![[mx]](/mwf/v12/flags/mx.png "Mexico")
Hi, i'm new here. I know nothing about programming and little about computer chess tournaments. However, the facts are clear. Rybka won 4 WCCC in a row, unmatched. Plagiarism?...really? Why not improve your engine and compete with Rybka, instead of accusing Rajlich of technical violations which have nothing to do with how Rybka operates. Banned for life?..to me this means everyone at the ICGA panel fears Rybka and know their engines, hence themselves, to be inferior to Rajlich. Who cares about the ICGA?
Maybe a private match between Rybka and Houdini, sponsord and regulated by another institution, might return public interrest to chess.
As a fan, I really don't care about codes and all that, I simply like Rybka because it is the best out there and want to see it compete with other engines of similar caliber.
As Kasparov once stated, Rybka is the best out there.
To programmers, live with it and keep working.
Maybe a private match between Rybka and Houdini, sponsord and regulated by another institution, might return public interrest to chess.
As a fan, I really don't care about codes and all that, I simply like Rybka because it is the best out there and want to see it compete with other engines of similar caliber.
As Kasparov once stated, Rybka is the best out there.
To programmers, live with it and keep working.
puedes tener por seguro,que en igualdad de condiciones de juego (de hardware me refiero,y por supuesto,de controles de tiempo),Houdini literalmente vapulearia a rybka...
y ni siquiera quiero que me creas,sino que puedes mirarlo tu mismo:si tienes rybka o Houdini(las ultimas versiones Rybka 4.1 y Houdini 2.0c,y sino,SOLO para hacer el test y NO para jugar,no apruebo la pirateria,bajalos pirateados de internet),enfrentalos en tu ordenador,sea cual sea,en el control de tiempo que creas mas conveniente,y con un libro de aperturas neutro,y me dices como queda tu 'pequeño torneo'
p.s. te aconsejo 3 minutos + 2 segundos (o 5 minutos + 2 segundos de incremento por bando) para que puedas jugar al menos 100 partidas,y luego me dices...y te apuesto lo que quieras que Houdini consigue el 60% de los puntos totales,y con un poco de suerte,incluso mas
y ni siquiera quiero que me creas,sino que puedes mirarlo tu mismo:si tienes rybka o Houdini(las ultimas versiones Rybka 4.1 y Houdini 2.0c,y sino,SOLO para hacer el test y NO para jugar,no apruebo la pirateria,bajalos pirateados de internet),enfrentalos en tu ordenador,sea cual sea,en el control de tiempo que creas mas conveniente,y con un libro de aperturas neutro,y me dices como queda tu 'pequeño torneo'
p.s. te aconsejo 3 minutos + 2 segundos (o 5 minutos + 2 segundos de incremento por bando) para que puedas jugar al menos 100 partidas,y luego me dices...y te apuesto lo que quieras que Houdini consigue el 60% de los puntos totales,y con un poco de suerte,incluso mas
Sabes cual es el significado de vapulear?..no aplica para un 60%. De cualquier forma no lo voy a hacer.
no quise poner un 60,sino un 70...y si no lo vas a hacer(solo te llevaria unas 3 o 4 horas),antes de escribir posts sin un minimo de seriedad,piensatelo dos veces,porque yo si he hecho los 'pequeños torneos' esos,y el porcentaje de puntos totales de Houdini contra rybka se acerca al 70 %(puntos totales,NO partidas ganadas)
que facil es hablar por hablar,sin un minimo de pruebas que sustenten lo que dices,eh?no solo no tienes datos,sino que pudiendolos obtener tu mismo en apenas horas y convencerte de tu error,no quieres,y prefieres seguir en el...
como dijo el genial poeta,(y no solo aqui en ESPAÑA),de cada 10 cabezas,9 embiesten y 1 piensa...y tu no eres la que piensa
que facil es hablar por hablar,sin un minimo de pruebas que sustenten lo que dices,eh?no solo no tienes datos,sino que pudiendolos obtener tu mismo en apenas horas y convencerte de tu error,no quieres,y prefieres seguir en el...
como dijo el genial poeta,(y no solo aqui en ESPAÑA),de cada 10 cabezas,9 embiesten y 1 piensa...y tu no eres la que piensa
Cual es mi error, decir que Rybka es el mejor?..es mi opinion. Si le gana Houdini a Rybka, que bueno. De todos modos seguiré utilizando Rybka.
Buen día Barnard.
Buen día Barnard.
no randi,tu error no es decir que rybka es el mejor,tu error es que pudiendo comprobar que no lo es,te obceques en que si que es el mejor,cuando es obvio que Houdini le saca entre 60-70 puntos elo a rybka
buen dia para ti tambien randi,saludos
buen dia para ti tambien randi,saludos
![[cr]](/mwf/v12/flags/cr.png "Costa Rica")
No solo puede hacer las pruebas cuando quiera, también puede comprobarlo echando un vistazo a las diferentes páginas encargadas de probar los módulos, que contienen cientos y hasta miles de partidas jugadas por el mismo programa. En efecto, Houdini 2.0c es líder y la diferencia con Rybka 4.1 es de aproximadamente unos 60-70 puntos Elo.
Houdini será sin duda el módulo a batir por Rybka 5, aunque esperemos un tiempo y veamos que desempeño tendrá el Komodo 4 MP. La versión SP actual está teniendo resultados buenos en tiempos de controles más lentos. Quizás sea la versión multiprocesador o una siguiente la que logre arrebatarle el liderato a Houdini antes de la salida de Rybka 5.
Gaмßito.
Houdini será sin duda el módulo a batir por Rybka 5, aunque esperemos un tiempo y veamos que desempeño tendrá el Komodo 4 MP. La versión SP actual está teniendo resultados buenos en tiempos de controles más lentos. Quizás sea la versión multiprocesador o una siguiente la que logre arrebatarle el liderato a Houdini antes de la salida de Rybka 5.
Gaмßito.
