Othello Quiz 2022

69 views
Skip to first unread message

Masanori Itikawa

unread,
May 6, 2022, 4:06:37 PMMay 6
to

Timothy Chow

unread,
May 7, 2022, 7:04:22 PMMay 7
to
On 5/6/2022 4:06 PM, Masanori Itikawa wrote:
> now available
>
> http://itikawa.com/fes2022quiz.html

Ah, very nice! So what is the story...was everything
canceled the last year (or two?) and are things back to
normal as far as the Japan Open is concerned?

As for the problems:

1. I got dinged for a massive blunder in a position
like this one recently. If I've learned anything, I
should get this one right: 7/4 6/4. We're in danger
of losing the priming battle and need to make a bid for
a prime of our own that is as good or better. Black's
sixes are duplicated to hit and make his own prime.

2. The score might make a difference because we're
itching for a gammon. Thus we need to consider the
banana split 5/4*. But Black has no other blots to
pick up, so I'm not sure if 5/4* is called for. If
we do play 5/4* then the question is whether 13/8
(for another builder, but not on a different point)
or 21/16 (for better board coverage) is the 5. I think
I would go with 13/8. But do we play 5/4* at all? We
could play safe and just try to consolidate our race
advantage, but that seems like the wrong idea with such
a strong board advantage and a high gammon value. I
would try 22/16, which triplicates Black's 4's and gives
us good chances of hitting a shot soon.

3. Black's position is fragile and we'd like him to play
his roll rather than dance. But making the 1pt looks
very inflexible. I think the choice is between making
the 3pt with 8/3 6/3 and hitting with 7/2* 6/3. Making
the 3pt obviously gives Black a good 6, but a lot of
other rolls crack Black's prime. If we play 7/2* 6/3
then I don't see any immediate cracking rolls for Black
because he has a spare on his 7pt. I think I'll go for
8/3 6/3.

4. Oh, Paul just posted one like this. Whatever we do,
we shouldn't touch the checker in the 13pt because that
is how we plan to play an ace. I also don't think there
is any real need to fill in the gaps on the 3pt and 5pt.
We'd like to save the 8pt checker to play our 2s. So
9/6 9/4.

5. We want Black to play 3/2* with an ace, so make the 2pt.
8/2 3/2.

6. My first instinct is to jump Black's prime and roll our
prime with 21/15 5/3. An alternative would be to step up
with the 2, 23/21, but then what's the 6? I think it has
to be 21/15 again because we don't want to break our solid
five-prime. It looks a little scary but at DMP we don't
care about getting gammoned. Let's try it. 23/15.

7. I would play to leave no shots if that were possible, but
it's not possible. I still think we are reluctant to leave
more shots than absolutely necessary. We have only two
choices for the 6. If we play 8/2, then 8/6 leaves 13 shots
and 6/4* leaves 15 shots (the other alternatives leave too
many shots IMO). If we play 7/1*, then 8/6 leaves 17 shots
and 3/1 leaves 13 shots, and it also has the advantage of
making the 1pt and forcing Black forward. 7/1* 3/1.

8. For money I'd be thinking about whether I can take the
upcoming double. We have a very vulnerable position so I
don't think it's time to be leaving extra blots around. The
only meaningful point we can make is the 9pt and that comes
at the cost of the 11pt and two blots. Not worth it IMO.
Even 8/2 risks another checker getting sent back in the
ensuing blot-hitting contest. It's ugly but I will go with
8/4 6/4.

9. There sure do seem to be a lot of priming problems this
year! Anyway, one thing I've discovered about my own game is
that I tend to underestimate the value of hitting loose when
I have a four-point board. So 7/1* is a candidate for sure.
If Black rolls an ace, we still have a (broken) five-prime so
it's not the end of the world, and if he rolls a 4---well, he
could have rolled a 3 if we hadn't hit. So the downside isn't
terrible. If we do hit then I think we take the opportunity to
break anchor with 22/18. What if we don't hit? 13/3 looks
awkward, disconnecting our position and misplacing a checker.
22/12 doesn't look better then 22/18 7/1*. I'm going with
22/18 7/1*.

