info
Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
BGBlitz 2.8 is online
285 views
Skip to first unread message
bgbl...@googlemail.com
Nov 2, 2014, 1:36:57 PM11/2/14
to
Hi all,
BGBlitz 2.8.0 is available. Finally! It took even longer than 2.7. And no, again I was not lazy. But when you burn the midnight oil for more than a decade you pile up debts in your program structure. I decided to take the time and clean up things, which took a lot of time. More than I liked. And there are around 150 improvements, bug fixes and new features. The most important:
The tutor of BGBlitz has a new ability rating your moves in real-time. This doesn't sound too spectacular, but the feedback is immediate after your checker move
- it doesn't interrupt your flow
- you have an idea of the error size
- you get not only critics but positive feedback too which is important under motivational aspects.
Hard to explain how it feels but imagine having a coach in the background, nodding his head or rising his eye brow....
There is a small screencast showing this feature in action. It is really addictive. ( http://www.bgblitz.com/screencast/TutorRating_EN.mp4)
Your mileage may vary, but for me, the aesthetics of the game were always important. For the last year, I've worked with Debbie Curtis, a U.S. designer and former Mac consultant, who created a bunch of marvelous boards -- around 50 big ones, at 1490 x 1036 pixels. In the hunt for the perfect board, she nudged me to expand BGBlitz's visual interface capabilities (e.g., multiple checker faces, flexible cube and point numbers locations, bigger thumbnail previews). She donated four boards free of charge, and you can buy the others on her website www.dcbackgammon.com . They're very affordable (costs less than a beer in most countries and you would be surprised how much work goes into them!). As a little thank you to her and to my BGBlitz customers, I bought 250 licenses to give away to the next 125 new buyers (upgrade from Player version counts, too) and the first 125 existing buyers. All you have to do is send me an EMail (subject: "free theme") and I'll give you a special code to get any theme of your choice -- free, no strings attached!
You can see some appetizers on my website.
For the same reason I added some more well looking look&feels so you have more choices to make your favorite theme look even better (Windows and Linux only).
Naturally some effort was put in the AI. I experimented a lot, trained more than a dozen neural nets (each trained about two month), fixed a weakness in the cube heuristics and here is the new AI, not surprisingly called TachiAI IV. In the Depreli 2010 Benchmark it achieves 43.4 (down from 66. Lower is better) with 3-ply and 37.5 with 4-ply which is a decent improvement. At the same time the speed, especially for 3 and 4-ply was improved (about twice as fast) and BGBlitz now supports Match Equity tables in the XG-format.
I play a lot on http://www.dailygammon.com. It's like correspondence Backgammon on steroids. How often you want to know "was you're move right?" or "Should I have dropped the cube?", but you are too lazy to setup the position. And waiting until the game is finished takes too long (sometimes weeks). BGBlitz comes to the rescue. Just drop the URL on BGBlitz board and you know whether you were right. Here is a screencast showing this feauture. (http://bgblitz.com/screencast/Dailygammon.mp4) Naturally you have to use it ethically and only look after you have moved or made your cube decision!!
I added another random number generator. It is recorded real random from "quantum randomness of photon arrival times" (see https://qrng.physik.hu-berlin.de). I included 1 million random dice, what should suffice for some time.....
Support for Longgammon is added (a Backgammon variation with a different starting position)
Numerous further improvements and some bug fixes among them 2 fatal errors and a dozen serious errors
Many thanks to Anders Olsen, Charlie Dick, Chrilly Donninger, Claudio Rescoigno, Dave Bellows, Debbie Curtis, Erich Bloch, Gert Ereth, Gregoire Cornu, Jörg Picard, Kevin Whyte, Markus Goemmel, Michael Petch, Norbert Dehnbostel, Otto Nadolny, Paul Churchill, Peter Grotrian, Philippe Michel, Radu-Dan Sabau, Ralph Landkammer, Sultan Rotrout, William Maslen and to all of the people who have provided error reports, making suggestions or helping in any other way to improve BGBlitz.
The update is free!
Just download the free version from http://www.bgblitz.com/download.html . If you have installed the license in your current version with drag&drop, or have at least 2.7.0 the license will be recognized automatically. In the worst case you have to drag&drop the license file again on the board if BGBlitz runs.
If you use Windows please uninstall the old version first.
I hope you enjoy the new version. I always glad getting some feedback. If you find some errors or quirks please drop me an email.
regards Frank
BGBlitz 2.8.0 is available. Finally! It took even longer than 2.7. And no, again I was not lazy. But when you burn the midnight oil for more than a decade you pile up debts in your program structure. I decided to take the time and clean up things, which took a lot of time. More than I liked. And there are around 150 improvements, bug fixes and new features. The most important:
The tutor of BGBlitz has a new ability rating your moves in real-time. This doesn't sound too spectacular, but the feedback is immediate after your checker move
- it doesn't interrupt your flow
- you have an idea of the error size
- you get not only critics but positive feedback too which is important under motivational aspects.
Hard to explain how it feels but imagine having a coach in the background, nodding his head or rising his eye brow....
There is a small screencast showing this feature in action. It is really addictive. ( http://www.bgblitz.com/screencast/TutorRating_EN.mp4)
Your mileage may vary, but for me, the aesthetics of the game were always important. For the last year, I've worked with Debbie Curtis, a U.S. designer and former Mac consultant, who created a bunch of marvelous boards -- around 50 big ones, at 1490 x 1036 pixels. In the hunt for the perfect board, she nudged me to expand BGBlitz's visual interface capabilities (e.g., multiple checker faces, flexible cube and point numbers locations, bigger thumbnail previews). She donated four boards free of charge, and you can buy the others on her website www.dcbackgammon.com . They're very affordable (costs less than a beer in most countries and you would be surprised how much work goes into them!). As a little thank you to her and to my BGBlitz customers, I bought 250 licenses to give away to the next 125 new buyers (upgrade from Player version counts, too) and the first 125 existing buyers. All you have to do is send me an EMail (subject: "free theme") and I'll give you a special code to get any theme of your choice -- free, no strings attached!
You can see some appetizers on my website.
For the same reason I added some more well looking look&feels so you have more choices to make your favorite theme look even better (Windows and Linux only).
Naturally some effort was put in the AI. I experimented a lot, trained more than a dozen neural nets (each trained about two month), fixed a weakness in the cube heuristics and here is the new AI, not surprisingly called TachiAI IV. In the Depreli 2010 Benchmark it achieves 43.4 (down from 66. Lower is better) with 3-ply and 37.5 with 4-ply which is a decent improvement. At the same time the speed, especially for 3 and 4-ply was improved (about twice as fast) and BGBlitz now supports Match Equity tables in the XG-format.
I play a lot on http://www.dailygammon.com. It's like correspondence Backgammon on steroids. How often you want to know "was you're move right?" or "Should I have dropped the cube?", but you are too lazy to setup the position. And waiting until the game is finished takes too long (sometimes weeks). BGBlitz comes to the rescue. Just drop the URL on BGBlitz board and you know whether you were right. Here is a screencast showing this feauture. (http://bgblitz.com/screencast/Dailygammon.mp4) Naturally you have to use it ethically and only look after you have moved or made your cube decision!!
I added another random number generator. It is recorded real random from "quantum randomness of photon arrival times" (see https://qrng.physik.hu-berlin.de). I included 1 million random dice, what should suffice for some time.....
