What's a novelty?

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.
Post Reply
User avatar
Ozymandias
Posts: 1157
Joined: Sun Oct 25, 2009 12:30 am

What's a novelty?

Post by Ozymandias » Wed Mar 21, 2018 4:57 pm

I was going to reply to Dann's thread Gentlemen, start your engines: A 2 ply novelty with this question, but I thought it would derail the conversation and besides, the topic deserves its own.

As Nelson mentions in the quoted article:
Traditionally, the word ‘novelty’ has only applied to new positions obtained by titled players.
The problem starts when he proposes an updated definition...
which takes into account sources of games that have traditionally not constituted “chess theory”–engine games, games between club players, games played on Internet servers by anonymous nicks
For centuries, games were played and forgotten by history: casual friendly games, blitz games, club games dating back almost 250 years, even serious games between masters... lost in the mists of time. From the late 1400s to the early 1800s, only a selection of several hundred games have made it trough to us; basically, games published in books, featuring masters playing either among themselves or (just as a easily) the infamous NN player. In the 19th century, things started to look better for chess statisticians, and as early as the Chicago 1874 tournament, recording the games played, began to be imperative, at the organizers' request. Finally, in the 20th century, the extensive use of clocks, score-sheets, SAN and chess software, favoured the increase of games being recorded.

If we look at our DB, we'll see that among the millions of games available, only a small portion, fit the old definition. But it was popular for a reason; the same way that traditionally, chess games from the best players, were portrayed in books, low quality games would NOT be accepted as part of chess opening theory. This made sense, when the games of the masters weren't the ones being exclusively kept for posterity, a selection criteria had to be imposed.

The new millennium, however, would bring forth another computer development, which didn't exactly help the term "novelty": engines. When you have at home, a player entity of the world's champion caliber, what happens to that well defined distinction, between "masters" and "amateurs"? The very same thing than with "novelty", it becomes irrelevant. The former because both of them, "masters" and "amateurs", are now in the same category (humans) and the later because it's suddenly impossible to properly define.

Nowadays, the most anyone can say is, "this move is a novelty... in my DB".

Dann Corbit
Posts: 10267
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: What's a novelty?

Post by Dann Corbit » Wed Mar 21, 2018 4:59 pm

I think the idea they are after is that it should be strong players so that the moves are fairly trustworthy.

A couple 800 Elo players on FICS don't really establish any chess theory.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

User avatar
Ovyron
Posts: 3304
Joined: Tue Jul 03, 2007 2:30 am

Re: What's a novelty?

Post by Ovyron » Wed Mar 21, 2018 5:08 pm

Dann Corbit wrote:I think the idea they are after is that it should be strong players so that the moves are fairly trustworthy.
Is this clearly defined? Like, a 2092 elo player makes some move never seen before, but nobody cares, then a player with 2093 elo makes the same move, and it makes it to the news? (feel free to adjust these elos accordingly, they just need to be 1 elo apart.)

Since elo fluctuates, doing it like this could cause some frustration (like a player knows they could have made the news if they played this novelty last week, but now their rating is too low.)
Great spirits have always encountered violent opposition from mediocre minds.

User avatar
Ozymandias
Posts: 1157
Joined: Sun Oct 25, 2009 12:30 am

Re: What's a novelty?

Post by Ozymandias » Wed Mar 21, 2018 5:09 pm

I agree, Nelson's way to generous, when it comes to accepting a game in his DB, but my point is, even if you only accept master-level games (as I do)... how can you know that a certain position hasn't been played before?

I'm sure I have games Nelson doesn't, and vice versa, but even if we were to merge them (not happening), someone else could come along and point to an earlier occurrence. Ultimately, you can have in your DB, in-house played games just as deserving of trustworthiness, as any other publicly available.
Ovyron wrote:
Dann Corbit wrote:I think the idea they are after is that it should be strong players so that the moves are fairly trustworthy.
Is this clearly defined? Like, a 2092 elo player makes some move never seen before, but nobody cares, then a player with 2093 elo makes the same move, and it makes it to the news? (feel free to adjust these elos accordingly, they just need to be 1 elo apart.)

