It is currently Sat Jul 20, 2019 1:18 am

All times are UTC - 8 hours [ DST ]




Post new topic Reply to topic  [ 6 posts ] 
Author Message
Offline
 Post subject: Assessing Human Error Against a Benchmark of Perfection
Post #1 Posted: Fri Jan 11, 2019 6:49 pm 
Honinbo

Posts: 8482
Liked others: 1468
Was liked: 1406
KGS: Kirby
Tygem: 커비라고해
Paper on human error in chess: https://arxiv.org/pdf/1606.04956.pdf

Kind of interesting. The authors use "tablebases" to get lookups for perfect play in chess positions having small numbers of pieces on the board. They analyze a large dataset of games, and attempt to find what factors correlate most strongly with blunders, mostly analyzing three categories:
* Player skill level
* Difficulty of position
* Time available to make moves

Interestingly, even more than skill level, the inherent difficulty of a position was seen to be the most strongly correlated with blunders. There were even "skill-anomalous" positions in which weaker players had fewer errors than those stronger than them. The empirical blunder rate was also shown to be higher in aggregate for players spending more time playing a move. Practically speaking, I suppose it means that in a complicated situation, both players are more prone to making blunders - maybe differences in skill become less noticeable? Not really sure.

A strong part of this analysis is that with the tablebases, the true optimal result is known, so blunders can be confidently categorized. On the other hand, since chess isn't solved for larger numbers of pieces on the board, maybe the analysis is limited to the end of the game.

Not really sure, but I found the article interesting.

_________________
it's be happy, not achieve happiness


This post by Kirby was liked by: Bill Spight
Top
 Profile  
 
Online
 Post subject: Re: Assessing Human Error Against a Benchmark of Perfection
Post #2 Posted: Fri Jan 11, 2019 9:45 pm 
Honinbo

Posts: 8592
Liked others: 2511
Was liked: 2962
Kirby wrote:
The empirical blunder rate was also shown to be higher in aggregate for players spending more time playing a move.


This is in line with the statement by Botvinnik, adopted by me in my own training, and applied by me here to some games analyzed by Leela 11, that plays you spend more time on are more likely to be mistakes than other plays. :)

_________________
There is one human race.
----------------------------------------------------

The Adkins Principle:

At some point, doesn't thinking have to go on?

— Winona Adkins

Top
 Profile  
 
Offline
 Post subject: Re: Assessing Human Error Against a Benchmark of Perfection
Post #3 Posted: Sat Jan 12, 2019 2:39 am 
Lives in gote

Posts: 602
Location: Indiana
Liked others: 114
Was liked: 174
Bill Spight wrote:
This is in line with the statement by Botvinnik, adopted by me in my own training, and applied by me here to some games analyzed by Leela 11, that plays you spend more time on are more likely to be mistakes than other plays. :)


This reminds me of a science fiction story from the 1950s in which someone notices that toilet paper seems to preferentially rip where there are no perforations. This "discovery" leads to a whole new field of materials science in which all structures are strengthened by removing increasing amounts of material. Think of the savings!

Top
 Profile  
 
Offline
 Post subject: Re: Assessing Human Error Against a Benchmark of Perfection
Post #4 Posted: Sat Jan 12, 2019 2:57 am 
Lives in gote
User avatar

Posts: 469
Liked others: 41
Was liked: 160
Read on page 7:

Quote:
we focus on the most commonly-occurring FICS time constraint — the large subset of games in which each player is allocated 3 minutes for the whole game. As a first object of study, let’s define the function g(t) to be the empirical blunder rate in positions where the player begins consid-
ering their move with t seconds left in the game. Figure 7 shows a plot of g(t);


Attachment:
time.png
time.png [ 8.63 KiB | Viewed 677 times ]

Top
 Profile  
 
Online
 Post subject: Re: Assessing Human Error Against a Benchmark of Perfection
Post #5 Posted: Sat Jan 12, 2019 4:20 am 
Honinbo

Posts: 8592
Liked others: 2511
Was liked: 2962
Aidoneus wrote:
This "discovery" leads to a whole new field of materials science in which all structures are strengthened by removing increasing amounts of material. Think of the savings!


Why, building bicycles out of hollow pipe saved both money and lives! :D

_________________
There is one human race.
----------------------------------------------------

The Adkins Principle:

At some point, doesn't thinking have to go on?

— Winona Adkins

Top
 Profile  
 
Offline
 Post subject: Re: Assessing Human Error Against a Benchmark of Perfection
Post #6 Posted: Sat Jan 12, 2019 5:08 am 
Lives in gote

Posts: 602
Location: Indiana
Liked others: 114
Was liked: 174
Bill Spight wrote:
Aidoneus wrote:
This "discovery" leads to a whole new field of materials science in which all structures are strengthened by removing increasing amounts of material. Think of the savings!


Why, building bicycles out of hollow pipe saved both money and lives! :D


We don't need no stinkin' spokes!
Image


This post by Aidoneus was liked by 4 people: Bill Spight, EdLee, jonsa, sybob
Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 6 posts ] 

All times are UTC - 8 hours [ DST ]


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group