Creo que el Komodo MP esta teniendo problemas para ser sacado a la venta;desde la versión 3 se espera la versión MP,pero siguen sin sacarla,y creo que la implementación no la están llevando todo lo bien que debieran,o al menos las pruebas que hacen enfrentando a la versión privada que ellos tengan del Komodo MP este teniendo resultados irregulares contra otros módulos que tienen una buena implementación del MP,como Houdini o rybka
Recuerdo un comentario que escribió Richard Lang(programador del Chess Genius),que dijo que su programa no escalaba nada bien cuando intento hacer la versión MP,y puede que simplemente el equipo de Komodo este teniendo problemas técnicos con la implementación de la versión MP
Recuerdo un comentario que escribió Richard Lang(programador del Chess Genius),que dijo que su programa no escalaba nada bien cuando intento hacer la versión MP,y puede que simplemente el equipo de Komodo este teniendo problemas técnicos con la implementación de la versión MP
Yo he usado el Houdini y el Rybka en mis análisis de juegos por correspondencia y concuerdo con mi paisano randi en que Rybka es mejor. Los resultados de juegos entre ellos no importan.
no importan los resultados de juegos entre ellos?entonces para que sirven las listas de fuerza como la 'IPON',para saber el tiempo que va a hacer?...otra cosa diferente es que tu te obceques en que rybka es mas fuerte que todos los engines,y que aunque juegue todos los torneos que sea,y que este por detras de ellos en todas las listas de fuerza,tu vas a seguir diciendo que rybka es mas fuerte que Houdini...
lo tuyo es como quien tiene fe en la religion;ya puedes decirle que dios no existe,y pedirle que te de pruebas de que existe,que seguira creyendo que existe,aunque nadie lo haya visto,ni una sola persona...
lo tuyo es como quien tiene fe en la religion;ya puedes decirle que dios no existe,y pedirle que te de pruebas de que existe,que seguira creyendo que existe,aunque nadie lo haya visto,ni una sola persona...
> es que tu te obceques en que rybka es mas fuerte que todos los engines,y que aunque juegue todos los torneos que sea,y que este por detras de ellos en todas las listas de fuerza,tu vas a seguir diciendo que rybka es mas fuerte que Houdini...
Sólo hablo de mi experiencia personal respecto a la utilidad que dan los engines para análisis de juegos por correspondencia. Y como están las cosas, engines "débiles" como Zappa Mexico II y Naum 4.2 son más útiles que Houdini.
>lo tuyo es como quien tiene fe en la religion;ya puedes decirle que dios no existe,y pedirle que te de pruebas de que existe,que seguira creyendo que existe,aunque nadie lo haya visto,ni una sola persona...
Si hablara todos los días con diós tendría la seguridad de que existe. Este es el caso con Houdini, ya he comprobado lo que digo personalmente y entonces no es cuestión de fé.
Uly,yo gano mas porcentaje de juegos usando ahora Houdini que cuando usaba el rybka(te hablo de hace años),como explicas eso?
y por otra parte,tampoco me enciego usando solo el Houdini...te enseñe un final de partida,en el que derrote a aquel jugador,el final de dama y tore(y peones),contra dama torre y peones,y use el HIARCS para analizarlo,que es bastante mas flojo que el Zappa o el Naum,pero que tiene muchos conocimientos de finales metidos dentro del engine...
y por otra parte,tampoco me enciego usando solo el Houdini...te enseñe un final de partida,en el que derrote a aquel jugador,el final de dama y tore(y peones),contra dama torre y peones,y use el HIARCS para analizarlo,que es bastante mas flojo que el Zappa o el Naum,pero que tiene muchos conocimientos de finales metidos dentro del engine...
> Uly,yo gano mas porcentaje de juegos usando ahora Houdini que cuando usaba el rybka(te hablo de hace años),como explicas eso?
Diferentes metodos de análisis llevan a diferentes resultados.
Ademas, ¿no que no te habías comprado el Rybka 4.1?
> te enseñe un final de partida,en el que derrote a aquel jugador,el final de dama y tore(y peones),contra dama torre y peones,y use el HIARCS para analizarlo
Las engines sólo son herramientas, esa victoria es más que nada credito tuyo que del Hiarcs.
si diferentes metodos de analisis llevan a diferentes resultados,puede que los que tu usas no sean los mas adecuados,y que por ello creas que rybka es mas fuerte que Houdini Uly,cierto?
cierto,no he comprado rybka 4,ni compre rybka 3,pero cuando salio la primera vez el rybka,si que me lo baje...y si,he dicho 'baje'...y no me convencio,asi que no lo compre...creo que no hace falta explicar de nuevo que siempre bajo las cosas antes de comprarlas,las uso un tiempo,y si me convencen,las compro,y si no me convencen,no las compro...
y como yo al ajedrez por correspondencia solo he vuelto desde hace relativamente poco tiempo(me lo deje,y lo retome en febrero del año pasado),ahi esta explicado lo del rybka que use,que no era la version 4,sino de hace años,como te habia dicho
bueno,lo de la victoria esa,si que te doy la razon;aunque use el HIARCS,me costo sangre sudor y lagrimas,todas las horas que estuve analizando esa partida,y ese final en concreto...
cierto,no he comprado rybka 4,ni compre rybka 3,pero cuando salio la primera vez el rybka,si que me lo baje...y si,he dicho 'baje'...y no me convencio,asi que no lo compre...creo que no hace falta explicar de nuevo que siempre bajo las cosas antes de comprarlas,las uso un tiempo,y si me convencen,las compro,y si no me convencen,no las compro...
y como yo al ajedrez por correspondencia solo he vuelto desde hace relativamente poco tiempo(me lo deje,y lo retome en febrero del año pasado),ahi esta explicado lo del rybka que use,que no era la version 4,sino de hace años,como te habia dicho
bueno,lo de la victoria esa,si que te doy la razon;aunque use el HIARCS,me costo sangre sudor y lagrimas,todas las horas que estuve analizando esa partida,y ese final en concreto...
> cierto,no he comprado rybka 4,ni compre rybka 3
Entonces no puedes saber de que estas hablando.