Support for Longgammon is added (a Backgammon variation with a different starting position)
Numerous further improvements and some bug fixes among them 2 fatal errors and a dozen serious errors
Many thanks to Anders Olsen, Charlie Dick, Chrilly Donninger, Claudio Rescoigno, Dave Bellows, Debbie Curtis, Erich Bloch, Gert Ereth, Gregoire Cornu, Jörg Picard, Kevin Whyte, Markus Goemmel, Michael Petch, Norbert Dehnbostel, Otto Nadolny, Paul Churchill, Peter Grotrian, Philippe Michel, Radu-Dan Sabau, Ralph Landkammer, Sultan Rotrout, William Maslen and to all of the people who have provided error reports, making suggestions or helping in any other way to improve BGBlitz.
The update is free!
Just download the free version from http://www.bgblitz.com/download.html . If you have installed the license in your current version with drag&drop, or have at least 2.7.0 the license will be recognized automatically. In the worst case you have to drag&drop the license file again on the board if BGBlitz runs.
If you use Windows please uninstall the old version first.
I hope you enjoy the new version. I always glad getting some feedback. If you find some errors or quirks please drop me an email.
regards Frank
Andrew B.
Nov 3, 2014, 2:06:36 PM11/3/14
to
>regards Frank
Nice to see you. I enjoy bgblitz as an opponent and advisor, even though I have
the others.
Nice to see you. I enjoy bgblitz as an opponent and advisor, even though I have
the others.
mu...@compuplus.net
Nov 4, 2014, 6:34:04 AM11/4/14
to
On Sunday, November 2, 2014 11:36:57 AM UTC-7, bgbl...@googlemail.com wrote:
> BGBlitz 2.8.0 is available. Finally!
I'm glad that some third party efforts are being made in the
> BGBlitz 2.8.0 is available. Finally!
two party dictatorship of Republican XGR and Democrat GNU...
Hoping that this bot is not just a wannabe of the others, I
have a bunch of questions:
1- Is this bot "genetically" same/similar or different than
GNU/XGR/Snowie/etc...?
A- If different:
A-a) How strong is it compared to others by its own measure?
Please feel free to elaborate on this as you may like...
A-b) Whether it is stronger or weaker than the others, is it
more humanly realistic and empirically proven?
A-c) Will you accept a non-monetary human challenge to your
bot and help conduct an open experiment?
A-d) Will you accept a monetary human challenge to your bot
based on prorated even odds, i.e. the proportionate winning
chances of the two sides according to your bot's rating of
their respective checker and cube skills?
B- If same/similar:
B-a) How strong is it compared to others by the measure/s of
GNY and/or XGR?
B-b) If strength is not an issue, what is the purpose of this
bot? Prettier boards??
B-c) If strength is an issue, will you accept a non-monetary
human challenge to your bot and help conduct an open experiment?
My purpose in this would be to take other bots (proportionately)
down along with yours, despite their avoiding such an experiment.
B-d) If strength is an issue, even if not claiming to be a strong
or strongest bot, will you accept a monetary human challenge to
your bot based on prorated even odds, i.e. proportionate winning
chances of the two sides according to either GNU's or XGR's rating
of their respective checker and cube skills?
C- Along and/or aside the foregoing:
C-a) Would anyone else like to substitute for or supplement the
developer of this bot as outlined above?
MK
Michael Petch
Nov 4, 2014, 5:51:38 PM11/4/14
to
On 2014-11-04 4:34 AM, mu...@compuplus.net wrote:
> On Sunday, November 2, 2014 11:36:57 AM UTC-7, bgbl...@googlemail.com wrote:
>
>> BGBlitz 2.8.0 is available. Finally!
>
> I'm glad that some third party efforts are being made in the
> two party dictatorship of Republican XGR and Democrat GNU...
>
> Hoping that this bot is not just a wannabe of the others, I
> have a bunch of questions:
You were probably too busy posting about Jellyfish to notice that
> On Sunday, November 2, 2014 11:36:57 AM UTC-7, bgbl...@googlemail.com wrote:
>
>> BGBlitz 2.8.0 is available. Finally!
>
> I'm glad that some third party efforts are being made in the
> two party dictatorship of Republican XGR and Democrat GNU...
>
> Hoping that this bot is not just a wannabe of the others, I
> have a bunch of questions:
BgBlitz came out in the mid 90s, maybe '95 or '96. GNU Backgammon isn't
even that old since it didn't show up until about '99.
bgbl...@googlemail.com
Nov 6, 2014, 4:30:26 PM11/6/14
to
Hi Murat,
> 1- Is this bot "genetically" same/similar or different than
> GNU/XGR/Snowie/etc...?
I'm not sure what is genetically means for you, but it is a neural net based. Different to Gnu it's only trained by self-play. XG/Snowie training details are not known
> A-a) How strong is it compared to others by its own measure?
> Please feel free to elaborate on this as you may like...
at the depreli 2010 benchmark it now has 43.5 at 3-ply and 37.5 at 4-ply. This is better than Snowie but worse than XG. New GnuBG seems to be nearly caught up with XG2
> A-b) Whether it is stronger or weaker than the others, is it
> more humanly realistic and empirically proven?
I doubt that there is a good measurement for that beyond: more errors ;)
> A-c) Will you accept a non-monetary human challenge to your
> bot and help conduct an open experiment?
if the setting and the amount of time is reasonable: why not.
> A-d) Will you accept a monetary human challenge to your bot
> based on prorated even odds, i.e. the proportionate winning
> chances of the two sides according to your bot's rating of
> their respective checker and cube skills?
don't got that.
> B-b) If strength is not an issue, what is the purpose of this
> bot? Prettier boards??
It's main purpose is that I have fun doing it and talking with interesting people. Be assured that if I invested a fraction of the time that I invested in BGBlitz in normal free lancer work I would have made a not so small fortune....
2nd: if you have one excellent bot this doesn't mean that you could achieve a deeper understanding if you have additional bots. As Ed'O coined it on TV: "cross check, cross check cross check". Read Bob Wachtels recent book. page 88f especially.
> B-c) If strength is an issue, will you accept a non-monetary
> human challenge to your bot and help conduct an open experiment?
> My purpose in this would be to take other bots (proportionately)
> down along with yours, despite their avoiding such an experiment.
see above.
> B-d) If strength is an issue, even if not claiming to be a strong
> or strongest bot, will you accept a monetary human challenge to
> your bot based on prorated even odds, i.e. proportionate winning
> chances of the two sides according to either GNU's or XGR's rating
> of their respective checker and cube skills?
see above.
ciao
Frank
> 1- Is this bot "genetically" same/similar or different than
> GNU/XGR/Snowie/etc...?
> A-a) How strong is it compared to others by its own measure?
> Please feel free to elaborate on this as you may like...
> A-b) Whether it is stronger or weaker than the others, is it
> more humanly realistic and empirically proven?
> A-c) Will you accept a non-monetary human challenge to your
> bot and help conduct an open experiment?
> A-d) Will you accept a monetary human challenge to your bot
> based on prorated even odds, i.e. the proportionate winning
> chances of the two sides according to your bot's rating of
> their respective checker and cube skills?
> B-b) If strength is not an issue, what is the purpose of this
> bot? Prettier boards??
2nd: if you have one excellent bot this doesn't mean that you could achieve a deeper understanding if you have additional bots. As Ed'O coined it on TV: "cross check, cross check cross check". Read Bob Wachtels recent book. page 88f especially.
> B-c) If strength is an issue, will you accept a non-monetary
> human challenge to your bot and help conduct an open experiment?
> My purpose in this would be to take other bots (proportionately)
> down along with yours, despite their avoiding such an experiment.
