"Fat fritz" stronger than Stockfish ?

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
chrisw
Posts: 2143
Joined: Tue Apr 03, 2012 2:28 pm

Re: "Fat fritz" stronger than Stockfish ?

Post by chrisw » Mon Oct 14, 2019 10:43 pm

Ovyron wrote:
Mon Oct 14, 2019 9:13 pm
Are you saying that it'd be fine to distribute the commercial opening chess book Goi 6.2.1 CTG (which costs $22) because the chess moves it contains and move rankings it contains can't be copyrighted?

I always thought distributing something like that would be piracy.
I understand where you are coming from because I was arguing with R de M exactly the same as you, well, sort of exactly. He made a very powerful case for there to be no copyright protection for neural net weights. Because no human hand.

If an “object“ has no copyright protection, it may be freely distributed. However A comes into physical possession of the object, A may copy and distribute it.

For your example of Goi. I have no idea. It would depend on whether there was human hand involved in its creation. If it’s a “book” created by applying an algorithm to a set of PGNs then I think the RdeM argument goes that unless there is some serious human creative selection process of the PGNs, then possibly not. Best ask a lawyer though.

User avatar
Ovyron
Posts: 2754
Joined: Tue Jul 03, 2007 2:30 am

Re: "Fat fritz" stronger than Stockfish ?

Post by Ovyron » Mon Oct 14, 2019 11:11 pm

Yeah, I'd say that if the neural net weights had some serious human creative selection process involved on their creation then they should be protected by copyright.

In the realm of opening chess books, the human creative selection process part can happen when she makes a selection of games to add to a book, after that, everything can become completely mechanized, and everyone that knows what games were selected can automate the process.

In the same way, anyone that knows how were the weights produced could do it at home, therefore, it should be able to be distributed because it's something that can be done without a human making decisions involved.

However, if nobody has any idea of the human creative selection process that happened to produce those weights (if any), then they'd be protected. Specifically, lc0 has many settings, the specific settings used to create the weights could have involved some human creative selection process, and then they'd be protected. It's problematic that we can't tell the difference between weights without it and weights with them, but it wouldn't be right to distribute them without even knowing how they were produced.
Great spirits have always encountered violent opposition from mediocre minds.

Uri
Posts: 423
Joined: Thu Dec 27, 2007 8:34 pm

Re: "Fat fritz" stronger than Stockfish ?

Post by Uri » Tue Oct 15, 2019 1:56 pm

The way I see it neural network chess engines is only for rich people.

Also the installation of neural network engines is difficult and the way I see it the technology is only in the beginning stage. In 50 years from now the technology could change completely and could become much more user friendly.

Also a fast Graphics card plus the Chessbase program costs tons of money if you buy it all together and most young chess players simply can't afford such a thing.

Jouni
Posts: 2011
Joined: Wed Mar 08, 2006 7:15 pm

Re: "Fat fritz" stronger than Stockfish ?

Post by Jouni » Tue Oct 15, 2019 3:48 pm

I fully agree with Uri! Unnecessary complicated to install. Can't hey finally release Lc0 version 1.0 and one exe-file!? No sense to buy fast GPU, when engine strength is +-20 to Stockfish.
Jouni

User avatar
Ovyron
Posts: 2754
Joined: Tue Jul 03, 2007 2:30 am

Re: "Fat fritz" stronger than Stockfish ?

Post by Ovyron » Sat Oct 19, 2019 12:09 am

Jouni wrote:
Tue Oct 15, 2019 3:48 pm
No sense to buy fast GPU, when engine strength is +-20 to Stockfish.
How much does a 10core CPU cost? Because if LC0 on that GPU plays +-20 to Stockfish, it might make sense to go for the GPU instead.
Great spirits have always encountered violent opposition from mediocre minds.

jorose
Posts: 268
Joined: Thu Jan 22, 2015 2:21 pm
Location: Zurich, Switzerland
Full name: Jonathan Rosenthal

Re: "Fat fritz" stronger than Stockfish ?

Post by jorose » Sat Oct 19, 2019 2:18 am

Ovyron wrote:
Sat Oct 19, 2019 12:09 am
Jouni wrote:
Tue Oct 15, 2019 3:48 pm
No sense to buy fast GPU, when engine strength is +-20 to Stockfish.
How much does a 10core CPU cost? Because if LC0 on that GPU plays +-20 to Stockfish, it might make sense to go for the GPU instead.
I only checked Newegg, which I have heard isn't close to the cheapest nowadays. You can get an 8core 3.3ghz processor for 65$ from Newegg. You can get a Ryzen processor with 12cores/24threads for 270$. There are Intel Xeons with 10 cores for around 120$, but they are for servers and have low clock speeds.