Es como si yo comentara en el mal desempeño de Houdini cuando solo lo use un rato.
otra vez tengo que volver a explicartelo Uly?en fin,lo hare...
cuando salio rybka 4,lo baje pirateado(no cabe esta nocion,pues no esta protegido),y lo estube probando...como no cumplio mis expectativas,simplemente lo borre,y deje de utilizarlo...asi que si que se de que estoy hablando,porque si que lo he utilizado,y como no me convencio,lo borre y no lo compre...
en el otro extremo esta Houdini,lo baje pirateado,y estuve utilizandolo para ver como era(mas que nada,en correspondencia,como el rybka),y como vi que si me gustaba,y me convencio,lo compre...
si lo que quieres es tener un tiempo de cuanto use el rybka,en tiempo,para opinar de el,aproximadamente 2 meses
cuando salio rybka 4,lo baje pirateado(no cabe esta nocion,pues no esta protegido),y lo estube probando...como no cumplio mis expectativas,simplemente lo borre,y deje de utilizarlo...asi que si que se de que estoy hablando,porque si que lo he utilizado,y como no me convencio,lo borre y no lo compre...
en el otro extremo esta Houdini,lo baje pirateado,y estuve utilizandolo para ver como era(mas que nada,en correspondencia,como el rybka),y como vi que si me gustaba,y me convencio,lo compre...
si lo que quieres es tener un tiempo de cuanto use el rybka,en tiempo,para opinar de el,aproximadamente 2 meses
¿Cuanto tiempo se te hace correcto usar el Rybka pirateado? 2 meses se me hace un tiempo muy arbitrario.
Quizas tus metodos no sean muy adecuados y por eso no aprendiste a sacarle provecho al Rybka.
Quizas tus metodos no sean muy adecuados y por eso no aprendiste a sacarle provecho al Rybka.
Uly,si contemplo la posibilidad que tus metodos sean incorrectos,tambien debo contemplar que los mios lo sean...por eso tambien miro analisis independientes,torneos independientes,etc,y en todos ellos,Houdini esta por delante de rybka
y lo de los 2 meses,fue por la siguiente razon:si fuera para ver su fuerza 'pura',con ponerlo a jugar un mini torneo,de unas 300 partidas contra otros engines,me hubiera bastado,y hubiera tardado,pongamos,unos 4 dias...pero para ajedrez por correspondencia,en menos de una semana,no puedes ver nada,asi que necesite jugar unas cuantas partidas de correspondencia 'rapidas';jugue el mismo numero con Houdini que con rybka,y con rybka me vi en mas apuros,pese a jugar igual,que con Houdini,asi que decidi que rybka no lo queria conmigo
y lo de los 2 meses,fue por la siguiente razon:si fuera para ver su fuerza 'pura',con ponerlo a jugar un mini torneo,de unas 300 partidas contra otros engines,me hubiera bastado,y hubiera tardado,pongamos,unos 4 dias...pero para ajedrez por correspondencia,en menos de una semana,no puedes ver nada,asi que necesite jugar unas cuantas partidas de correspondencia 'rapidas';jugue el mismo numero con Houdini que con rybka,y con rybka me vi en mas apuros,pese a jugar igual,que con Houdini,asi que decidi que rybka no lo queria conmigo
> por eso tambien miro analisis independientes,torneos independientes,etc,y en todos ellos,Houdini esta por delante de rybka
Afirmas eso, pero no puede ser cierto puesto que no tienes el Rybka.
> con ponerlo a jugar un mini torneo,de unas 300 partidas contra otros engines,me hubiera bastado
Análisis de juegos y juegos contra otros engines no tiene nada que ver, deberías saberlo mejor.
>Afirmas eso, pero no puede ser cierto puesto que no tienes el Rybka.
Uly,como no tengo el rybka,ya no es cierto lo que afirmo?pues si que es endeble tu argumento...
con el mismo razonamiento,tampoco tengo un doctorado en quimicas,y se que añadiendo un acido a una base,se obtiene una sal...y ahora refutame el argumento diciendome que no tengo un doctorado en quimicas,y que por eso no es cierto mi afirmacion...
>> con ponerlo a jugar un mini torneo,de unas 300 partidas contra otros engines,me hubiera bastado
>Análisis de juegos y juegos contra otros engines no tiene nada que ver, deberías saberlo mejor.
sacas mi frase de contexto;lo que digo es que use 2 meses el rybka para ver como se desenvolvia en correspondencia,que si hubiera querido ver la fuerza pura del engine,lo hubiera puesto a jugar engine vs engine en un mini torneo de 300 partidas,dado que en juego por correspondencia,la iteraccion del jugador humano con el engine,es,a mi entender,mas del 75 % del resultado del juego,y si no,mira el resultado de tu match con el centauro y lo veras
¿Cuantos juegos tuviste en esos 2 meses?
10 0 12 con el Houdini y otros 10 o 12 con el rybka...se que me vas a objetar que no son suficientes juegos para sacar conclusiones validas,pero para jugar mas juegos,tenia que tenerlo mas tiempo,y para ello hubiera tenido que comprarlo,y como en los 10 o 12 juegos que jugue con el rybka que vi en mas apuros que en los que jugue con el Houdini,y como en las listas de fuerza el Houdini estaba tambien por delante del rybka,juntando ambos parametros decidi que el rybka no lo queria porque ra mas flojo que el Houdini,y lo borre...ademas,si lees un post que escribio el Sedat Canbaz(creo que se escribe asi),sobre que hacer para tener un nivel alto de elo en el playches,mira que programa recomienda,y no es el rybka precisamente,sino el Houdini...
por supuesto,aqui en el foro de rybka,hay gente como tu que defiende a ultranza a rybka,pero si preguntas a jugadores de ajedrez por correspondencia fuera de este foro,te apuesto lo que quieras que la gente usa mas el Houdini que el rybka,y sino haz la prueba Uly
por supuesto,aqui en el foro de rybka,hay gente como tu que defiende a ultranza a rybka,pero si preguntas a jugadores de ajedrez por correspondencia fuera de este foro,te apuesto lo que quieras que la gente usa mas el Houdini que el rybka,y sino haz la prueba Uly
> ademas,si lees un post que escribio el Sedat Canbaz(creo que se escribe asi),sobre que hacer para tener un nivel alto de elo en el playches,mira que programa recomienda,y no es el rybka precisamente,sino el Houdini...
¿Por qué sigues revolviendo juegos con análisis? Esos argumentos no son válidos.
