the O. That appeared to tip it BUGGAN's way, but then I realised that JUN on to the N of BUGGAN would give us a similar problem. I can't do the maths to back up my gut feeling, but I think BUGGAN is right. Let's spin Raymond's roulette wheel and hope that neither PRIVADO nor JUN comes up. That's a vote for BUGGAN, for better or worse. Well done on coming up with it. I didn't know it or the E extension.
Except that GUNGED doesn't exist. Change my vote back to GUNGE. For the moment at least. :)
I feel a bit like this myself. Nevertheless I'll take the plunge & vote for BUGGAN 3i (26). While agreeing with Pete tha ta reply of JUN n1 (20) could be ominous if Maven has the O a pick of one of the D, I, P or V which I make just under 45.5% likely enables us to block.
OK,OK, I've read your comments, Pete and Raymond, with much interest. Of course, Maven might play JIN instead of JUN. Would that make it more or less tricky? Anyway, I'll follow my usual instincts, and take the risk. So, I vote for BUGGAN. I think the E hook tips it for me. At least if I get shot at dawn I won't be alone facing the firing squad.
Belated bravery, another vote for BUGGAN.
the vote is for B U G G A N (i3a) , here is MAVEN's move 11, JIVE (G2d) .
Our Rack : A E R U
Uk Scrabble Maven
1. BLOC F8a 16 16 (Change 5) 0 0
2. CLARTING G7d 66 82 SHEENIER B12a 74 74
3. DINKUM J5d 36 116 DISTAVES B5d 72 146
4. QIS A6d 55 171 FEY K6d 33 179
5. MED C13a 31 202 EURO i10d 9 188
6. ZIT E11a 52 254 SENATOr i14a 71 259
7. HArO O12d 30 284 TAXI J15a 42 301
8. DOLINE B5a 14 298 GRIFT L3d 30 331
9. REWAX L11d 30 328 WAP C4a 29 360
10. cOLLATE N6d 63 391 YO O7d 26 386
11. BUGGAN i3a 26 417 JIVE G2d 15 401
unseen tiles : D O P R (4 tiles)
UREA, O1d, 26 and out, and well done for the BUGGAN find
UREA o1 (26) to win. Congrats to for finding BUGGAN. Also many thanks to for running what turned out to be an excellent game with many interesting choices to be made. Hope you can run another one soon.
the vote is for U R E A (O1d) , UK Scrabble wins. I would like to thank you all for joining in and making this game so enjoyable, and I will be happy to
run another soon.
Uk Scrabble Maven
1. BLOC F8a 16 16 (Change 5) 0 0
2. CLARTING G7d 66 82 SHEENIER B12a 74 74
3. DINKUM J5d 36 116 DISTAVES B5d 72 146
4. QIS A6d 55 171 FEY K6d 33 179
5. MED C13a 31 202 EURO i10d 9 188
6. ZIT E11a 52 254 SENATOr i14a 71 259
7. HArO O12d 30 284 TAXI J15a 42 301
8. DOLINE B5a 14 298 GRIFT L3d 30 331
9. REWAX L11d 30 328 WAP C4a 29 360
10. cOLLATE N6d 63 391 YO O7d 26 386
11. BUGGAN i3a 26 417 JIVE G2d 15 401
12. UREA O1d 26 443
++DOPR 14 457 401
unseen tiles : D O P R (4 tiles)
|
Site Navigation