Clear beginner's error by XG in competition mode -- XG's programmers need to pay urgent attention.

[Paul]

I didn't know that XG has a feature where it impersonates TV characters.
However, in the below position, XG did an excellent impersonation of
Joey Tribbiani from the sitcom "Friends" by playing 5/4 1/off(3). With the
gammon count being hugely in XG's favour, XG needed to clear points
by 5/4(4).
How bizarre, how bizarre!

Paul

XGID=-I-----------------AdbcDeA:1:1:-1:11:12:10:0:13:10

X:eXtremeGammon O:Daniel
Score is X:10 O:12 13 pt.(s) match.
+13-14-15-16-17-18------19-20-21-22-23-24-+
| | | O | +---+
| | | O | | 2 |
| | | O | +---+
| | | O |
| | | 9 |
| |BAR| |
| | O | X |
| | | X O X |
| | | X X O X |
| | | X X X O X |
| | | O X X X O X |
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 42 O: 145 X-O: 10-12/13
Cube: 2, O own cube
X to play 11

1. 3-ply 5/4 1/Off(3) eq:+2.818
Player: 99.80% (G:91.11% B:12.39%)
Opponent: 0.20% (G:0.00% B:0.00%)

2. 3-ply 5/4(2) 1/Off(2) eq:+2.803 (-0.015)
Player: 99.69% (G:90.45% B:8.41%)
Opponent: 0.31% (G:0.00% B:0.00%)

3. 3-ply 5/4 5/3 1/Off eq:+2.779 (-0.039)
Player: 99.57% (G:89.41% B:6.34%)
Opponent: 0.43% (G:0.00% B:0.00%)

4. 3-ply 5/3 1/Off(2) eq:+2.771 (-0.047)
Player: 99.61% (G:88.96% B:9.14%)
Opponent: 0.39% (G:0.00% B:0.00%)

5. 2-ply 5/4(4) eq:+2.749 (-0.070)
Player: 99.59% (G:87.84% B:4.47%)
Opponent: 0.41% (G:0.00% B:0.00%)


eXtreme Gammon Version: 2.10, MET: Kazaross XG2

[Michael]

I see your point but it's also a matter of backgammons.The fastest it bears off the more the bg wins.
I believe what it actually played is both reasonably safe and reasonably fast

[Paul]

No, there aren't any backgammons at this match score. In my opinion, XG
shouldn't include the bg stats in the table because it's misleading, but
that's a matter of taste.
For money (or other scores where backgammons count), XG's play might be
best. I really can't understand it. It must be a bug.

Paul

[BlueDice]

Yes, very strange, gnubg rollout agrees with XG rollout and by a similar equity difference! {5/4 1/off(3) vs 5/4(4)}
--
BD

[bananab...@gmail.com]

I am having a real problem understanding what you're saying here. 5/4 1/off (3) is the right play and that's coming from me, not XG. I also don't get the Friends reference.

Stick

[Michael]

Yes you are right backgammons don't count here.
Although I agree with you that the way it played doesn't look "reasonable" the only way to check if it's in fact the best move or not is to actually count rolls in depth of possibly 5 plies.Who would care do such a boring thing anyway so let's rely on bot's results. The bot says that about 1 out of 10 times O saves the gammon! For that to happen she has to hit.
So is what it played safer than 5/4(4) when analyzed in depth of 5??
I personally don't know, but considering that if X rolls a 2 in both next 2 rolls he exposes a blot, it might be.
If you want to keep it simple, just think that about 50% of the times X would expose a blot anyhow. Strategic-wise, the more checkers he would have beared off by that time, the more his gammon chances are.

I hope my explanation helps.

[Michael]

You could help by explaining the reasons.

[bananab...@gmail.com]

I am waiting to get verification that he (everyone??) is saying that Xg's play is wrong and that 5/4(4) should be the best play. I don't want to waste time otherwise. XG's play is certainly correct.

Stick

[peps...@gmail.com]

The right play is counter-intuitive to non-experts because most would assume that we get a gammon if we avoid leaving a shot and should therefore clear points.

I thought XG was badly wrong, here. Joey, from Friends, says dumb things
all the time, and I thought XG was making a dumb play. That's the Friends
reference. I think the point is that running to save the gammon, even without getting and hitting a shot, is more likely than it seems. That's why XG needs to rip checkers off.

Paul

[Paul]

Yes, I thought 5/4(4) was the right play.
I did catch XG in an error once, and that's why I was (wrongly) confident.
I found a position in money play where XG does a play which loses a
backgammon 100% of the time whereas XG can save the backgammon with a
probability of around 1%. Some of those backgammon saves are gammon saves
or even wins. If I have a checker on my two point (and no other checkers) and XG has made my ace point, than XG can (quite bizarrely) leave the acepoint anchor to guarantee a bg loss, rather than hoping for me to roll 11.