> B-d) If strength is an issue, even if not claiming to be a strong
> or strongest bot, will you accept a monetary human challenge to
> your bot based on prorated even odds, i.e. proportionate winning
> chances of the two sides according to either GNU's or XGR's rating
> of their respective checker and cube skills?
ciao
Frank
Tim Chow
Nov 6, 2014, 4:39:13 PM11/6/14
to
On Thursday, November 6, 2014 4:30:26 PM UTC-5, bgbl...@googlemail.com wrote:
> > A-d) Will you accept a monetary human challenge to your bot
> > based on prorated even odds, i.e. the proportionate winning
> > chances of the two sides according to your bot's rating of
> > their respective checker and cube skills?
> don't got that.
He means, will BGBlitz estimate, based on error-rate calculations, what your odds are of beating it in (say) an 11-point match? And if so, are you willing to bet money that those estimates are accurate? For example, say that BGBlitz looks at Murat's error rate and decides that BGBlitz would be a 3-to-1 favorite against him in an 11-point match; would you be willing to give 3-to-1 odds against Murat?
> > A-d) Will you accept a monetary human challenge to your bot
> > based on prorated even odds, i.e. the proportionate winning
> > chances of the two sides according to your bot's rating of
> > their respective checker and cube skills?
> don't got that.
Murat doesn't believe in these estimates and thinks that someone else in the world besides himself would be impressed if he were able to beat those estimated odds.
---
Tim Chow
greedygammon
Nov 7, 2014, 4:48:50 PM11/7/14
to
On Tuesday, November 4, 2014 2:51:38 PM UTC-8, Michael Petch wrote:
> On 2014-11-04 4:34 AM, muratwrote:
when gnubg splashed onto the scene with a deafening roar, snowie and jellyfish were sent scampering back into their den. Yeah... OPEN SOURCE BITCHES!! HAHAHAHA. It was a while after that when Frank transitioned his "swiss knife of backgammon" not sure what it was database of position? did it play? into the fine AI player that it is now. I believe he got some help from the honorable Professor Gary Wong. And XG followed after. Gnubg was then and still is a priceless gift to the bg world. For all those suckers who got snowballed out of $200-$380 on the commercial versions of Jf/Snowie, STOP HATING ON GNUBG!!!. Although I must admit, If I paid $300 for something that got duplicated by a free version, I would be a bit miffed too. What about the owners of snowie and jf, I wonder if they considered hiring a hit man to take out Gary Wong? I heard he was in hiding for a while. LOL.
The only backgammon program that offers both online and offline games GREEDYGAMMON yeah!! www.greedygammon.com
maareyes
> On 2014-11-04 4:34 AM, muratwrote:
The only backgammon program that offers both online and offline games GREEDYGAMMON yeah!! www.greedygammon.com
maareyes
michae...@gmail.com
Nov 8, 2014, 9:02:05 AM11/8/14
to
On Friday, November 7, 2014 11:48:50 PM UTC+2, greedygammon wrote:
> The only backgammon program that offers both online and offline games GREEDYGAMMON yeah!! www.greedygammon.com
>
> maareyes
Yep! And you will not even notice the difference :-)
> The only backgammon program that offers both online and offline games GREEDYGAMMON yeah!! www.greedygammon.com
>
> maareyes
mu...@compuplus.net
Nov 11, 2014, 5:41:35 AM11/11/14
to
On Thursday, November 6, 2014 2:30:26 PM UTC-7, bgbl...@googlemail.com wrote:
> Hi Murat,
Thanks for your responses. I since downloaded your bot and played with it
for a few days. It looks nice. It flows well. It's quite strong and very
fast at the same time.
Apparently it has many special features not found in other bots but except
a couple of them, they are not enough to prefer it over gnubg (actually, I
don't use the similar features on gnubg either) but I want to encourage you
keep improving it and may even try to offer you some suggestions.
One thing I liked is being able to set and see the random dice counter. It
was already available even in jellyfish and I never understood why it was
left out in other bots. Without repeating the reasons I had given earlier,
I think it is a valuable feature.
I was also intrigued by the doubling settings of cautious/normal/aggressive,
which caused me to reinstall your bot to take a second look at it after I
had already uninstalled it. I assume it allows the user to select the "cube
window"? I'll come back to this later on.
>> 1- Is this bot "genetically" same/similar or different than
>> GNU/XGR/Snowie/etc...?
> I'm not sure what is genetically means for you, but it is a neural
> net based. Different to Gnu it's only trained by self-play. XG/Snowie
> training details are not known
Okay, that much was already obvious from your post so maybe I should have
asked instead if it's "philosophically" different? In other words, do you
take the same approach to making checker/cube decisions, etc.?
>> A-c) Will you accept a non-monetary human challenge to your
>> bot and help conduct an open experiment?
> if the setting and the amount of time is reasonable: why not.
Let's put this aside for now, as I feel we may be able to cooperate to do
some interesting experiments without resorting to this.
>> A-d) Will you accept a monetary human challenge to your bot
>> based on prorated even odds, i.e. the proportionate winning
>> chances of the two sides according to your bot's rating of
>> their respective checker and cube skills?
> don't got that.
Did you understand better after Chow's explanation? Anyway, same as above,
let's postpone this for now.
> Be assured that if I invested a fraction of the time that I invested in BGBlitz in normal free lancer work I would have made a not so small fortune....
It's great that you are not purely motivated by money. Perhaps you will be
more open to alternative ideas and put some time/effort into experimenting
with them.
If you have followed some of my arguments here over the years, you probably
know my lack of esteem for of the so-called "cube skill". In a few words,
it is based on arguments like: "Everything else being equal between two bots
and/or bot-like humans, the side who gains the initial advantage will never
lose it!" (after 4 billion games;)...
Therefore, I argue that the "cube window" totally bogus. If we are going to
insist on using the cube, we need to handle it differently and accept that
it may not even require all that great of a skill.
Again, I don't want to repeat what I had posted about my experimenting with
doubling within the first 2-3 rolls against gnubg depending on the dice and
I don't know if anyone else experimented with it with a genuinely open mind
but doing it with two slightly different versions of the same bot may be
more valuable/convincing to bot worshippers.
How hard would it be for you to allow the user to adjust the "cubing window"
to whatever the value he wishes (instead of just cautious/normal/aggressive)?
It looks like the functionality is there already and it should be a matter
of setting the variable to any value instead of preselected three.
Apparently, your bot also already has the feature to play against another
bot(?) If so, I would be very much interested in having the normal bgblitz
bot play against another bgblitz bot identical except in its cube handling
like I would... :)
Would you be willing to help with this?
MK
> Hi Murat,
Thanks for your responses. I since downloaded your bot and played with it
for a few days. It looks nice. It flows well. It's quite strong and very
fast at the same time.
Apparently it has many special features not found in other bots but except
a couple of them, they are not enough to prefer it over gnubg (actually, I
don't use the similar features on gnubg either) but I want to encourage you
keep improving it and may even try to offer you some suggestions.
One thing I liked is being able to set and see the random dice counter. It
was already available even in jellyfish and I never understood why it was
left out in other bots. Without repeating the reasons I had given earlier,
I think it is a valuable feature.
I was also intrigued by the doubling settings of cautious/normal/aggressive,
which caused me to reinstall your bot to take a second look at it after I
had already uninstalled it. I assume it allows the user to select the "cube
window"? I'll come back to this later on.
>> 1- Is this bot "genetically" same/similar or different than
>> GNU/XGR/Snowie/etc...?
