Rybka vs Fat Fritz controversies

[Madeleine Birchfield]

How similar is the current controversey over Fat Fritz 2, a commercial derivative of the free and open source Stockfish, to the controversey in 2011 over Rybka, a commercial derivative of the free and open source Fruit?

[towforce]

Very different. Rybka won 4 world championships: one of the rules for those championships was that entrants must declare if any of the code was taken from another program. Rybka made no such declaration. The championships' governing body then made a ruling that Rybka contained code from Fruit and Crafty, so Rybka was stripped of its titles.

In contrast, the facts about FF2 are barely in dispute at all. The problems are:

1. FF2 has a price tag of €100 for a product that is mostly open source code

2. Chessbase videos and web pages have exaggerated its strength and originality

[daniel71]

I think Rybka only used a few lines of code from Fruit and was totally rewritten by version 3. People accused Vasik of totally stealing Fruit but Rybka still today is the only commercial program able to use 2048 cores. Fritz 1 and 2 just completely copied all the code and changed only the nets for NNUE. I remember when Rybka first came out I purchased it directly from Vasik Rajlich and I was so impressed with how nice he was to me and his new engine, we exchanged several emails and I was a regular customer of Rybka. I used my new purchase to play chess on Play.com engine room and got blocked and threatened with a ban. I remember how nasty the Playchess.com moderator was to me like it was yesterday. I think IM Vasik Rajlich suffered greatly from jealousy from other people more than anything.

[Frank Quisinsky]

Open-Source software produced a "knock-on" effect if I am thinking on all the available clones with or without necessary sources (first causal view) ... for the glory of a faster development!

Rybka – Fruit / Stockfish or Lc0 – Fat Fritz … two different situations split the community but without comparability!
It is counterproductive if a commercial project split the community. Maybe the directors of Chessbase do not have the knowledge about the last developments in computer chess, like NNUE? Can be an explanation, each other explanation will give me after all the marketing information I found in www a negative memory.

Really complicated to set a symbol as answer: I do not like the product! Once I can do as user of chess-software is not to buy Fat Fritz II. I have it in my own hand. The problem with this view: I will give commercial companys "freedom to do whatever one wants".

Now, fun should be in foreground if I have time for my hobby!

If I am looking in the past, I am quite sure that such a product will be not the last one. Solidary from the community is not to await (to many different people with different own interest) and a way of solidary seems unconvertible / unthinkable.

We must life in “Open-Source” times with such products and should learn to see the positive site. Customers from the German company will get to know “Stockfish”. Unfortunately, under the name Fat Fritz.

To use the work from Fabien without to give the name of the programmer and his ideas, the sources is a complete other situation. After all I understand GPLv3 is not the topic for Fat Fritz II product. Open-Source programmers should be happy if a commerical company created a new product around it. But not without rhyme or reason!

[mar]

The championships' governing body then made a ruling that Rybka contained code from Fruit and Crafty, so Rybka was stripped of its titles.

it's not that simple.
Rybka 1.6.1 (not sure why Vas sent this to Olivier to participate in CW or OW, don't remember) was a verbatim copy of one of the older Craftys.
Rybka 1.0 (that was examined) had only 1 thing in common with Crafty - Bob's rotated bitboards IIRC.

this topic was discussed to death here so I see little point of going deeper.
I will only say this:
- a wrong version of Rybka was examined (not sure if Watkins examined Rybka2.x later), but Rybka 1.0 beta didn't participate in any ICGA tournaments IIRC
- comparing the FF2 situation to Rybka-Fruit is bonkers

last thing - if you're a dev and do hyperbullet testing or use singular extensions the way SF does (not the way Hsu did in Deep blue), then you owe Vas a beer

[noobpwnftw]

1) Rybka had clear improvements over any other chess engine at the time, while FF2 doesn't.
2) To this day, many people still acknowledge Vas's contributions despite all the controversies, while ASilver has never contributed one single game to Stockfish or Leela training.

As for the rest, I think the dust will eventually settle, and there will always be new fellows in the OSS community as well as new ASilver who rips them off.