Since elo fluctuates, doing it like this could cause some frustration (like a player knows they could have made the news if they played this novelty last week, but now their rating is too low.)
That would be another valid discussion point.

FICGS
Posts: 14
Joined: Wed Mar 13, 2013 6:20 pm
Contact:

Re: What's a novelty?

Post by FICGS » Mon Mar 26, 2018 8:04 pm

"made the news", really ? :)

Anyway, when there are news, there is money, so there is no rule anymore IMHO...
Play chess online on the FICGS applications & website - Correspondence chess tournaments & championship

User avatar
Ozymandias
Posts: 1157
Joined: Sun Oct 25, 2009 12:30 am

Re: What's a novelty?

Post by Ozymandias » Sat Jun 30, 2018 9:59 am

Example from the Leela blog, the real "novelty" comes at move 9:

Colin-G
Posts: 154
Joined: Mon Oct 31, 2016 5:30 pm
Location: England

Re: What's a novelty?

Post by Colin-G » Sat Jun 30, 2018 11:04 am

For me it is a novelty if the move has not been played before in my engine-engine match database of over 30,000 games.
Scid finds the novelty for me.
e.g. 12...Rd8 was the novelty played by Fire 7.1 in this position against Phalanx XXV yesterday.

the full game is below

Just testing posting with fen and pgn tags. All seems ok.

yanquis1972
Posts: 1762
Joined: Tue Jun 02, 2009 10:14 pm

Re: What's a novelty?

Post by yanquis1972 » Sat Jun 30, 2018 1:39 pm

Definition of novel
1 : new and not resembling something formerly known or used
2 : original or striking especially in conception or style

the 2nd definition is the more useful one, the 1st i find interesting in terms of trivia.

if the primary intention of noting novelty's is a quick & dirty mark of advancing chess theory, ozymandias (& the status quo) has it right. if the primary intention is one of statistical curiosity/trivia, nelson does.

but i don't interpret the quote as nelson proposing a change to the definition as-is, rather offering an alternative that has its own worth w/r/t, at very least, trivia ("a position never before seen in the catalogued history of chess") with a happy side-effect of occasionally revealing potential sources of inspiration & serving to mark advances taking place outside of sanction, OTB chess.

in chess notation they could be distinguished as "TN" & "N" (?)

User avatar
Ozymandias
Posts: 1157
Joined: Sun Oct 25, 2009 12:30 am

Re: What's a novelty?

Post by Ozymandias » Sat Jun 30, 2018 3:30 pm

Colin-G wrote:
Sat Jun 30, 2018 11:04 am
For me it is a novelty if the move has not been played before in my engine-engine match database of over 30,000 games.
Illustrating my point about the definition being quite liquid nowadays.
yanquis1972 wrote:
Sat Jun 30, 2018 1:39 pm
Definition of novel
1 : new and not resembling something formerly known or used
2 : original or striking especially in conception or style

the 2nd definition is the more useful one, the 1st i find interesting in terms of trivia.

if the primary intention of noting novelty's is a quick & dirty mark of advancing chess theory, ozymandias (& the status quo) has it right. if the primary intention is one of statistical curiosity/trivia, nelson does.

but i don't interpret the quote as nelson proposing a change to the definition as-is, rather offering an alternative that has its own worth w/r/t, at very least, trivia ("a position never before seen in the catalogued history of chess") with a happy side-effect of occasionally revealing potential sources of inspiration & serving to mark advances taking place outside of sanction, OTB chess.

in chess notation they could be distinguished as "TN" & "N" (?)
We all agree that some filter must be stablished. The part where everyone disagrees, is where the limit should be placed. Nelson accepts games by players with an online rating below 2000 (although I should clarify that he doesn't accept every recorded game), others stick to GM games (or a small engine-engine match database). Who's right? Everyone and no one. The delimitation was always artificial, but for a long time, it was practical. Now is artificial to the point of nonsense and hardly practical.

In the example I provided above, it could very well be argued, that Ufim 8.02, playing on the equivalent to an AMD64X2 4200+ with a TC or 40/20, just doesn't have the quality required to claim the novelty. The fact that it lost (while Leela won) is further argument. But it's all a question of where you drew the line in the sand.

Post Reply