Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Here is a fair bet against bots.
30 views
Skip to first unread message
muratk
Apr 17, 2012, 3:34:52 AM4/17/12
to
I believe I had made a rating based offer in the past but let's try
again anyway.
Who cares if the bots cheat? Who cares if can predict future dice
rolls based on how the bot plays a position? Etc...
Let's make things simple. And afterwards, everybody can make whatever
they want to make out of it.
Here is what I propose:
1 - I will play 100 matches of 25-points for $100 points per match,
against your choice of either Gnugb, XG or Snowie.
2 - I will play over an Internet connection. You will have the choice
of your PRN and seed for each match, which I will not need to know
until the end.
3 - As long as we used an already existing/compiled bot, your bot can
even look ahead and cheat all it wants... ;) You just have to tell me
the PRNs and seeds that you used, so that I can verify the rolls and
make sure that you didn't roll the dice manually... :)
4 - We will not worry about the odds until after the end, at which
time we will derive the odds from the ELO ratings that the bots
assigns to each player, according to the table at
http://www.bkgm.com/faq/Ratings.html#what_are_my_chances_of_winning_.
5 - We will then multiply the stakes by the odds and determine how
much the players will owe each other.
For example:
Let's suppose that at the end the bot comes up with a rating
difference of 200 between me and itself. According to the conversion
table, this would mean that I would have a 24% chance of winning.
Thus, the odds will be 24 to 76, and I will be betting a total of
$2,400 against your $7,600.
Based on this, if end up winning 15 matches against your bot, I will
pay you the $900 difference. On the other hand, if I end up winning 35
matches against your bot, you will pay me the $1,100 difference.
Please keep arguments like "why should anybody bet on even odds", etc.
to yourselves and go flip coins...
I would at least expect the bot developers to take advantage of this
offer, to promote their products by proving the value of their bots as
analysis tools, learning tools, etc. even if not to prove that their
bots are the strongest players on the planet...
MK
again anyway.
Who cares if the bots cheat? Who cares if can predict future dice
rolls based on how the bot plays a position? Etc...
Let's make things simple. And afterwards, everybody can make whatever
they want to make out of it.
Here is what I propose:
1 - I will play 100 matches of 25-points for $100 points per match,
against your choice of either Gnugb, XG or Snowie.
2 - I will play over an Internet connection. You will have the choice
of your PRN and seed for each match, which I will not need to know
until the end.
3 - As long as we used an already existing/compiled bot, your bot can
even look ahead and cheat all it wants... ;) You just have to tell me
the PRNs and seeds that you used, so that I can verify the rolls and
make sure that you didn't roll the dice manually... :)
4 - We will not worry about the odds until after the end, at which
time we will derive the odds from the ELO ratings that the bots
assigns to each player, according to the table at
http://www.bkgm.com/faq/Ratings.html#what_are_my_chances_of_winning_.
5 - We will then multiply the stakes by the odds and determine how
much the players will owe each other.
For example:
Let's suppose that at the end the bot comes up with a rating
difference of 200 between me and itself. According to the conversion
table, this would mean that I would have a 24% chance of winning.
Thus, the odds will be 24 to 76, and I will be betting a total of
$2,400 against your $7,600.
Based on this, if end up winning 15 matches against your bot, I will
pay you the $900 difference. On the other hand, if I end up winning 35
matches against your bot, you will pay me the $1,100 difference.
Please keep arguments like "why should anybody bet on even odds", etc.
to yourselves and go flip coins...
I would at least expect the bot developers to take advantage of this
offer, to promote their products by proving the value of their bots as
analysis tools, learning tools, etc. even if not to prove that their
bots are the strongest players on the planet...
MK
Michael Petch
Apr 17, 2012, 4:28:43 AM4/17/12
to
On 17/04/2012 1:34 AM, muratk wrote:
> Who cares if the bots cheat?
You did.
> Who cares if the bots cheat?
> Who cares if can predict future dice
> rolls based on how the bot plays a position?
I wouldn't take the bet because even I can get pretty decent win
percentage with pumping up the cube on matches, and winning considerably
more matches than my error rate against XG says I should.
Bots assume equal skill but a crafty underdog can play to the bots
weakness. A bot generally assumes both players are equal strength, which
can get it into some trouble.
I might be inclined to believe the bot would have a better chance if it
was all 1ptrs.
If you were to make the claim that Backgammon bots DON'T cheat but have
many weaknesses when compared to humans, I would say "I'd agree
wholeheartedly".
I don't think anyone here says the way a bot plays is infallible when it
comes to their analysis, however I believe they are an invaluable tool
for understanding certain aspects of the game. We may have gained some
insight that we may not have seen without them.
Bots are simply ONE tool, albeit a powerful one.
rebcalale
Apr 18, 2012, 9:12:27 PM4/18/12
to
BOTS cheat pure and simple and yes it is predictable. Now if you want
to argue about how or why they cheat, as always knock yourself out.
But you know, the previous poster knows, anyone who takes the time to
analyze BOT games knows, BOTs CHEAT (in other words, they do not play
as a human is forced to). The previous poster's challenge is a good
one but we both know you will not take it because you as skilled at
lame excuses as anyone I have ever seen.
Michael Petch
Apr 18, 2012, 11:07:03 PM4/18/12
to
On 18/04/2012 7:12 PM, rebcalale wrote:
> Leave for years come back, same old, same old!