Lo que sí the puedo decir es que yo tengo mucha más experiencia con el Rybka que tú, lo he venido usando desde antes de que saliera a la venta, como testador Beta de Rybka 4. Y comparandolo con Houdini, puedo decir los siguientes puntos:
- Ambos escojen planes muy similares, así que no es muy efectivo usarlos juntos en análisis por la redundancia de ideas.
- Houdini tiene sus evaluaciones de posición más aplastadas (más cerca del cero) por lo que es más dificil encontrar posiciones atractivas a las que se quiera llegar, que con el Rybka.
- Por el mismo aplastamiento, Houdini tiende a dar la misma puntuación, o puntuación muy similar, a varias jugadas depues de análizarlas interactivamente, lo que no es nada útil pues su análisis se vuelve una lista de jugadas. El Rybka en cambió tiene más preferencia por determinados movimientos y un plan que seguir.
- Houdini es demaciado pesimista con respecto a posiciones que deberían tener más alto puntaje, he llegado a ver posiciones en donde negras gana que Houdini les da un bajo -0.60 cuando el Rybka les da un más correcto -1.50.
- Muy mala evaluación de algunas posiciónes en el medio juego, Rybka tiene este problema en los finales de juego que les da 2.00 a posiciones de empate, pero este problema es más serio en el medio juego en el que depender de Houdini puede ser muy peligroso.
Y tú, ¿qué problemas especificos puedes mencionar que Rybka tuvo en tus juegos? ¿qué tipo de cálidad piensas que pudieron tener 12 juegos en 2 meses? ¿Qué tipo de metodo de análisis usas en el que Rybka según no lo hizo bien? Mi tipo de análisis tiene unas 6 engines colaborando en un mismo juego, en el que todas tienen un papel importante, y de esta forma Rybka es mucho mejor jugador de equipo que Houdini.
vamos a ir al grano Uly,y dejemonos de teorizar...quieres un desafio conmigo?10 partidas,yo SOLO usare Houdini,y tu SOLO usaras rybka,sin ayudarnos de ningun otro engine,y vemos como queda el match?
Creo que tu desafío no tiene sentido, puesto que metodos basados en un sólo engine son inferiores a los que usan más de uno, no importando el resultado los juegos serian de pobre calidad.
Un desafío más correcto sería: Yo no puedo usar Houdini. Tú no puedes usar Rybka. Creo que mi fuerza de juego no se vería afectada en no usar a Houdini, si tú admites que tu fuerza tampoco sería peor por no usar a Rybka, entonces un desafío así no sería diferente que una serie de partidos normales.
Un desafío más correcto sería: Yo no puedo usar Houdini. Tú no puedes usar Rybka. Creo que mi fuerza de juego no se vería afectada en no usar a Houdini, si tú admites que tu fuerza tampoco sería peor por no usar a Rybka, entonces un desafío así no sería diferente que una serie de partidos normales.
Uly,una pregunta,y se sincero,los juegos que has visto mios aqui,contra ti y contra Paul,crees que han sido de pobre calidad?
¿Sólo usaste un engine en ellos?
si Uly,solo use el Houdini en ellos...y puedes comprobarlo analizando mi partidas con el Houdini,y veras que las jugadas coinciden con los analisis del Houdini(salvo novedades que yo introduje,como cuando jugue Dama 'b3' en la Chigorin contra Paul)
quise hacer la prueba de si usando un solo engine estaria en desventaja frente a gente que usaba varios,y a esto añade que tengo un ordenador que tiene 5 años
quise hacer la prueba de si usando un solo engine estaria en desventaja frente a gente que usaba varios,y a esto añade que tengo un ordenador que tiene 5 años
Bueno, entonces no tendrás problemas en aceptar un desafío en donde yo no puedo usar Houdini o ningúno de los derivados de Ippolit (ni Robbolito, ni Deep Saros, ni Fire, ni IvanHoe, etc.) y tú sólo usas Houdini.
Acepto el desafío Uly,10 partidas,tu no puedes usar Houdini ni ninguno de los derivados de Ippolit,y yo solo usare Houdini en mis análisis
Te parecen 10 partidas,así jugamos 5 cada uno con blancas y negras?y el ritmo el que tu y yo usábamos hasta ahora,jugar cuando tengamos la jugada
Te parecen 10 partidas,así jugamos 5 cada uno con blancas y negras?y el ritmo el que tu y yo usábamos hasta ahora,jugar cuando tengamos la jugada
![[us]](/mwf/v12/flags/us.png "United States")
Not bad for a newcomer!
You are not the first to suggest precisely this!
> Maybe a private match between Rybka and Houdini, sponsord and regulated by another institution, might return public interrest to chess.
You are not the first to suggest precisely this!
Hi Nelson,
I have one general comment on all this bruhaha that I am sure you can agree with:
We have computer chess history, and then we have computer chess history after the coming of the interwebs. In the interweb history, nothing like Vas and Rybka has managed to attract people to be actively involved in and follow computer chess events, as fans and participants. We have had freestyle stuff, the match with Zappa, the handicap matches with GMs, all the discussions in this forum (far ahead of any other computer chess community in terms of spread and quality), books by Jeroen, opening theory discussion and playchess and CC games, Convekta software etc. etc.
In light of that, it's absurd what is happening at the moment with the vendetta against Vas, as recently examplified with the vote against Rybka participation in whatever the name was (CCPT or USSR or something). Negative versus positive.
Also, I think you had a great comment on some guy's contribution recently, paraphrasing: All this talent put into such a silly cause???
I enjoyed the interview with Vas a lot, thanks for that, and also all your other amusing and insightful contributions. Always good for a good read :-)
I have one general comment on all this bruhaha that I am sure you can agree with:
We have computer chess history, and then we have computer chess history after the coming of the interwebs. In the interweb history, nothing like Vas and Rybka has managed to attract people to be actively involved in and follow computer chess events, as fans and participants. We have had freestyle stuff, the match with Zappa, the handicap matches with GMs, all the discussions in this forum (far ahead of any other computer chess community in terms of spread and quality), books by Jeroen, opening theory discussion and playchess and CC games, Convekta software etc. etc.
In light of that, it's absurd what is happening at the moment with the vendetta against Vas, as recently examplified with the vote against Rybka participation in whatever the name was (CCPT or USSR or something). Negative versus positive.