> I'm not sure what is genetically means for you, but it is a neural
> net based. Different to Gnu it's only trained by self-play. XG/Snowie
> training details are not known
asked instead if it's "philosophically" different? In other words, do you
take the same approach to making checker/cube decisions, etc.?
>> A-c) Will you accept a non-monetary human challenge to your
>> bot and help conduct an open experiment?
> if the setting and the amount of time is reasonable: why not.
some interesting experiments without resorting to this.
>> A-d) Will you accept a monetary human challenge to your bot
>> based on prorated even odds, i.e. the proportionate winning
>> chances of the two sides according to your bot's rating of
>> their respective checker and cube skills?
> don't got that.
let's postpone this for now.
> Be assured that if I invested a fraction of the time that I invested in BGBlitz in normal free lancer work I would have made a not so small fortune....
more open to alternative ideas and put some time/effort into experimenting
with them.
If you have followed some of my arguments here over the years, you probably
know my lack of esteem for of the so-called "cube skill". In a few words,
it is based on arguments like: "Everything else being equal between two bots
and/or bot-like humans, the side who gains the initial advantage will never
lose it!" (after 4 billion games;)...
Therefore, I argue that the "cube window" totally bogus. If we are going to
insist on using the cube, we need to handle it differently and accept that
it may not even require all that great of a skill.
Again, I don't want to repeat what I had posted about my experimenting with
doubling within the first 2-3 rolls against gnubg depending on the dice and
I don't know if anyone else experimented with it with a genuinely open mind
but doing it with two slightly different versions of the same bot may be
more valuable/convincing to bot worshippers.
How hard would it be for you to allow the user to adjust the "cubing window"
to whatever the value he wishes (instead of just cautious/normal/aggressive)?
It looks like the functionality is there already and it should be a matter
of setting the variable to any value instead of preselected three.
Apparently, your bot also already has the feature to play against another
bot(?) If so, I would be very much interested in having the normal bgblitz
bot play against another bgblitz bot identical except in its cube handling
like I would... :)
Would you be willing to help with this?
MK
michae...@gmail.com
Nov 11, 2014, 11:37:54 AM11/11/14
to
Thank God BG is a fast game in real life or online play and perhaps this is the only reason the computers did not manage to kill it yet like they killed chess.
Reading about Murat's questions on how BGBlitz handles the cube I also have the following question:
From the old times of Snowie and those nice graphs of how equity/MWC changed during the game: Many Snowie users thought that cube decisions made while MWC had a history of constant rising had much more chance than the theoretical.
I am wondering whether this hypothesis has been tested statistically by Frank as developer of BGblitz or any other developer.
Reading about Murat's questions on how BGBlitz handles the cube I also have the following question:
From the old times of Snowie and those nice graphs of how equity/MWC changed during the game: Many Snowie users thought that cube decisions made while MWC had a history of constant rising had much more chance than the theoretical.
I am wondering whether this hypothesis has been tested statistically by Frank as developer of BGblitz or any other developer.
mu...@compuplus.net
Nov 12, 2014, 7:25:04 AM11/12/14
to
On Tuesday, November 4, 2014 3:51:38 PM UTC-7, Michael Petch wrote:
> You were probably too busy posting about Jellyfish to notice
> that BgBlitz came out in the mid 90s, maybe '95 or '96. GNU
> Backgammon isn't even that old since it didn't show up until
> about '99.
I have always lauded and supported gnubg against the other bots,
> You were probably too busy posting about Jellyfish to notice
> that BgBlitz came out in the mid 90s, maybe '95 or '96. GNU
> Backgammon isn't even that old since it didn't show up until
> about '99.
despite how I felt about some of the false-scientist cocksuckers
who are for some reason unwilling to take the next steps for the
sake of the game but I also understand their defensiveness...
After concocting bullshit for years and years, acknowledging that
they were wrong from the step one would surely be devastating. :(
The good thing is, they still have time to go jump off a cliff. :)
What's most interesting here, though, along with greedygammon's
post, is that the pieces of a puzzle may be coming together for
me after so many years of unsuspecting wondering...
After the motherfucking scumbag jellyshit was caught cheating with
his pants down, I never understood and kept asking how he could
fall of the face of the planet just like that...!??
Snowjob was obviously superior to it but surely not so much to
wipe it out of competition overnight. Personally I have always
believed that snowjob was built on reverse engineered jellyshit
and have posted several unlikely yet common bugs shared between
the two.
And Petch is right that bgshitz had escaped my attention, maybe
because it just wasn't news enough, because it didn't pose any
considerable challenge to snowjob, let alone the much better and
stronger gnugb.
After reading what you guys posted, I reinstalled bgshitz for a
third time and lo and behold, it sure plays like the jellyshit
of yore... :)
It's so fucking obvious that I can't believe I missed it when I
played it the very first time. Oh well, I guess we are all apt
to fall asleep at the wheel... :(
No wonder the piece of bgshitz cost only $23. Cheaper than the
original jellyshit... :))
MK
michae...@gmail.com
Nov 12, 2014, 8:53:27 AM11/12/14
to
Tim Chow
Nov 12, 2014, 8:27:25 PM11/12/14
to
On Tuesday, November 11, 2014 11:37:54 AM UTC-5, michae...@gmail.com wrote:
> From the old times of Snowie and those nice graphs of how equity/MWC changed
> during the game: Many Snowie users thought that cube decisions made while MWC
> had a history of constant rising had much more chance than the theoretical.
I don't understand your description here. What is the phenomenon? If, during a game or a match, the MWC/equity rises constantly, then...?
> From the old times of Snowie and those nice graphs of how equity/MWC changed
> during the game: Many Snowie users thought that cube decisions made while MWC
> had a history of constant rising had much more chance than the theoretical.
---
Tim Chow
mu...@compuplus.net
Nov 13, 2014, 10:07:10 AM11/13/14
to
On Thursday, November 6, 2014 2:39:13 PM UTC-7, Tim Chow wrote:
> For example, say that BGBlitz looks at Murat's error rate and
> decides that BGBlitz would be a 3-to-1 favorite against him in
> an 11-point match; would you be willing to give 3-to-1 odds
> against Murat?
>
> Murat doesn't believe in these estimates and thinks that
> someone else in the world besides himself would be impressed
> if he were able to beat those estimated odds.
Based on your example, if I ended up beating the odds by only a
> For example, say that BGBlitz looks at Murat's error rate and
> decides that BGBlitz would be a 3-to-1 favorite against him in
> an 11-point match; would you be willing to give 3-to-1 odds
> against Murat?
>
> Murat doesn't believe in these estimates and thinks that
> someone else in the world besides himself would be impressed
> if he were able to beat those estimated odds.
few points, like 34-66 or 35-65 instead of 33-66, I myself would
not be impressed either.
However, if I were able to beat the odds by let's say 10% points,
i.e. 43-57, I sure would be impressed like hell.
If I started beating the odds by 1%, then 2%, 3%, 5%, 10%, etc.
I assume that you would start becoming impressed at some point??
So let me ask you a simple question: "What would that point be?"
I have another related question but I don't want to strain your
feeble mind all at once, so I'll save that one for later... ;)
MK
Bradley K. Sherman
Nov 13, 2014, 10:36:23 AM11/13/14
to
<mu...@compuplus.net> wrote:
>On Thursday, November 6, 2014 2:39:13 PM UTC-7, Tim Chow wrote:
>
>> For example, say that BGBlitz looks at Murat's error rate and
>> decides that BGBlitz would be a 3-to-1 favorite against him in
>> an 11-point match; would you be willing to give 3-to-1 odds
>> against Murat?