10. Gammon wins are meaningless ATS. The pip count is even.
We just don't want to get hit. Unfortunately there is no
completely safe play. Fewest shots looks like 6/3*/1, which
has the downside of exposing another blot. A "nothing" play
such as 8/5 6/4 gives Black 16 shots (including 55!) and I
think the only other way to leave fewer shots is to play
17/15 6/4 (or 17/15 8/6) which still leaves 15 shots. I will
minimize shots with 6/3*/1. If Black doesn't hit, we should
be able to clean up fairly easily.

---
Tim Chow

Timothy Chow

unread,
May 7, 2022, 7:24:27 PMMay 7
to
It seems that I scored 7/10, which may be the highest score that I've
ever achieved on an Othello quiz. I see that one of the ones I got
wrong was the "easiest" problem as measured by how many participants
got it right!

Congratulations to Yokota Kazuki, who has won the competition three
times in a row (allowing for the fact that there was no competition
in 2020 or 2021).

---
Tim Chow

Timothy Chow

unread,
May 7, 2022, 7:43:14 PMMay 7
to
On 5/7/2022 7:04 PM, I wrote:
> 2. The score might make a difference because we're
> itching for a gammon.  Thus we need to consider the
> banana split 5/4*.  But Black has no other blots to
> pick up, so I'm not sure if 5/4* is called for.  If
> we do play 5/4* then the question is whether 13/8
> (for another builder, but not on a different point)
> or 21/16 (for better board coverage) is the 5.  I think
> I would go with 13/8.  But do we play 5/4* at all?

Not that it matters, because I decided not to play 5/4*,
but I just noticed that I mistyped something. I mean to
say that if I were to play 5/4* then I would go with 21/16.

The Othello Quiz problems are posed at a variety of match
scores, but for many years, the correct answer remained the
same if you were to change all the match scores to unlimited
games. This year, I see that problems 2 and 7 do depend on
the match score, at least in the sense that if you change
them to unlimited, the top play is no longer so clear.

---
Tim Chow

Masanori Itikawa

unread,
May 7, 2022, 8:00:07 PMMay 7
to
2022年5月8日日曜日 8:04:22 UTC+9 Tim Chow:

> Ah, very nice! So what is the story...was everything
> canceled the last year (or two?) and are things back to
> normal as far as the Japan Open is concerned?

Unfortunately, due to COVID-19, the Japan Open was cancelled twice.

peps...@gmail.com

unread,
May 8, 2022, 5:39:45 AMMay 8
to
Since the rollouts are the arbitrator, I think the natural way to rank the participants would
be as follows:
1) For every wrong answer, give the negative evaluation for that answer as in the rollout.
2) For every omitted answer, replace the omission by the worst of the candidate plays.
3) Sum and take the highest sum.
4) In the case of ties, rank the participants according to their solving time.

Of course, if there's only one player with 10/10, and we're only interested in first place, this doesn't matter.

There are some interesting (to me) points of comparison between competitive backgammon-problem-solving and competitive chess-problem-solving.
(As I see it), some points are as follows: [If I use any terms you don't know (and want to know), the answers can easily be ascertained by googling --
for example "chess helpmate".]

1) Backgammon problems seem heavily oriented towards practically useful positions. All of the positions look like they could easily have arisen in practice.
Here, the comparison with chess problem solving could hardly be more marked. Standard chess problem solving has direct mates, endgame studies, helpmates and selfmates.
Helpmates and selfmates don't even make any sense from an ordinary chess-playing point of view. Direct mates make almost no sense, given that chess doesn't confer much
benefit to minimising distance-to-mate. Furthermore, the positions are hardly ever plausible. In particular, Black nearly always has a should-have-resigned position, and White
has a position which makes White's inability to win earlier implausible. Endgame studies do indeed ask for optimal practical play. However, many of the positions are unnatural
here too, as they are selected for being interesting and difficult and clear, with the practical value not a factor in selection.

2) There are a huge number of prestigious (although not heavily funded) chess-problem-solving events. If someone doesn't need to earn money (and there are a surprisingly
large number of such people) and has the funds to travel, it is quite possible to be a full-time chess-problem-solver as there is a yearly calendar of events.