[M ANSARI]

Totally different things. Vas is probably the most wronged person in computer chess and also is one of the star people that really pushed the limits to chess engine development at a time everyone thought stagnation was inevitable and chess engine strength was reaching its limits. I still remember those days and it really rekindled my interest in chess engines. He revolutionized how chess engines could be strengthened by using millions of fast game to tune a chess engine. This is still being used today by all the top engines including Lc0 and and SF. The unjust treatment Vas got probably turned him off chess engine development and as a result we lost a truly valuable contributor. In simple terms Vas came up with an engine that just kicked butt and many authors that were left in the dust got really pissed off!

With Fat Fritz 2 things are completely different. There is no custom chess engine that has been developed and it is quite clear on the package that the engine is the free SF that anyone can download. However there is a custom network which was developed. Supposedly this custom network was tuned and makes the engine play stronger moves than the original (according to the author) and might choose some moves that the default network might not and thus might give you an edge in analysis of chess positions. I really don't get the controversy regarding CB and Fat Fritz as everything is very clearly stated. Now some might think that CB should not charge anything for their software as SF is free and I can see their point. But CB also has a GUI and a Playchess subscription in the price as well as an extremely robust and powerful chess database tools ... they are notjust selling an engine. I would totally agree that the Playchess subscription is not worth it as Lichess or Chess.com are way better and Lichess (by far the best) also happens to be free !!! I would happily pay double for Lichess for what CB charges for Playchess and still think it is an awesome deal. But I have to admit that I very much like the CB GUI and database functionality. If you think the price of their package is not worth it, then just don't buy it! Anyone can charge anything for their work and it is up to the buyer to accept that or reject that.

[supersharp77]

How similar is the current controversey over Fat Fritz 2, a commercial derivative of the free and open source Stockfish, to the controversey in 2011 over Rybka, a commercial derivative of the free and open source Fruit?

Totally different things. Vas is probably the most wronged person in computer chess and also is one of the star people that really pushed the limits to chess engine development at a time everyone thought stagnation was inevitable and chess engine strength was reaching its limits. I still remember those days and it really rekindled my interest in chess engines. He revolutionized how chess engines could be strengthened by using millions of fast game to tune a chess engine. This is still being used today by all the top engines including Lc0 and and SF. The unjust treatment Vas got probably turned him off chess engine development and as a result we lost a truly valuable contributor.

.......Agreed 100%

...Post MB...
Very similar.....quite a few of these 'authors' 'borrow' 'use' or 'sample' code from other chess engines..IMHO this dispute with Rybka bleeds into the Chessbase dispute because when the chess engine "purists" (Most of which dominate & control THIS chess forum) Tried to Ban and Blacklist the author of the extremely commercially successful Rybka chess engine https://en.wikipedia.org/wiki/Vasik_Rajlich Vasik Rajlich...after years of wandering in the wilderness guess where he ended up finding a job? (Chessbase!!) Fritz 15 and Fritz 16 are seemingly carryovers of the famous/infamous Rybka Chess engines...That relationship ended with Fritz 17 (pressure perhaps?) Chessbase also caused quite a Bit of Controversy with the Release of their "Fat Fritz 1 Engine" (wholly based on LC0 (Leela) and they made similar fantastic claims on engine strength...
But we suspect they made quite a nice bundle of Cash selling their Chessbase Products around the holidays..Cut to 2021..The "release" of another Controversial Product "Fat Fritz 2" "Worlds Strongest Chess Engine" (err..Based almost wholly on Stockfish) Well Folks...."Here We Go Again"......"Round And Around"......

[towforce]

I didn't take much interest in the Rybka controversy when it happened, but I've had a quick look at the evidence (link), and it does look as though both V1 and later versions contained code from Fruit and Crafty. Apologies to Rybka supporters for any upset this causes - my intention, as ever, is an improved understanding of things.

One BIG mistake that the ICGA made is that the rules give the tournament director the right, but not the obligation, to ask contenders for source code: it was obvious to me when I entered the 2000 WCCC (as the Crafty operator) - a decade before this controversy - that the tournament director should be OBLIGED to collect source code from each competitor. When the source is modified between rounds, the changed source should also be handed over. There's no earthly reason to not do this!!!

I think that in this thread we have uncovered the source of Rybka's strength: his big innovation was quite likely to have been automated tuning.

[Gian-Carlo Pascutto]

or use singular extensions the way SF does (not the way Hsu did in Deep blue), then you owe Vas a beer

I don't think you need to send Vas beers just because you've read the Multi-ProbCut paper and made an obvious connection, but that's just me.