It would not be a complete reunion without Crabtree, Reb and Murat on
> Leave for years come back, same old, same old!
here at the same time. Two for three so far lol
Michael Petch
Apr 18, 2012, 11:11:22 PM4/18/12
to
On 18/04/2012 7:12 PM, rebcalale wrote:
> BOTs CHEAT (in other words, they do not play
> as a human is forced to)
I understand "BOTs CHEAT" but I in your other words I have no idea what
> as a human is forced to)
you are trying to say with "in other words, they do not play as a human
is forced to". Can you rephrase it some other words? (Seriously, I
really don't know what is meant, I doubt I am the only one).
muratk
Apr 20, 2012, 6:29:57 AM4/20/12
to
On Apr 17, 2:28 am, Michael Petch <mpe...@capp-sysware.com> wrote:
.
other people do care also. I meant to say, "since you will never doubt
it, let's forget that argument and move on without it"... You could
have understood this, no...?
>> Who cares if can predict future dice
>> rolls based on how the bot plays a position?
>
> You did.
Come on, Petch, it's a literary question. Of course I do and a lot of
other people do care also. I meant to say, "since you will never doubt
it, let's forget that argument and move on without it"... You could
have understood this, no...?
> I wouldn't take the bet because even I can get pretty decent win
> percentage with pumping up the cube on matches, and winning considerably
> more matches than my error rate against XG says I should.
Okay, so, you are admitting that the error rate (if not the rest of
the similar analyses) calculated by XG is plain bullshit...??
So, how much do we deduct for this bullshit from the $39.95 that the
XG costs...???
> Bots assume equal skill but a crafty underdog can play to the bots
> weakness. A bot generally assumes both players are equal strength, which
> can get it into some trouble.
Suck my cock math phd faggot... :)) Is that what they thought you at
school...?
Why doesn't the puto improve his product to input an arbitrary
strength of the human player and then do all its calculations based on
that constant, instead of assuming that both players are of equal
strength...?
How long would it take you putos to implement this improvement...? I
am not asking to rush you. I am asking just to have an idea of what to
expect... ;)
> I might be inclined to believe the bot would have a better chance if it
> was all 1ptrs.
Why...?
> If you were to make the claim that Backgammon bots DON'T cheat but have
> many weaknesses when compared to humans, I would say "I'd agree
> wholeheartedly".
Okay, here is my offer to you: I will retract my claim that the bots
cheat, if you agree that even average human players can beat the shit
out of the bots...
> I don't think anyone here says the way a bot plays is infallible when it
> comes to their analysis, however I believe they are an invaluable tool
> for understanding certain aspects of the game. We may have gained some
> insight that we may not have seen without them.
You contradict yourself and that is a real danger! The bots are not
"infallible" but they are still "invaluable"...? Hah hahaa... You have
a weird notion of value, I would say...
$39.95 is all you "salesmen" can get for your "fallible"
garbage...! :))
> Bots are simply ONE tool, albeit a powerful one.
Bullshit...!! How can you claim that they are powerful if they can't
even gauge the strength of their opponents and play accordingly...?
I mean, we are not even scratching the surface of the "world-class
bullshit" called cube-skill yet... :))
MK
.
> On 17/04/2012 1:34 AM, muratk wrote:
>
>> Who cares if the bots cheat?
>
> You did.
Come on, Petch, it's a literary question. Of course I do and a lot of
>
>> Who cares if the bots cheat?
>
> You did.
other people do care also. I meant to say, "since you will never doubt
it, let's forget that argument and move on without it"... You could
have understood this, no...?
>> Who cares if can predict future dice
>> rolls based on how the bot plays a position?
>
> You did.
other people do care also. I meant to say, "since you will never doubt
it, let's forget that argument and move on without it"... You could
have understood this, no...?
> I wouldn't take the bet because even I can get pretty decent win
> percentage with pumping up the cube on matches, and winning considerably
> more matches than my error rate against XG says I should.
the similar analyses) calculated by XG is plain bullshit...??
So, how much do we deduct for this bullshit from the $39.95 that the
XG costs...???
> Bots assume equal skill but a crafty underdog can play to the bots
> weakness. A bot generally assumes both players are equal strength, which
> can get it into some trouble.
school...?
Why doesn't the puto improve his product to input an arbitrary
strength of the human player and then do all its calculations based on
that constant, instead of assuming that both players are of equal
strength...?
How long would it take you putos to implement this improvement...? I
am not asking to rush you. I am asking just to have an idea of what to
expect... ;)
> I might be inclined to believe the bot would have a better chance if it
> was all 1ptrs.
> If you were to make the claim that Backgammon bots DON'T cheat but have
> many weaknesses when compared to humans, I would say "I'd agree
> wholeheartedly".
cheat, if you agree that even average human players can beat the shit
out of the bots...
> I don't think anyone here says the way a bot plays is infallible when it
> comes to their analysis, however I believe they are an invaluable tool
> for understanding certain aspects of the game. We may have gained some
> insight that we may not have seen without them.
"infallible" but they are still "invaluable"...? Hah hahaa... You have
a weird notion of value, I would say...
$39.95 is all you "salesmen" can get for your "fallible"
garbage...! :))
> Bots are simply ONE tool, albeit a powerful one.
even gauge the strength of their opponents and play accordingly...?
I mean, we are not even scratching the surface of the "world-class
bullshit" called cube-skill yet... :))
MK
mu...@compuplus.net
May 3, 2012, 4:31:09 AM5/3/12
to
MK
0 new messages