Paul

[bananab...@gmail.com]

I watch Friends almost every day of my life. Joey says dumb things!?

https://www.youtube.com/watch?v=8bo54wE_rJU

He's the only one to say anything intelligent in this entire convo from how long to wait before hitting on someone, to what really matters, to why we don't listen to cows!

Notice XG's play wins more games and more gammons.

Stick

[Paul]

I think the point is that O doesn't need to hit a shot to save the gammon.
With my play, the crossover count is 20 to 14 with O on roll, and X will
often be only to bear one off.
So X needs to remove as many checkers as possible.
I think that my play is by far the safest from the point of view of avoiding
leaving shots.

Paul

[bananab...@gmail.com]

Your play is not 'by far the safest' from the point of view of leaving shots. First, after your play right away when you throw a two you see you're not happy as it generally leaves you odd on the top and strips the point in front of the anchor meaning the next deuce will either create a gap itself or leave a shot.

What your play also fails to do is remove checkers from the board so that if you are ever hit you'll have a lot more men off so you'll be a lot more likely to win the game. (and gammon if needed) Even if you give the opp. as perfect of a board as you can and make gammons not matter (dmp) you'll see that 4 1/off(3) is right. XGID=---BBBBA-----------AdbcDeA:0:0:-1:11:0:0:0:1:10

Stick

[Tim Chow]

On Sunday, February 19, 2017 at 6:11:43 PM UTC-5, bananab...@gmail.com wrote:
> 5/4 1/off (3) is the right play and that's coming from me, not XG.

That's what I would have played, too. A standard thing to do is to look at
the distance between my highest point and the anchor and see how those numbers
play. If I clear the 5pt then the distance between my highest point and the
anchor is 2. How do my 2's play? Not so well; they're forced: 3/1. A couple
of these and I could be looking at a blot.

---
Tim Chow

[mu...@compuplus.net]

On February 19, 2017 at 1:46:13 PM UTC-7, BlueDice wrote:

> On February 19, 2017 at 10:42:22 AM UTC, Paul wrote:

>> Joey Tribbiani from the sitcom "Friends" by playing 5/4 1/off(3).
>> With the gammon count being hugely in XG's favour, XG needed to
>> clear points by 5/4(4). How bizarre, how bizarre!

> Yes, very strange, gnubg rollout agrees with XG rollout and by
> a similar equity difference! {5/4 1/off(3) vs 5/4(4)}

Since I am personally convinced, from the evidence I had shared in
other threads, that XG was built on "stolen" (i.e. in violation of
GPL) GNUBG code, I was alerted that in this position Snowie 3-ply
does indeed consider 5/4(4) as the top choice.

Another little incest that the "bot-whores" in the "XG bordello"
won't want to tal about...??

MK

[Tim Chow]

On Friday, February 24, 2017 at 3:37:58 AM UTC-5, mu...@compuplus.net wrote:
> Since I am personally convinced, from the evidence I had shared in
> other threads, that XG was built on "stolen" (i.e. in violation of
> GPL) GNUBG code, I was alerted that in this position Snowie 3-ply
> does indeed consider 5/4(4) as the top choice.
>
> Another little incest that the "bot-whores" in the "XG bordello"
> won't want to tal about...??

Hilarious!! So now if two bots both like making a good play they must be
stealing from each other? Did you notice that both XG and GNU play 8/5 6/5
with an opening 31? Clearly, that is absolute proof XG has stolen GNU's code.

Keep the entertainment coming, Jack!
---
Tim Chow

[mu...@compuplus.net]

On February 24, 2017 at 5:03:15 PM UTC-7, Tim Chow wrote:

> On February 24, 2017 at 3:37:58 AM UTC-5, mu...@compuplus.net wrote:

>> ... Snowie 3-ply does indeed consider 5/4(4) as the top choice.

>> Another little incest that the "bot-whores" in the "XG bordello"
>> won't want to tal about...??

First, let me give you another chance to notice that I had noted
the above just as "another little" example in addition to other
previously stronger evidences like resigning bugs and still with
lingering question marks at the end...!!

> Hilarious!! So now if two bots both like making a good play they
> must be stealing from each other?

Hilarious is your settling to call XG/GNUBG's top choice "a good
play" instead of "the best play". :)

Are you then saying that Snowie's top choice 5/4(4) "a bad play"?

I can't wait to be entertained by seeing you and your ilk try to
"prove" that XG/GNUBG's top choice is better than Snowie's.

Until then, XG/GNUBG's top choice that the original poster thought
to be a programming "bug" may indeed be an acquired (thus shared)
"bug" through "bot fucking" (code stealing;)...

In fact, which play is best may not even matter when both XG and
GNUBG's top 7 choices (from 5/4 1/off(3) down to 5/4(4)) are in
the same order (except for one in the middle) with almost exactly
the same equity differences between them, is in itself a little
too much of a coincidence to suspect "bot fucking".

> Did you notice that both XG and GNU play 8/5 6/5 with an opening
> 31? Clearly, that is absolute proof XG has stolen GNU's code.