On the same site, an RTX2060 will set you back 330$ and you'll still need a CPU on top of that.
-Jonathan

User avatar
Ovyron
Posts: 2754
Joined: Tue Jul 03, 2007 2:30 am

Re: "Fat fritz" stronger than Stockfish ?

Post by Ovyron » Sat Oct 19, 2019 3:29 am

jorose wrote:
Sat Oct 19, 2019 2:18 am
You can get a Ryzen processor with 12cores/24threads for 270$.
I think this is where the lines cross, LC0 on a GTX 1650 would have a performance comparable to that 12core, and you can buy the GPU for $150, saving $120.
Great spirits have always encountered violent opposition from mediocre minds.

kramnik
Posts: 28
Joined: Fri Mar 29, 2019 8:21 pm
Full name: Massimiliano Goi
Contact:

Re: "Fat fritz" stronger than Stockfish ?

Post by kramnik » Tue Oct 29, 2019 4:21 pm

I guess no serious opening book is made just translating games into book format, there's always:

a) a human selection of games
b) a human sorting of that selection (as chess opening books compilers use the First Comes First Served algorythm, so there is a hierarchy bound to the sorting in the making process)
c) a human correction of bad/weak moves
d) a human testing of the product and an evenctual following correction (points c and d repeat many times)

For these reasons the most of the books are copyrighted, even though they are free, otherwise I could just steal a statue and not getting arrested just claiming that the marble which the statue was made is just material of that mountain... nonsense.

And usually the higher the price the more work was done, except maybe for the Apple products, which is crap.

I've not knowledge of neural networks, but they could be private just if the computers with which they were made were rented, or they belonged to a private network. And there could be a hand made selection of the nodes...

Since the basic of the modern language programming courses they say that private and public classes are the opposite, so what is private should not be released to the public as a base concept.
chrisw wrote:
Mon Oct 14, 2019 10:43 pm
Ovyron wrote:
Mon Oct 14, 2019 9:13 pm
Are you saying that it'd be fine to distribute the commercial opening chess book Goi 6.2.1 CTG (which costs $22) because the chess moves it contains and move rankings it contains can't be copyrighted?

I always thought distributing something like that would be piracy.
I understand where you are coming from because I was arguing with R de M exactly the same as you, well, sort of exactly. He made a very powerful case for there to be no copyright protection for neural net weights. Because no human hand.

If an “object“ has no copyright protection, it may be freely distributed. However A comes into physical possession of the object, A may copy and distribute it.

For your example of Goi. I have no idea. It would depend on whether there was human hand involved in its creation. If it’s a “book” created by applying an algorithm to a set of PGNs then I think the RdeM argument goes that unless there is some serious human creative selection process of the PGNs, then possibly not. Best ask a lawyer though.

User avatar
Ovyron
Posts: 2754
Joined: Tue Jul 03, 2007 2:30 am

Re: "Fat fritz" stronger than Stockfish ?

Post by Ovyron » Tue Oct 29, 2019 6:51 pm

kramnik wrote:
Tue Oct 29, 2019 4:21 pm
And usually the higher the price the more work was done
This is usually fallacious. Since they're children, people learn to associate price with value, so when they are adults they think the pricey thing has more quality. In reality, the pricey thing is most likely pricey because it belongs to a brand that had to pay for advertisements to be well known and such, and nothing else, so if you buy the cheap product that the store makes you'll be surprised that it's as good or better than the expensive one.

One brand of alcoholic beverages here in Mexico was in economic trouble, so what did they do to solve it? They increased the prices until they were the priciest around, and the sales soared because people hadn't seen a price so high before so they assumed it should be a product of unprecedented quality. It was the same product that wasn't selling because of low quality.

This is specially true for medicines, where there's a "active substance" that heals people, but a huge industry has been built around supply and demand where the most well known brands can inflate the price, while "similar" products work as well and cost a fraction, because the active substance has nothing to do with what they're charging.

The best things aren't often the priciest, after all, Stockfish is the best thing around and it is free.

kramnik
Posts: 28
Joined: Fri Mar 29, 2019 8:21 pm
Full name: Massimiliano Goi
Contact:

Re: "Fat fritz" stronger than Stockfish ?

Post by kramnik » Tue Oct 29, 2019 10:44 pm

You just can't put anything on the same plane and for the most of things you are wrong, people is not so ignorant to not get informed on what they are spending. funny that alcoholic aneddoct, thought :lol:

Post Reply