Also, I think you had a great comment on some guy's contribution recently, paraphrasing: All this talent put into such a silly cause???
I enjoyed the interview with Vas a lot, thanks for that, and also all your other amusing and insightful contributions. Always good for a good read :-)
Vas's overall contribution to the hobby is beyond reckoning, and that is precisely why he has become such a lightning rod. The paradoxical lesson is that punishment is meted out commensurate with the magnitude of one's good deeds. Vas is the Dreyfus of our times, sent to the hobby's equivalent of Devil's Island by these regal-minded, condescending louts. What I am still unable to get my head around is the intense willful malevolence of Vas's self-appointed persecutors. I have never in my life encountered such concentrated and sustained evil.
Just a bit too florid for my tastes.
But overall, pretty good.
I especially enjoyed the Dreyfus reference.
But overall, pretty good.
I especially enjoyed the Dreyfus reference.
Yes, I agree with your literary assessment. Over the top. I guess Soren is our Emile Zola.
You are much better than I am. You write with style, flair. My writing can best be described as "constipated".
For that I would recommend powerful mental laxatives. The traditional ones employed by writers are bourbon, gin, rum, vodka, tequila. After getting yourself nicely enameled you might start writing with an expansive style worthy of the great Russian prose stylists or perhaps Hunter S. Thompson.
Hunter S. was on a lot more "mental laxatives" than alcohol :-)
Excellent summary! After reading it, I have a much better appreciation for why other elements in the computer chess community felt a need to destroy this small island of success with charges that Vas copied his way to the top. When you compare the major interest in the Mexican match with the total lack of interest in recent ICGA events, it all becomes crystal clear.
> Maybe a private match between Rybka and Houdini, sponsord and regulated by another institution, might return public interrest to chess.
Call it the "Rybka vs Houdini Revenge Match of the 21th century" played in spite of ICGA sanctions.
Vasik,
I have been a customer since the very 1st commercial version. We once drew a hard fought game at the World Open. I remember our correspondence where you explained how Rybka used probability instead of pawn=1, so I am sure that you have made unique contributions to computer chess. I have rooted for Rybka since day 1. However, as a published scientist, where the 30 percent rule mentioned here is the norm, your statement below is the most absurd excuse since Bill Clinton said "it depends on what is is" during his defense of his lies about Monica Lewinsky. You have basically given away your own moral standard for plagiarism and I don't think I am unique to say this is ridiculous. I am not much of a computer programmer having peaked at Pascal, so I don't understand every detail of this debate, but the premise that only copy and paste is plagiarism or that retyping something yourself allows freedom to take others code is a very weak defense. I expected better. Even with the 30% standard for publication in science journals, the simple law is that you can not take someone else's intellectual property without proper reference and/or permission, not even 1%. Your catch phrase "original at the source code" seems a very flimsy fig leaf.
Dr. Brian McCarthy
I doubt that all of that code was typed by hand. If it was, then sure, it's "original at the source code level".
Lots of brilliant ideas!
Vas
I have been a customer since the very 1st commercial version. We once drew a hard fought game at the World Open. I remember our correspondence where you explained how Rybka used probability instead of pawn=1, so I am sure that you have made unique contributions to computer chess. I have rooted for Rybka since day 1. However, as a published scientist, where the 30 percent rule mentioned here is the norm, your statement below is the most absurd excuse since Bill Clinton said "it depends on what is is" during his defense of his lies about Monica Lewinsky. You have basically given away your own moral standard for plagiarism and I don't think I am unique to say this is ridiculous. I am not much of a computer programmer having peaked at Pascal, so I don't understand every detail of this debate, but the premise that only copy and paste is plagiarism or that retyping something yourself allows freedom to take others code is a very weak defense. I expected better. Even with the 30% standard for publication in science journals, the simple law is that you can not take someone else's intellectual property without proper reference and/or permission, not even 1%. Your catch phrase "original at the source code" seems a very flimsy fig leaf.
Dr. Brian McCarthy
> Regarding Strelka/IPPOLIT: as the author(s) seem to have typed their own source code (or code to generate this), how are they not "original" under your definition?
I doubt that all of that code was typed by hand. If it was, then sure, it's "original at the source code level".
> Do these Rybka versions [the ones that were cloned] have any additional creative content beyond the source code?
Lots of brilliant ideas!
Vas
![[pl]](/mwf/v12/flags/pl.png "Poland")
Hi Brian,
yeah, I remember you. You played that game really fast. I assume you don't often get into time trouble.
Source code originality isn't at all a fig leaf, it's the crux of the issue. David Levy as well as several other programmers accused me of copying source code. This is an important thing for programmers.
One thing I'll agree with you about is that perhaps it shouldn't be the main issue. In computer chess, ideas do have more value than code. Nevertheless, computer chess is a programming discipline, so code gets the central role.
Best regards,
Vas
yeah, I remember you. You played that game really fast. I assume you don't often get into time trouble.
Source code originality isn't at all a fig leaf, it's the crux of the issue. David Levy as well as several other programmers accused me of copying source code. This is an important thing for programmers.
One thing I'll agree with you about is that perhaps it shouldn't be the main issue. In computer chess, ideas do have more value than code. Nevertheless, computer chess is a programming discipline, so code gets the central role.
Best regards,
Vas
![[no]](/mwf/v12/flags/no.png "Norway")
> One thing I'll agree with you about is that perhaps it shouldn't be the main issue. In computer chess, ideas do have more value than code. Nevertheless, computer chess is a programming discipline, so code gets the central role.