3) Because of point 1), there is quite a difference between the chess communities and chess-problem-solving communities. Over 99% of the world's top chess players don't compete
in chess-problem-solving events. Probably many of them would do poorly even if they did. However, the converse question is also interesting:
How many of the world's top chess-problem-solvers are expert in playing chess? Here, everything
depends on what we mean by "expert". If we mean at least the strength of "expert" level as defined by the USCF, then the answer is probably: "At least 90% of them".
If we mean IM strength or better, then it may be around 25 to 30%. There are many names that are well-known in both communities -- in particular John Nunn and Piotr Murdzia.

4) The prestige of chess-problem-solving in the chess community is far less than backgammon-problem-solving in the backgammon community. I don't think Piotr Murdzia is
particularly well-known despite having been the world's best chess-problem solver.

5) Chess-problem organisations maintain a rating list for chess-problem solvers, which is designed to have a similar scale to OTB chess ratings. I'm not sure if this is done in
backgammon.

Paul

Timothy Chow

unread,
May 8, 2022, 8:54:23 AMMay 8
to
On 5/8/2022 5:39 AM, peps...@gmail.com wrote:
> There are some interesting (to me) points of comparison between competitive backgammon-problem-solving and competitive chess-problem-solving.

I would say that the biggest difference is that backgammon simply
offers less scope for competitive problem solving, by the nature
of the game.

One way to see this is to pose the question, how many backgammon
problems can you compose without appealing to a bot as an oracle?
The answer is, very few. For the standard "what's the best move"
problem, you're pretty much limited to a few non-contact problems.
There is also not a lot of scope for retros or proof games, as you
will quickly discover if you try to compose such a problem. I have
seen a few nice nonstandard problems ("Which play leaves the fewest
shots?") but they are not rich enough to form an entire genre.

Okay, let's say we're willing to countenance a bot as oracle. This
is a huge aesthetic flaw, because for example, standard "what's
the best move" problems cannot be solved with certainty. In chess, if
you find the mate in 3, you can confirm the correctness of the answer
during the test. This is not possible in backgammon if the bot is an
oracle. It also raises the question of which bot you're going to use
and which settings and so on. This is one nice feature of the Othello
Quiz: Othello is very good about choosing positions where there is a
huge equity difference between the right play and any other play. This
means that the problem is "robust" to changes in bot technology. If
we were to use your proposed scoring system, people's rankings might
change with changes in the bot---another inelegant feature.

I have occasionally had the opportunity to compose puzzles for
competitions, in which the type of puzzle was almost completely up
for grabs---it could be a word puzzle, a logic puzzle, a math puzzle,
a trivia puzzle, etc. I definitely considered making a backgammon
puzzle, but couldn't come up with a good one. Unless you have a
population of solvers who is docile enough to accept a bot verdict as
an oracle, it's just not possible.

There's one genre which could possibly work: Proof games or retros
where the preceding moves are all stipulated to have been made by a
bot, and the solver has full access to a bot during the test. The
number of people who would find such problems interesting would,
however, be extremely small.

---
Tim Chow

peps...@gmail.com

unread,
May 8, 2022, 7:14:57 PMMay 8
to
How small would the number be?
I would think either 0 or 1. It would depend on whether you would
find the problems interesting. If so, I'd guess 1. Otherwise, 0.

I like the way people's rankings would change as the technology changes.
That's not a bug -- it's a feature.
It's no different to the way, we reassess how good the past work of writers
and mathematicians was, depending on future developments.

Paul

Timothy Chow

unread,
May 9, 2022, 8:49:53 AMMay 9
to
Nack Ballard finds such problems interesting, probably more than I do.
Stick finds them slightly interesting, though probably less than I do.

> I like the way people's rankings would change as the technology changes.
> That's not a bug -- it's a feature.
> It's no different to the way, we reassess how good the past work of writers
> and mathematicians was, depending on future developments.

Comparison to writers and mathematicians does not seem right to me.
A better comparison would be to other competitions in sports and games.
If new measurement technology overturns an old verdict, that is usually
considered an annoyance that we have to live with in an imperfect world,
and not a desirable feature.

In the case of backgammon, there is a further aesthetic flaw in that the
technology is not just providing a more accurate way of measuring
something that you could measure without it; it has become the focus of
the competition. A sports analogy might be, someone designs a robot
golfer and then we hold competitions where people guess how many strokes
the robot will take to make the hole. It's kind of fun but earning a
grandmaster title in guessing doesn't seem very appealing.

---
Tim Chow
Reply all
Reply to author
Forward
0 new messages