>>
>> Murat doesn't believe in these estimates and thinks that
>> someone else in the world besides himself would be impressed
>> if he were able to beat those estimated odds.
>
>Based on your example, if I ended up beating the odds by only a
>few points, like 34-66 or 35-65 instead of 33-66, I myself would
>not be impressed either.
>
>However, if I were able to beat the odds by let's say 10% points,
>i.e. 43-57, I sure would be impressed like hell.
>
>If I started beating the odds by 1%, then 2%, 3%, 5%, 10%, etc.
>I assume that you would start becoming impressed at some point??
>
>So let me ask you a simple question: "What would that point be?"
Whatever that point is, you would never reach it. Long before
>On Thursday, November 6, 2014 2:39:13 PM UTC-7, Tim Chow wrote:
>
>> For example, say that BGBlitz looks at Murat's error rate and
>> decides that BGBlitz would be a 3-to-1 favorite against him in
>> an 11-point match; would you be willing to give 3-to-1 odds
>> against Murat?
>>
>> Murat doesn't believe in these estimates and thinks that
>> someone else in the world besides himself would be impressed
>> if he were able to beat those estimated odds.
>
>Based on your example, if I ended up beating the odds by only a
>few points, like 34-66 or 35-65 instead of 33-66, I myself would
>not be impressed either.
>
>However, if I were able to beat the odds by let's say 10% points,
>i.e. 43-57, I sure would be impressed like hell.
>
>If I started beating the odds by 1%, then 2%, 3%, 5%, 10%, etc.
>I assume that you would start becoming impressed at some point??
>
>So let me ask you a simple question: "What would that point be?"
you got there it would be clear that BGBlitz needed improvement
in its algorithm for estimating opponent's strength, or that
you had "thrown" the test matches.
If you could merely win 45 out of 100 11-point matches
against gnubg, we would be much more impressed, much
more quickly.
--bks
mu...@compuplus.net
Nov 13, 2014, 10:43:00 AM11/13/14
to
On Wednesday, November 12, 2014 6:53:27 AM UTC-7, michae...@gmail.com wrote:
> Do you think BG software is big business?
I have no idea about the actual market size but it could
> Do you think BG software is big business?
be big I suppose. If one could sell 1000 copies at $300,
I would consider that a pretty big chunk of money.
In general, software is very lucrative because you don't
really manufacture anything. Nowadays you don't even put
it on a disk. They download it and pay online. You email
a license key. Almost free money. :) Especially if you
stole part of the code from somebody else... ;)
In BG, software is as a gambling tool and as a way of
getting people to use a certain server to gable online,
etc.
I pity gamblers, as they spend their lives away trying
to make a few bucks which they could have made by mowing
lawns or delivering paper for $5/hour.
It may not be much different for bot peddlers either, as
Frank claimed he could made a fortune if he had spent as
much time doing something else.
What pisses me off and sends me swearing is that they all
pretend to be benevolent while trying to make a few
miserable dollars through deceptive means, just like the
gamblers who manipulate the dice, etc.
The sad thing is the fate of BG, especially the future of
the bots are in the hands of this small "incestuous circle"
of mentally ill gamblers... :(
I have no respect for cheaters. Piss on them!
MK
Michael Petch
Nov 13, 2014, 10:52:23 AM11/13/14
to
On 2014-11-13 8:42 AM, mu...@compuplus.net wrote:
> What pisses me off and sends me swearing is that they all
> pretend to be benevolent while trying to make a few
> miserable dollars through deceptive means, just like the
> gamblers who manipulate the dice, etc.
I've never received payment for my work on GNUbg, I can say money is not
> What pisses me off and sends me swearing is that they all
> pretend to be benevolent while trying to make a few
> miserable dollars through deceptive means, just like the
> gamblers who manipulate the dice, etc.
a motivating factor for me to work on the project.
michae...@gmail.com
Nov 13, 2014, 10:59:48 AM11/13/14
to
..... and reaches say to the point that a STATIC analysis by BG software says your winning chances are say 70% then the hypothesis is (i repeat hypothesis) that your chances are not really 70% but maybe 75 or 80%. It takes into account the DYNAMIC development of a position.That dynamic development could be related e.g to the playing strength of each player.
I hope it makes sense.
michae...@gmail.com
Nov 13, 2014, 11:39:47 AM11/13/14
to
I live in a European Country. Just recently we had the online National BG championship. I estimated the % of participants relative to the population. Less than one person per 200K. I know most of the participants only a few ever bought a BG software. Lets take the whole Europe+ the US population. What is it 500-600M? How many are the serious players around 3000? It is questionable if more than 20% of those ever bought a BG software.
I urge Frank of BG Blitz to tell us (not how many copies he sold) but how many downloads he had. Then you make your own estimates how much money he made.
It's not big business my friend trust me, I am in business myself.
I quite agree with you that BG software has been used by gamblers to cheat. I share your feelings Vs gamblers, gambling is totally against my ethics too.
But in Europe today you cannot even access gambling sites. There has been a law, and I think there is a similar law today in the US. So to which cheaters/gamblers are those BG software developers going to sell their products if not to those in Europe and the US?? To those in Africa?? :-)
I 've always played backgammon for the excitement of the game. However to keep my brain sharp I preferred chess. I am so disappointed that chess has been killed by computers.
Imo you should not get so frustrated against BG software.They have not killed the game, on the contrary they helped advance it.
Cheers man.
michae...@gmail.com
Nov 13, 2014, 12:01:46 PM11/13/14
to
On Thursday, November 13, 2014 5:52:23 PM UTC+2, Michael Petch wrote:
>
> I've never received payment for my work on GNUbg, I can say money is not
> a motivating factor for me to work on the project.
Speaking of downloads (re my previous post) could you please inform us how many downloads the latest version of GNUbg had? I need them to make an estimate.
>
> I've never received payment for my work on GNUbg, I can say money is not
> a motivating factor for me to work on the project.
Thanks.
Michael Petch
Nov 13, 2014, 2:20:33 PM11/13/14
to
On 2014-11-13 10:01 AM, michae...@gmail.com wrote:
> Speaking of downloads (re my previous post) could you please inform us how many downloads the latest version of GNUbg had? I need them to make an estimate.
> Thanks.
GNUbg doesn't follow traditional patterns that one sees with commercial
> Speaking of downloads (re my previous post) could you please inform us how many downloads the latest version of GNUbg had? I need them to make an estimate.
> Thanks.
software downloads so is much harder to track. GNUbg is an upstream
maintainer that generally provides two main things when it comes to
product distribution: Source code and Official binaries. We provide
official binaries for OS/X and Windows only. The bulk of users though
are on Linux distributions that take our source code, build binaries and
distribute the source code and the binaries from their own package
repositories. I have a hunch that our single largest base of users is
probably on Debian based systems, most especially Ubuntu.
Please note that 1.04.000 was only officially released (2 www.gnubg.org)
for 2 platforms - OS/X Yosemite and MS Windows. Older versions of OS/X
still use 1.03.00x as their latest versions. I don't generally even
track downloads or hits on our site however here is some raw data from
October and November (1.04.000 was released on Oct 21st).