>
This is the main thing for 99% of the fans at least. It should be about the chess. If it really was purely about the code there really shouldn't even be necessary play a single game of chess as the one with the cleanest code should just be given the trophy. But that would be kind of ridiculous, wouldn't it. What most people want to see is the strongest chess entity (software+hardware+opening book) play. Then what really matters is who has the groundbreaking ideas that makes a program stronger. This would also be a lot more beneficial commercially in my opinion. The vast majority wont buy a second rate engine for the engines sake (but perhaps for other features, but that is beside the point here). The ICGA seem to treat their tournaments as just some homegame, just like if me and some friends met for an evening of poker. Do they even need an organisation for that? What they seem to forget is that as soon as they publish something, e.g tournament results, then they can no longer claim they are running the tournament JUST for themselves. Their WCCC should be renamed to something like the world chess programming championship the way it is now, and perhaps that can be one discipline with very strict rules. And then have one open championship were derivatives are allowed to compete to see who truly has created the strongest chess entity. And so what if there are 5 Rybka or whatever derivatives that play in this category if they play differently? Of course engines have to follow copyright laws when it comes to commercial engines, and verbatim copying should not be allowed. I know for sure which discipline would be most interesting from a spectators point of view. In this scenario you will have both a chess programmer world champion AND a chess engine/entity world champion.
Well...these are just some thoughts on computer chess in general seen from a chess fans point of view.
The ICGA has accused Vas of deriving a program from Fruit 2.1. Not having access to the Rybka source code, which Vas says was "lost", we had to do the next best thing and reverse engineer the code. Our reports show some direct code copying (especially in the pre Rybka 1.0 programs). More importantly, it shows a huge change in Rybka to make it closely resemble Fruit. The reports clearly state code had to be rewritten from mailbox to bitboard, but a huge part of the evaluation of Fruit 2.1 was copied, converted then tuned. Just about every evaluation feature was copied. We could find no similar duplication in other open source programs. (Please note some panel members are aware of even more blatant copying in other programs, but these programs no longer seem to be active).
As for Strelka, when I first saw the code it was clear to be the basic evaluation was based on Fruit 2.1. Some programmers who probably were not familiar with Fruit did not see the similarity, but the panel agreed that Vas took Fruits evaluation and did not declare this on the entry forms.
So let me ask Vas a question. Do you think it is OK to read another programs source code then reconstruct it, step by step in another form (like C to Pascal or converting it from mailbox to bitboard)? No copy and paste, just retyping? You defend yourself saying everything was "typed" by you. But that is not defense at all. Monks hand copy manuscripts for centuries but that did not make the words original.
My answer would be no, that is stealing. If you took all the ideas or a big set of them without declaring it on the entry form, that is wrong.
I think reusing basic ideas is OK, like knights should be in the center. But copying a scheme adding rank and file specific ratios from another program is not. And doing this for all the pieces for both opening and ending eval is not right. Taking too much without permission is theft.
Vas says 75% of the code change around the time of Fruit release. Change from what to what? Out EXE analysis says it changed to closely resemble Fruit (and in the process gained something like 800 ELO).
As for Strelka, when I first saw the code it was clear to be the basic evaluation was based on Fruit 2.1. Some programmers who probably were not familiar with Fruit did not see the similarity, but the panel agreed that Vas took Fruits evaluation and did not declare this on the entry forms.
So let me ask Vas a question. Do you think it is OK to read another programs source code then reconstruct it, step by step in another form (like C to Pascal or converting it from mailbox to bitboard)? No copy and paste, just retyping? You defend yourself saying everything was "typed" by you. But that is not defense at all. Monks hand copy manuscripts for centuries but that did not make the words original.
My answer would be no, that is stealing. If you took all the ideas or a big set of them without declaring it on the entry form, that is wrong.
I think reusing basic ideas is OK, like knights should be in the center. But copying a scheme adding rank and file specific ratios from another program is not. And doing this for all the pieces for both opening and ending eval is not right. Taking too much without permission is theft.
Vas says 75% of the code change around the time of Fruit release. Change from what to what? Out EXE analysis says it changed to closely resemble Fruit (and in the process gained something like 800 ELO).
During the ICGA panel investigation of Rybka, on April 4th, 2011 the ICGA panel sent to Vasik Rajlich the following questions (via David Levy). We never received any answer to them. Now that Vas is answering some questions, perhaps he will answer these now?:
The relation of Rybka 1.0 Beta to earlier versions
In a CCC post of Dec 11 2005, you stated "The Rybka source code is original and pre-dates all of the Fruit releases."[1]
From this, one can infer that Rybka 1.0 Beta descended from Rybka 1.6.1 -- if so, can you answer the following?
1. Pre-Rybka 1.0 versions which were run in online tournaments seem to be roughly 600 elo weaker that Rybka 1.0 beta. It seems in less than a year, Rybka gained enormously in strength. Do you have intermediate versions showing the progression?
2. Are there particular reasons why you chose to use setjmp and incremental strtok (for UCI parsing) in Rybka 1.0 Beta? Similarly for the change in node counting and the disappearance of underpromotions -- for that matter, the disappearance of the UCI option searchmoves.
3. In particular, this represents an extremely significant departure from 1.6.1 (pre Rybka 1 beta) which had a structure much like that of Crafty, while Rybka 1.0 beta has one which possesses great similarity to Fruit 2.1. What would cause a complete rewrite of the basic program framework / organization, and over what sort of time frame was it accomplished?
4. Large examples of code copying from Crafty in Rybka 1.6.1 have been put into evidence. Given your statement that "Rybka is and always was completely original code, with the exception of various low-level snippets which are in the public domain,"[2] , do you contest that this is evidence is false?
Side question(s). Do you have any other old private Rybka versions that may be of use to this Panel, and if so, are you willing to submit them to its inspection? Furthermore, are you willing to allow the ICGA to make Rybka 1.6.1 public, so that this Panel can more fully meet the "publicity" requirement of any Rybka/Fruit findings (see 3.h.ii of The Charter), in particular to allow outside persons to verify said findings?
Fruit 2.1
1. If you say that no Fruit 2.1 code was copied with Rybka 1.0 Beta, how do you explain the evidence in Appendix A of the PDF of Watkins, in particular the second half of the iterative deepening code segment of page 17 (starting at the condition of EasyThreshold). Is the reconstruction on page 20 a fair representation of the ASM code? Is there a reason why the 6 variables that are used occur as with Fruit 2.1 (see bottom page 20)?
2. In a CCC post of Dec 16 2005, you stated "As far as I know, Rybka has a very original search and evaluation framework."[3] Can you expand on this?
3. The enumeration of Fruit 2.1 / Rybka 1.0 Beta evaluation features by both Wegner and Watkins shows a substantial, indeed mostly complete match. Are you able to give evidence of any other pair of engines "accepted by the computer chess community" which possess such an overlap of evaluation features? [NB: See evidential burden concerning the obligation of the defense to produce such evidence].