Now, this is a good example of your simpleton brain's inability
to offer a better strawman argument. Sad, senile idiot! :(

MK

[Paul]

On Sunday, February 26, 2017 at 6:11:05 AM UTC, mu...@compuplus.net wrote:

> Are you then saying that Snowie's top choice 5/4(4) "a bad play"?

I think that Snowie's top choice 5/4(4) is a bad play (taking your word
for it that you used Snowie correctly and are correct that this is
Snowie's choice).

From their postings on the thread, Tim and Stick seem to believe that
Snowie's choice is a bad play. Any play for which there is strong evidence
that it is suboptimal (loses MWC) is bad by definition.

Paul

[Tim Chow]

On Sunday, February 26, 2017 at 1:11:05 AM UTC-5, mu...@compuplus.net wrote:
> "bot fucking"
[...]
> "bot fucking".

10 points for profanity, Murat! I'll throw in another 5 points for your
total logical incoherence and for "waiting to be entertained" by things that
Stick and I already posted a while ago. Keep it coming, Jack!

---
Tim Chow

[mu...@compuplus.net]

On February 19, 2017 at 5:34:43 PM UTC-7, Paul wrote:
> ... If I have a checker on my two point (and no other

> checkers) and XG has made my ace point, than XG can
> (quite bizarrely) leave the acepoint anchor to guarantee
> a bg loss, rather than hoping for me to roll 11.

Do you remember when was this and/or what was the position?

Was it talked about here in RGB or in other forums?

Has is been fixed since then or can it still be reproduced?

MK

[mu...@compuplus.net]

On February 26, 2017 at 3:10:38 AM UTC-7, Paul wrote:

> On February 26, 2017 at 6:11:05 AM UTC, mu...@compuplus.net wrote:

>> Are you then saying that Snowie's top choice 5/4(4) "a bad play"?

> I think that Snowie's top choice 5/4(4) is a bad play (taking
> your word for it that you used Snowie correctly and are correct
> that this is Snowie's choice).

I was commenting based on what you/others have said. Personally,
I can't say that I understand the plies, rollout variables, etc.
for any given bot, let alone how they correspont among many bots,
and neither do I attach much value to such subjective numbers.

BlueDice sounded like he had done XG and GNUBG rollouts but he
hasn't posted them. It would be good to those, along with one
done for Snowie, done by some people competent in rollouts.

> From their postings on the thread, Tim and Stick seem to believe
> that Snowie's choice is a bad play. Any play for which there is
> strong evidence that it is suboptimal (loses MWC) is bad by
> definition.

How do you know that the "evidence" fits the "definition"?

Those minute differences in winning chances calculations would
only change here after contact and would be based on "one bot
playing against itself" using arbitrarily concicted formulas,
and thus will be "subjective", "unreliable", "unverifyable",
"unprovable"...

So, the best that we can do is compare bots to one another and
compare our own opinions.

What does it matter if some people think rolling a 2 twice in
a row may leave a blot, etc...? With XG's play, a 65 on the
very next roll leaves a blot. Snowie's play (also mine and your
initial choice) opens up the 5 point increasing the odds for
the opponent to enter sooner and thus reduce his hitting
combinations, etc. etc....

What really interests me here is whether XG's play may indeed
be a "bug". Because if it is, I would take GNUBG's having the
same bug as "another strong evidence" that XG is still based
on GNUBG's GPL code.

MK

[mu...@compuplus.net]

On February 26, 2017 at 5:12:56 PM UTC-7, Tim Chow wrote:

> On February 26, 2017 at 1:11:05 AM UTC-5, mu...@compuplus.net wrote:

> > "bot fucking"
> [...]
> > "bot fucking".

> 10 points for profanity, Murat!

I'm glad you like it. Often times I think I could be a good
creative writer. Well, good enough to write country songs,
anyway... :)

> I'll throw in another 5 points for your total logical
> incoherence and for "waiting to be entertained" by
> things that Stick and I already posted a while ago.

I saw the "*things*" that Stick and you posted but I didn't
think they added anything to the discussion.

They were just a bunch of wortless, well, "*things*". :)

You know that calling them analysis, opinion, speculation,
etc. would be exaggeration and that's why you resort to
using meaningless, filler words like "*things*". :(

Let's see you "thingers" post some rollouts and such... ;)

MK

[Tim Chow]

On Sunday, February 26, 2017 at 9:22:20 PM UTC-5, mu...@compuplus.net wrote:
> I saw the "*things*" that Stick and you posted but I didn't
> think they added anything to the discussion.

Of course not, since you're incapable of accurate thinking.

You should stick to doing what you know best---swearing incoherently at
anyone who makes a good point!

---
Tim Chow

[mu...@compuplus.net]

On February 26, 2017 at 9:24:02 PM UTC-7, Tim Chow wrote:

> Of course not, since you're incapable of accurate thinking.

"Thinking"..? Or did you mean "thinging"..??

As in: I thing, you thing, Stick things... :))

MK