Top 10 October downloads:
# Hits KBytes URL
1 98 3.65% 3169555 20.38%
/media/windows/gnubg-release-1_03_000-20140804-setup.exe
2 99 3.68% 1175113 7.55% /media/windows/gnubg-MAIN-20121023-setup.exe
3 17 0.63% 496357 3.19% /media/macos/gnubg-1_03_000-mac-x86_64.dmg
4 36 1.34% 441534 2.84% /media/windows/gnubg-1_04_000-20141021-setup.exe
5 12 0.45% 386926 2.49%
/media/windows/gnubg-release-1_02_000-20130728-setup.exe
6 5 0.19% 241811 1.55% /media/macos/gnubg-1_03_000-mac-ppc.dmg
7 8 0.30% 235282 1.51% /media/macos/gnubg-1_04_000-mac-1010-x86_64.dmg
8 8 0.30% 234419 1.51% /media/macos/gnubg-1_02_000-mac-x86_64.dmg
9 4 0.15% 195156 1.25% /media/macos/gnubg-1_02_000-mac-ppc.dmg
10 6 0.22% 170389 1.10% /media/macos/gnubg-1_03_000-mac-i386.dmg
Top 10 November downloads:
# Hits KBytes URL
1 47 4.92% 1599216 22.63% /media/windows/gnubg-1_04_000-20141021-setup.exe
2 10 1.05% 483621 6.84% /media/macos/gnubg-1_03_000-mac-ppc.dmg
3 19 1.99% 455526 6.45% /media/windows/gnubg-MAIN-20121023-setup.exe
4 8 0.84% 390312 5.52% /media/macos/gnubg-1_02_000-mac-ppc.dmg
5 13 1.36% 385813 5.46% /media/macos/gnubg-1_04_000-mac-1010-x86_64.dmg
6 8 0.84% 234419 3.32% /media/macos/gnubg-1_02_000-mac-x86_64.dmg
7 8 0.84% 233580 3.31% /media/macos/gnubg-1_03_000-mac-x86_64.dmg
8 7 0.73% 211051 2.99%
/media/windows/gnubg-release-1_02_000-20130728-setup.exe
9 5 0.52% 176110 2.49%
/media/windows/gnubg-release-1_03_000-20140804-setup.exe
10 6 0.63% 174914 2.48% /media/macos/gnubg-1_02_000-mac-i386.dmg
Ubuntu and other Debian distros happens to have a popularity contest for
packages. The service is opt-in so the numbers will only be a subset of
actual users. However for those people who share their information
Debian(Ubuntu/Kubuntu etc) one can find some raw information here:
http://popcon.ubuntu.com/by_vote
I'm not sure one can say much about the raw counts but it is interesting
to look at the comparison between gnuchess and gnubg.
More about what Popcon can be found here: http://popcon.debian.org/FAQ
It is rare (with the exception of the 1.00.000 release) for me to
actually advertise in public newsgroups when we put out new releases.
Our downstream maintainers (the distros) have automated bots that
generally pull any new source releases and do builds automatically for
their distributions. I generally only post to the mailing list and
update the website with little fan fare.
michae...@gmail.com
Nov 13, 2014, 3:37:43 PM11/13/14
to
Thank you very much M. Petch.
I am sure Murat is clever enough to make his own calculations considering the following factors.
a)Old users updating their computers
b)Trial users who uninstalled it eventually
c)Total downloads (over a period of 2-3 years) until next release .
d)What percentage of those who use it would pay to buy another software.
I am sure Murat is clever enough to make his own calculations considering the following factors.
a)Old users updating their computers
b)Trial users who uninstalled it eventually
c)Total downloads (over a period of 2-3 years) until next release .
d)What percentage of those who use it would pay to buy another software.
michae...@gmail.com
Nov 13, 2014, 3:54:46 PM11/13/14
to
XG--> 2009-2014 (5 years from first release and the author is still working on it)
BGBlitz -->1996-2014 (8 years!!)
Certainly BGBlitz is not for the money. It's just an intellectual work, the only thing that will survive after we die.
BGBlitz -->1996-2014 (8 years!!)
Certainly BGBlitz is not for the money. It's just an intellectual work, the only thing that will survive after we die.
Tim Chow
Nov 13, 2014, 4:50:46 PM11/13/14
to
On Thursday, November 13, 2014 10:07:10 AM UTC-5, mu...@compuplus.net wrote:
> If I started beating the odds by 1%, then 2%, 3%, 5%, 10%, etc.
> I assume that you would start becoming impressed at some point??
>
> So let me ask you a simple question: "What would that point be?"
Forget about computer-estimated odds. I would become impressed if you started consistently beating, say, XGR+ at 11-point matches or 5-point matches with greater than 50% probability. Even if you could do this against GNU grandmaster with your favorite random number generator, Mersenne Twister, that would be impressive.
> If I started beating the odds by 1%, then 2%, 3%, 5%, 10%, etc.
> I assume that you would start becoming impressed at some point??
>
> So let me ask you a simple question: "What would that point be?"
---
Tim Chow
mu...@compuplus.net
Nov 13, 2014, 11:06:54 PM11/13/14
to
On Thursday, November 13, 2014 8:52:23 AM UTC-7, Michael Petch wrote:
> I've never received payment for my work on GNUbg, I can say
> money is not a motivating factor for me to work on the project.
I didn't mean literally everybody and I'm sure there are many
> I've never received payment for my work on GNUbg, I can say
> money is not a motivating factor for me to work on the project.
more people like you contributing to gnubg. Even I try making
suggestion to improve it. I could try more but I don't think
I could do more because I don't agree with the claim that the
bot style checker play is the best way. Unfortunately I can't
suggest any better approach instead.
The same is even more true about the so-called "cube skill".
I know it's just elaborate bullshit and I can only contribute
to debunk it, hoping that others can come up with something
better.
What I don't understand is the unwillingness of people like
you (who claim to want nothing more than improve the bots)
to help me with it..??
Can you answer this at least for yourself? Why won't you in
fact even acknowledge the problem with the "cube skill" to
begin with??
MK
mu...@compuplus.net
Nov 13, 2014, 11:08:53 PM11/13/14
to
at all in these? I don't care the least bit.
MK
mu...@compuplus.net
Nov 13, 2014, 11:21:43 PM11/13/14
to
On Thursday, November 13, 2014 8:36:23 AM UTC-7, Bradley K. Sherman wrote:
>> If I started beating the odds by 1%, then 2%, 3%, 5%, 10%, etc.
>> I assume that you would start becoming impressed at some point??
>> So let me ask you a simple question: "What would that point be?"
> Whatever that point is, you would never reach it. Long before
> you got there it would be clear that BGBlitz needed improvement
> in its algorithm for estimating opponent's strength,
This doesn't concern just bgblitz but all bots in general and yes
>> If I started beating the odds by 1%, then 2%, 3%, 5%, 10%, etc.
>> I assume that you would start becoming impressed at some point??
>> So let me ask you a simple question: "What would that point be?"
> Whatever that point is, you would never reach it. Long before
> you got there it would be clear that BGBlitz needed improvement
> in its algorithm for estimating opponent's strength,
this is exactly what I am trying to show.
But since estimating opponent's strength is based on error rates,
etc. it would mean that the bots are playing wrongly (at various
degrees, depending on how badly I can "beat the odds")...!
So, are you suggesting that new ways of estimating the opponent's
strength will be fabricated to accommodate the wrong plays...?
> or that you had "thrown" the test matches.
after they are played.
> If you could merely win 45 out of 100 11-point matches
> against gnubg, we would be much more impressed, much
> more quickly.
discussing here. I'm not trying to prove that I'm a better bot!
I'm trying to prove that the "cube skill", "error rate", etc.
are bullshit... I hope you can grasp the difference??