4. Do you have any kind of legal agreement that would allow you to release a work derived from Fruit under a non-GPL license? (proposed by Christophe Théron (http://www.talkchess.com/forum/viewtopic.php?t=23252)
Later Rybkas
1. It is claimed that the evaluation function of Rybka 2.3.2a has only a handful of dissimilarities when compared to the evaluation function in Rybka 1.0 Beta -- would you agree with this claim?
ICGA Tournaments
1. Are you able to provide us with the exact .exes that played in all ICGA events particularly those from 2006 and 2007? If not, is it a fair approximation to use Rybka 2.3.2a for the 2007 event?
Notions of clones and derivatives
1. You have declared Strelka first and then IPPOLIT to be clones/derivatives of various Rybka versions -- can you give some of indication of the standard of comparison you used to make these determinations? Claims that Strelka is Rybka
^ Rybka - How much Fruit is inside ?, CCC post of Dec 11, 2005
^ Is Rybka a Derivative of Fruit, Rybkaforum.net, Jul 14 2007
^ Unmasking the Secrets of Rybka and Fruit, CCC post, Dec 16 2005.
The relation of Rybka 1.0 Beta to earlier versions
In a CCC post of Dec 11 2005, you stated "The Rybka source code is original and pre-dates all of the Fruit releases."[1]
From this, one can infer that Rybka 1.0 Beta descended from Rybka 1.6.1 -- if so, can you answer the following?
1. Pre-Rybka 1.0 versions which were run in online tournaments seem to be roughly 600 elo weaker that Rybka 1.0 beta. It seems in less than a year, Rybka gained enormously in strength. Do you have intermediate versions showing the progression?
2. Are there particular reasons why you chose to use setjmp and incremental strtok (for UCI parsing) in Rybka 1.0 Beta? Similarly for the change in node counting and the disappearance of underpromotions -- for that matter, the disappearance of the UCI option searchmoves.
3. In particular, this represents an extremely significant departure from 1.6.1 (pre Rybka 1 beta) which had a structure much like that of Crafty, while Rybka 1.0 beta has one which possesses great similarity to Fruit 2.1. What would cause a complete rewrite of the basic program framework / organization, and over what sort of time frame was it accomplished?
4. Large examples of code copying from Crafty in Rybka 1.6.1 have been put into evidence. Given your statement that "Rybka is and always was completely original code, with the exception of various low-level snippets which are in the public domain,"[2] , do you contest that this is evidence is false?
Side question(s). Do you have any other old private Rybka versions that may be of use to this Panel, and if so, are you willing to submit them to its inspection? Furthermore, are you willing to allow the ICGA to make Rybka 1.6.1 public, so that this Panel can more fully meet the "publicity" requirement of any Rybka/Fruit findings (see 3.h.ii of The Charter), in particular to allow outside persons to verify said findings?
Fruit 2.1
1. If you say that no Fruit 2.1 code was copied with Rybka 1.0 Beta, how do you explain the evidence in Appendix A of the PDF of Watkins, in particular the second half of the iterative deepening code segment of page 17 (starting at the condition of EasyThreshold). Is the reconstruction on page 20 a fair representation of the ASM code? Is there a reason why the 6 variables that are used occur as with Fruit 2.1 (see bottom page 20)?
2. In a CCC post of Dec 16 2005, you stated "As far as I know, Rybka has a very original search and evaluation framework."[3] Can you expand on this?
3. The enumeration of Fruit 2.1 / Rybka 1.0 Beta evaluation features by both Wegner and Watkins shows a substantial, indeed mostly complete match. Are you able to give evidence of any other pair of engines "accepted by the computer chess community" which possess such an overlap of evaluation features? [NB: See evidential burden concerning the obligation of the defense to produce such evidence].
4. Do you have any kind of legal agreement that would allow you to release a work derived from Fruit under a non-GPL license? (proposed by Christophe Théron (http://www.talkchess.com/forum/viewtopic.php?t=23252)
Later Rybkas
1. It is claimed that the evaluation function of Rybka 2.3.2a has only a handful of dissimilarities when compared to the evaluation function in Rybka 1.0 Beta -- would you agree with this claim?
ICGA Tournaments
1. Are you able to provide us with the exact .exes that played in all ICGA events particularly those from 2006 and 2007? If not, is it a fair approximation to use Rybka 2.3.2a for the 2007 event?
Notions of clones and derivatives
1. You have declared Strelka first and then IPPOLIT to be clones/derivatives of various Rybka versions -- can you give some of indication of the standard of comparison you used to make these determinations? Claims that Strelka is Rybka
^ Rybka - How much Fruit is inside ?, CCC post of Dec 11, 2005
^ Is Rybka a Derivative of Fruit, Rybkaforum.net, Jul 14 2007
^ Unmasking the Secrets of Rybka and Fruit, CCC post, Dec 16 2005.
Finally some real questions :
1: No.
2: setjmp: I got this from Fruit. Before that Rybka unwound the search stack. I prefer setjmp because it gives terser search code (without requiring macros).
strtok UCI parsing: I don't remember.
3: There was no particular time frame; there were just ongoing changes.
4: I consider that (for now) to be a private matter.
Side question: No and no.
1. Please post the stuff here and I'll try to figure it out.
2. The Rybka search and evaluation code is not just original but also very unique. I have separate routines for different types of search and my evaluation code is written using a high level of abstraction. In addition I believe that my search and eval algorithms are relatively unique.
3. I haven't tried to quantify this similarity or compare it to that of other engines, but formal statistics such as ponderhit rates don't support this theory.
4. No.
Later Rybkas: I haven't tried to make a quantitative comparison, but this claim doesn't sound right.
ICGA Tournaments: No, and yes.
Notions of clones: My standard was code originality. I don't claim any particular rigor, it's just my personal view.
Vas
:1: No.
2: setjmp: I got this from Fruit. Before that Rybka unwound the search stack. I prefer setjmp because it gives terser search code (without requiring macros).
strtok UCI parsing: I don't remember.
3: There was no particular time frame; there were just ongoing changes.