MK
mu...@compuplus.net
Nov 13, 2014, 11:33:30 PM11/13/14
to
On Thursday, November 13, 2014 2:50:46 PM UTC-7, Tim Chow wrote:
> On Thursday, November 13, 2014 10:07:10 AM UTC-5, mu...@compuplus.net wrote:
>> If I started beating the odds by 1%, then 2%, 3%, 5%, 10%, etc.
>> I assume that you would start becoming impressed at some point??
>> So let me ask you a simple question: "What would that point be?"
> Forget about computer-estimated odds.
Why should we all of a sudden forget about the main issue at hand!?
> On Thursday, November 13, 2014 10:07:10 AM UTC-5, mu...@compuplus.net wrote:
>> If I started beating the odds by 1%, then 2%, 3%, 5%, 10%, etc.
>> I assume that you would start becoming impressed at some point??
>> So let me ask you a simple question: "What would that point be?"
> Forget about computer-estimated odds.
How long do you think you will run and hide from the reality? I got
you guys really bad, didn't I...? :)
> I would become impressed if you started consistently beating, say,
> XGR+ at 11-point matches or 5-point matches with greater than 50%
> probability.
about here. I'm not trying to be "another stick in the wall"...! :)
(and pun intended too:))
It really puzzles me that especially people who claim and posture as
scientists can be in such a deep denial... What do you have to lose
at stake?? The more you fight back and delay accepting my arguments,
the more ashamed you will be of yourselves later on... :((
MK
Bradley K. Sherman
Nov 14, 2014, 12:07:04 AM11/14/14
to
<mu...@compuplus.net> wrote:
> ...
You may have a point there. As far as I know the bots do
not modify their play against humans to allow for the fact
that they are way better than humans.
--bks
> ...
>But since estimating opponent's strength is based on error rates,
>etc. it would mean that the bots are playing wrongly (at various
>degrees, depending on how badly I can "beat the odds")...!
> ...
>etc. it would mean that the bots are playing wrongly (at various
>degrees, depending on how badly I can "beat the odds")...!
You may have a point there. As far as I know the bots do
not modify their play against humans to allow for the fact
that they are way better than humans.
--bks
Tim Chow
Nov 14, 2014, 1:17:36 PM11/14/14
to
On Thursday, November 13, 2014 11:33:30 PM UTC-5, mu...@compuplus.net wrote:
> On Thursday, November 13, 2014 2:50:46 PM UTC-7, Tim Chow wrote:
> > Forget about computer-estimated odds.
>
> Why should we all of a sudden forget about the main issue at hand!?
It may be the main issue for you, but it's not the main issue for me. The main issue for me is understanding how to play better, not how to calculate the odds of one player beating another.
> On Thursday, November 13, 2014 2:50:46 PM UTC-7, Tim Chow wrote:
> > Forget about computer-estimated odds.
>
> Why should we all of a sudden forget about the main issue at hand!?
> How long do you think you will run and hide from the reality? I got
> you guys really bad, didn't I...? :)
---
Tim Chow
bgbl...@googlemail.com
Nov 15, 2014, 10:08:07 AM11/15/14
to
Am Donnerstag, 13. November 2014 16:43:00 UTC+1 schrieb mu...@compuplus.net:
> I have no idea about the actual market size but it could
> be big I suppose. If one could sell 1000 copies at $300,
> I would consider that a pretty big chunk of money.
And how many years do you need for development? How many of the revenues will be lost through cracked copies? You don't need HW? That makes the numbers much less impressive...
> I have no idea about the actual market size but it could
> be big I suppose. If one could sell 1000 copies at $300,
> I would consider that a pretty big chunk of money.
You're much better of nowadays if you charge $1 but have millions of users.
> In general, software is very lucrative because you don't
> really manufacture anything. Nowadays you don't even put
> it on a disk. They download it and pay online. You email
> a license key. Almost free money. :)
> What pisses me off and sends me swearing is that they all
> pretend to be benevolent while trying to make a few
> miserable dollars through deceptive means, just like the
> gamblers who manipulate the dice, etc.
Just take 2 hours/day (which is to few) times 365 *19 * whatever_you_think_a_qualified_freelancer_earns gives a number that at least impresses me.
I don't complain. Most hobbies cost money, some a lot. BGBlitz earns me money. It's enough to pay me the hardware/sw/books and a bit more, but making a living from it? dream on...
BGBlitz is fun for me (even answering the mails is most often fun). It's an intellectual challenge, its a testbed to try out cool things, its a tool I carve for my usage and a vehicle to communicate with a lot of interesting peoples around the globe. But as a vehicle to get rich, forget it.
mu...@compuplus.net
Nov 20, 2014, 7:15:48 AM11/20/14
to
On Thursday, November 13, 2014 10:07:04 PM UTC-7, Bradley K. Sherman wrote:
> <mu...@compuplus.net> wrote:
>>But since estimating opponent's strength is based on error rates,
>>etc. it would mean that the bots are playing wrongly (at various
>>degrees, depending on how badly I can "beat the odds")...!
of weaker humans but they could " or that "the bots could modify
their play to take advantage of weaker humans but they do not"??
Hah hah hhaaaa... :) Masses of dumb asses... :))
MK
> <mu...@compuplus.net> wrote:
>>But since estimating opponent's strength is based on error rates,
>>etc. it would mean that the bots are playing wrongly (at various
>>degrees, depending on how badly I can "beat the odds")...!
> You may have a point there. As far as I know the bots do
> not modify their play against humans to allow for the fact
> that they are way better than humans.
Is it that "the bots do not modify their play to take advantage
> not modify their play against humans to allow for the fact
> that they are way better than humans.
of weaker humans but they could " or that "the bots could modify
their play to take advantage of weaker humans but they do not"??
Hah hah hhaaaa... :) Masses of dumb asses... :))
MK
mu...@compuplus.net
Nov 20, 2014, 7:25:45 AM11/20/14
to
On Friday, November 14, 2014 11:17:36 AM UTC-7, Tim Chow wrote:
> On Thursday, November 13, 2014 11:33:30 PM UTC-5, mu...@compuplus.net wrote:
>> On Thursday, November 13, 2014 2:50:46 PM UTC-7, Tim Chow wrote:
>>> Forget about computer-estimated odds.
>> Why should we all of a sudden forget about the main issue at hand!?
> It may be the main issue for you, but it's not the main issue for me.
> The main issue for me is understanding how to play better, not how to
> calculate the odds of one player beating another.
Okay, then, how do you go about "understanding how to play better"..?
> On Thursday, November 13, 2014 11:33:30 PM UTC-5, mu...@compuplus.net wrote:
>> On Thursday, November 13, 2014 2:50:46 PM UTC-7, Tim Chow wrote:
>>> Forget about computer-estimated odds.
>> Why should we all of a sudden forget about the main issue at hand!?
> It may be the main issue for you, but it's not the main issue for me.
> The main issue for me is understanding how to play better, not how to
> calculate the odds of one player beating another.
>> How long do you think you will run and hide from the reality? I got
>> you guys really bad, didn't I...? :)
> Reality of what? Reality that the Elo computations are bogus?
> I knew that before I knew who Fraidy-Rat was.
for a rock to crawl under and hide... :(
Scroll up and read what you wrote in your post that I responded to.
Let me help you. You wrote:
"He means, will BGBlitz estimate, based on error-rate
"calculations, what your odds are of beating it...
So, now, what about that "error rate" thingy..? Is it accurate..??
How accurate..???
Once again, you demonstrate that you are nothing more than one of
the mentally ill gambling bunch and a senile slimy asshole... :((
MK
mu...@compuplus.net
Nov 20, 2014, 7:43:31 AM11/20/14
to
On Saturday, November 15, 2014 8:08:07 AM UTC-7, bgbl...@googlemail.com wrote:
> And how many years do you need for development? How many of the
> revenues will be lost through cracked copies? You don't need HW?
> That makes the numbers much less impressive... You're much better
> of nowadays if you charge $1 but have millions of users.
Why don't you tell it to people like the guy who sells extreme bg?
> And how many years do you need for development? How many of the
> revenues will be lost through cracked copies? You don't need HW?
> That makes the numbers much less impressive... You're much better
> of nowadays if you charge $1 but have millions of users.
He sure spends a lot of time and effort for copy protection when
it's so easy to circumvent. Anybody paying a penny for it must be
really stupid or doing an act of charity to the poor peddler...
BTW, he is breaking the law by spying on people who try his bot,
by leaving trackers on their computers without their knowledge
and/or consent...!!
If anyone wants to know how to get rid of his leftover garbage,
I'd be glad to let them know openly right here... ;)
> BGBlitz is fun for me (even answering the mails is most often fun).
> It's an intellectual challenge, its a testbed to try out cool things,
> its a tool I carve for my usage and a vehicle to communicate with a
> lot of interesting peoples around the globe. But as a vehicle to get
> rich, forget it.
allowing full-range selection of doubling threshold/window... Even
if it won't make you a millionaire, I would pay for a copy of your
bot if it had anything out of the ordinary than other bots, such
as the free gnubg which is as good as or better than any paid ones!
MK
Bradley K. Sherman
Nov 20, 2014, 9:21:15 AM11/20/14
to
<mu...@compuplus.net> wrote:
> ...
The bots could be programmed to recognize that they're playing
a weak player, and they could adjust for that, but they do not
(AFAICT).
--bks
> ...
>Is it that "the bots do not modify their play to take advantage
>of weaker humans but they could " or that "the bots could modify
>their play to take advantage of weaker humans but they do not"??
> ...
>of weaker humans but they could " or that "the bots could modify
>their play to take advantage of weaker humans but they do not"??
The bots could be programmed to recognize that they're playing
a weak player, and they could adjust for that, but they do not
(AFAICT).
--bks
michae...@gmail.com
Nov 20, 2014, 10:44:15 AM11/20/14
to
On Thursday, November 20, 2014 2:43:31 PM UTC+2, mu...@compuplus.net wrote:
> BTW, he is breaking the law by spying on people who try his bot,
> by leaving trackers on their computers without their knowledge
> and/or consent...!!
>
> If anyone wants to know how to get rid of his leftover garbage,
> I'd be glad to let them know openly right here... ;)
Yes I am interested. Please provide the information.
> BTW, he is breaking the law by spying on people who try his bot,
> by leaving trackers on their computers without their knowledge
> and/or consent...!!
>
> If anyone wants to know how to get rid of his leftover garbage,
> I'd be glad to let them know openly right here... ;)
Tim Chow
Nov 23, 2014, 3:48:06 PM11/23/14
to
On Thursday, November 20, 2014 7:25:45 AM UTC-5, mu...@compuplus.net wrote:
> > Reality of what? Reality that the Elo computations are bogus?
> > I knew that before I knew who Fraidy-Rat was.
>
> Who said anything about "Elo"...? Just like I said, you are looking
> for a rock to crawl under and hide... :(
>
> Scroll up and read what you wrote in your post that I responded to.
> Let me help you. You wrote:
>
> "He means, will BGBlitz estimate, based on error-rate
> "calculations, what your odds are of beating it...
>
> So, now, what about that "error rate" thingy..? Is it accurate..??
> How accurate..???
Error-rate, Elo, who cares? Obviously, odds-calculations are all just hocus-pocus. It's totally bizarre that you care so much about it.
> > Reality of what? Reality that the Elo computations are bogus?
> > I knew that before I knew who Fraidy-Rat was.
>
> Who said anything about "Elo"...? Just like I said, you are looking
> for a rock to crawl under and hide... :(
>
> Scroll up and read what you wrote in your post that I responded to.
> Let me help you. You wrote:
>
> "He means, will BGBlitz estimate, based on error-rate
> "calculations, what your odds are of beating it...
>
> So, now, what about that "error rate" thingy..? Is it accurate..??
> How accurate..???
---
Tim Chow
mu...@compuplus.net
Nov 24, 2014, 2:56:07 PM11/24/14
to
On Sunday, November 23, 2014 1:48:06 PM UTC-7, Tim Chow wrote:
>> So, now, what about that "error rate" thingy..?
>> Is it accurate..?? How accurate..???
> Error-rate, Elo, who cares? Obviously, odds-calculations
> are all just hocus-pocus. It's totally bizarre that you
> care so much about it.
You have avoided all my questions by trying to reduce the
>> So, now, what about that "error rate" thingy..?
>> Is it accurate..?? How accurate..???
> Error-rate, Elo, who cares? Obviously, odds-calculations
> are all just hocus-pocus. It's totally bizarre that you
> care so much about it.
subject to "odds-calculations" and thus make it go away.
Surely you and your ilk must "care" about Error-rate, Elo,
etc. or else you wouldn't spend so much time talking about
them.
Surely you are not using them as random words without any
defined meanings, are you??
So, now I am asking what do you guys understand from each
other's using those terms?
Clarify for me. Are you saying that odds calculations using
"error rates" are just hocus-pocus even though the "error
rates" themselves are not? Or are you saying that odds
calculations are just hocus-pocus because the "error rates"
used are hocus-pocus themselves??
Sorry, to traumatize your mentally ill gambler's brain.. :(
MK
Tim Chow
Nov 24, 2014, 8:25:15 PM11/24/14
to
On Monday, November 24, 2014 2:56:07 PM UTC-5, mu...@compuplus.net wrote:
> Clarify for me. Are you saying that odds calculations using
> "error rates" are just hocus-pocus even though the "error
> rates" themselves are not? Or are you saying that odds
> calculations are just hocus-pocus because the "error rates"
> used are hocus-pocus themselves??
I wouldn't say that "error rates are hocus-pocus themselves" because error rates can be useful when interpreted cautiously and applied in the right way.
> Clarify for me. Are you saying that odds calculations using
> "error rates" are just hocus-pocus even though the "error
> rates" themselves are not? Or are you saying that odds
> calculations are just hocus-pocus because the "error rates"
> used are hocus-pocus themselves??
When the computer recommends Play A and flags Play B as an error, it just means that if the rest of the game or match were played out by computers on both sides, then Play A would lead to better results. This doesn't mean, of course, that if the rest of the game or match were played out with a human on one side or both sides, then Play A would lead to better results. So if your goal is better results (and what other goal would make sense?), always going with the computer play isn't necessarily the right strategy.
However, when the computer flags Play B as a big error, it's usually possible to discern why. In many cases, once the "why" is understood, one can see that Play B would be a mistake against most players, and not just a mistake when a bot plays a bot. One can also sometimes discern the circumstances under which Play B would actually be a good play.
So, the computer's notion of an error is correlated with, but not exactly aligned with, "true" errors (i.e., the wrong play when all circumstances are taken into account, including who is playing whom, psychological factors, etc.). Because it is correlated, it is a useful tool for improving one's understanding of the game, provided one does not trust it blindly but uses it as a guide. But because it is not exactly aligned, taking it too literally and using it to calculate odds is going to lead to unreliable results.
---
Tim Chow
0 new messages