4: I consider that (for now) to be a private matter.
Side question: No and no.
1. Please post the stuff here and I'll try to figure it out.
2. The Rybka search and evaluation code is not just original but also very unique. I have separate routines for different types of search and my evaluation code is written using a high level of abstraction. In addition I believe that my search and eval algorithms are relatively unique.
3. I haven't tried to quantify this similarity or compare it to that of other engines, but formal statistics such as ponderhit rates don't support this theory.
4. No.
Later Rybkas: I haven't tried to make a quantitative comparison, but this claim doesn't sound right.
ICGA Tournaments: No, and yes.
Notions of clones: My standard was code originality. I don't claim any particular rigor, it's just my personal view.
Vas
First, thanks for some answers even if the answer was often no answer. And thanks for confirming that Rybka 2.3.2a was accurate for what played in 2007. The panel spent a lot of time on that EXE.
Some clarification questions:
For this "1. Please post the stuff to here and I will try to figure it out", this site does not seem to allow PDF attachments, but here is a copy online:
http://icga.wikispaces.com/file/view/RYBKA_FRUIT_Mar11.pdf/291349523/RYBKA_FRUIT_Mar11.pdf
Note this was sent to you by David Levy last April, along with other documents. Please review it and answer the original questions about the match in iterative deeping code between Rybka and Fruit.
For "3. I haven't tried to quantify this similarity or compare it to that of other engines, but formal statistics such as ponderhit rates don't support this theory." do you have any evidence that ponder hit rates are useful in detecting similar programs? The evidence seems very lacking except for precise copies. I find merely changing a few evaluation values changes search results greatly without effecting strength much. There are just too many false matches and false indications of originality, hence the panel did not use them much.
For "4. No", can you take a look at this evaluation comparison: http://icga.wikispaces.com/file/view/RybkaEvalCompare.pdf which details the evaluation comparison between Rybka 1.0 beta and 2.3.2a." After reviewing this, can you confirm it is accurate, and then answer the question again? I am sorry but " but this claim doesn't sound right." is another non-answer, so lets look at the date.
For "Notions of clones: My standard was code originality. I don't claim any particular rigor, it's just my personal view.", I would like to better understand what you believe. Is this accurate "a. You feel is it not OK to directly copy code. b. You feel is is OK to translate code to another language without permission. c. You seem to feel it is OK to take algorithms from other programs. d. And that doing b and c are not "derivatives".
One last question. Why do you consider the pre Rybka 1.0 question to be a "private matter"? Did you have an agreement with Bob Hyatt to reuse his code/algorithms?
Some clarification questions:
For this "1. Please post the stuff to here and I will try to figure it out", this site does not seem to allow PDF attachments, but here is a copy online:
http://icga.wikispaces.com/file/view/RYBKA_FRUIT_Mar11.pdf/291349523/RYBKA_FRUIT_Mar11.pdf
Note this was sent to you by David Levy last April, along with other documents. Please review it and answer the original questions about the match in iterative deeping code between Rybka and Fruit.
For "3. I haven't tried to quantify this similarity or compare it to that of other engines, but formal statistics such as ponderhit rates don't support this theory." do you have any evidence that ponder hit rates are useful in detecting similar programs? The evidence seems very lacking except for precise copies. I find merely changing a few evaluation values changes search results greatly without effecting strength much. There are just too many false matches and false indications of originality, hence the panel did not use them much.
For "4. No", can you take a look at this evaluation comparison: http://icga.wikispaces.com/file/view/RybkaEvalCompare.pdf which details the evaluation comparison between Rybka 1.0 beta and 2.3.2a." After reviewing this, can you confirm it is accurate, and then answer the question again? I am sorry but " but this claim doesn't sound right." is another non-answer, so lets look at the date.
For "Notions of clones: My standard was code originality. I don't claim any particular rigor, it's just my personal view.", I would like to better understand what you believe. Is this accurate "a. You feel is it not OK to directly copy code. b. You feel is is OK to translate code to another language without permission. c. You seem to feel it is OK to take algorithms from other programs. d. And that doing b and c are not "derivatives".
One last question. Why do you consider the pre Rybka 1.0 question to be a "private matter"? Did you have an agreement with Bob Hyatt to reuse his code/algorithms?
Mark,
Please ask questions directly here in plain English. I want this discussion to be self-sufficient for future readers..
I do think that ponderhit rates are useful. For example, different versions of the same program do typically have high ponderhit rates, despite significant search and evaluation changes and significant Elo gains between those versions.
I also like the fact that ponderhit rates are objective. There is no room for subjective interpretation and for people seeing what they want to see.
I do agree that ponderhit rates are not perfect, and I wish we had something better which was still objective.
A is true. C is generally true (with some exceptions). B and D are generally false (with some exceptions).
I'll probably say more about those later. They are not relevant to the ICGA case.
Vas
> Note this was sent to you by David Levy last April, along with other documents. Please review it and answer the original questions about the match in iterative deeping code between Rybka and Fruit.
Please ask questions directly here in plain English. I want this discussion to be self-sufficient for future readers..
> do you have any evidence that ponder hit rates are useful in detecting similar programs? The evidence seems very lacking except for precise copies. I find merely changing a few evaluation values changes search results greatly without effecting strength much. There are just too many false matches and false indications of originality, hence the panel did not use them much.
I do think that ponderhit rates are useful. For example, different versions of the same program do typically have high ponderhit rates, despite significant search and evaluation changes and significant Elo gains between those versions.
I also like the fact that ponderhit rates are objective. There is no room for subjective interpretation and for people seeing what they want to see.
I do agree that ponderhit rates are not perfect, and I wish we had something better which was still objective.
> I would like to better understand what you believe. Is this accurate "a. You feel is it not OK to directly copy code. b. You feel is is OK to translate code to another language without permission. c. You seem to feel it is OK to take algorithms from other programs. d. And that doing b and c are not "derivatives".
A is true. C is generally true (with some exceptions). B and D are generally false (with some exceptions).
> One last question. Why do you consider the pre Rybka 1.0 question to be a "private matter"? Did you have an agreement with Bob Hyatt to reuse his code/algorithms?
I'll probably say more about those later. They are not relevant to the ICGA case.
Vas
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill