Latest Updates:
Page Index Toggle Pages: [1] 2 
Topic Tools
Hot Topic (More than 10 Replies) Sam Collins' bias towards White (Read 23319 times)
fjd
God Member
*****
Offline



Posts: 551
Location: Ottawa
Joined: 09/22/16
Re: Sam Collins' bias towards White
Reply #20 - 06/27/18 at 00:18:04
Post Tools
IsaVulpes wrote on 06/22/17 at 11:19:09:
I'm not a subscriber so I can't give any sort of "proof", just a random thought: Does White scoring well necessarily have to be some kind of bias, and not just that those are more crucial games to cover?

Say you have a well known equalish position.
If Black "finds" something new, it'll just be "In this position, which is already known to be a draw, Black now tries something new (why even?), and it's also a draw.", which isn't the most interesting thing to go over.
On the other hand, if White tries something new, and it works (!), it becomes crucial. Does this actually bring an advantage? Was it just a surprise? How should black respond to equalize? etc

Which basically means every White win with a new/rediscovered try has to be put under close scrutiny (to find a way for Black to equalize / figure out if it's actually the way forward for White), while Black wins will usually be born from outplaying White in equal positions (not too relevant for an opening page), and draws will often just be "You've seen this before, and I've talked about this before, they got an equal position and the play that followed was logical".

This way, you'd naturally get a vast plus score for White in the analyzed games - without any sort of bias (as in "This opening is the greatest, look, 70% win in the game I look at!"), but just because those games tend to be the most important ones to look at (for both sides).


Good points all around by Isa and Eric, IMO
  
Back to top
 
IP Logged
 
ErictheRed
God Member
*****
Offline


USCF National Master

Posts: 2533
Location: USA
Joined: 10/02/05
Re: Sam Collins' bias towards White
Reply #19 - 06/26/18 at 21:41:27
Post Tools
Pessoa wrote on 06/26/18 at 12:27:27:
Looking at the 12 Anti-Sicilian updates by Sam Collins from July 2017 to June 2018, I find:

   1 – 0   :    49  games
  ½ – ½  :    31  games
   0 – 1   :    16  games

  White's score:  67%

  Averages:
  White:    ELO  2669,   performance  2736
  Black:     ELO  2611,   performance  2545

He doesn't get it.
I don't get it.


There has been good discussion and interesting points have been raised in this thread, but I take issue with using raw statistics to say that an author "doesn't get it" or insinuate that he is overly biased.  In a system of openings such as the Anti-Sicilians, where Black has long been considered to be fine, I suspect that most theoretically relevant games will be the ones where White has been able to press. 

Also, the end result of a game is often not very indicative of who won the opening battle. 

Lastly, even if Sam Collins does choose a few more White wins than you presume he should, does that mean that he is hiding Black's best defenses?  Does he overlook them in the games' notes?  If he does not, and if his analysis is essentially unbiased, then I don't understand the fuss. 

Saying that an author "doesn't get it" crosses the line, in my opinion, from civil discussion and criticism to ad hominem.  All because of the end results of games that the author has chosen to illustrate his points, instead of looking at the individual points themselves? 
« Last Edit: 06/26/18 at 23:23:01 by ErictheRed »  
Back to top
 
IP Logged
 
ReneDescartes
God Member
*****
Offline


Qu'est-ce donc que je
suis? Une chose qui pense.

Posts: 1236
Joined: 05/17/10
Gender: Male
Re: Sam Collins' bias towards White
Reply #18 - 06/26/18 at 20:51:28
Post Tools
Maybe those were theoretically significant games.
  
Back to top
 
IP Logged
 
Pessoa
Junior Member
**
Offline


As coisas não têm significação:
têm existência.

Posts: 58
Joined: 06/29/07
Gender: Male
Re: Sam Collins' bias towards White
Reply #17 - 06/26/18 at 12:27:27
Post Tools
Pessoa wrote on 06/21/17 at 09:57:58:
For quite some time now I have been having the feeling that Sam Collins, in his updates for the Anti-Sicilian section, has a rather strong – and hence, in my view, unjustified – bias towards White. [...]
Come on, Mr. Collins, the Anti-Sicilians are not that good for White! Please, redress the balance! 

Looking at the 12 Anti-Sicilian updates by Sam Collins from July 2017 to June 2018, I find:

   1 – 0   :    49  games
  ½ – ½  :    31  games
   0 – 1   :    16  games

  White's score:  67%

  Averages:
  White:    ELO  2669,   performance  2736
  Black:     ELO  2611,   performance  2545

He doesn't get it.
I don't get it.
  
Back to top
 
IP Logged
 
ReneDescartes
God Member
*****
Offline


Qu'est-ce donc que je
suis? Une chose qui pense.

Posts: 1236
Joined: 05/17/10
Gender: Male
Re: Sam Collins' bias towards White
Reply #16 - 06/27/17 at 18:21:13
Post Tools
TopNotch wrote on 06/27/17 at 03:57:05:
Pessoa wrote on 06/21/17 at 09:57:58:
For quite some time now I have been having the feeling that Sam Collins, in his updates for the Anti-Sicilian section, has a rather strong – and hence, in my view, unjustified – bias towards White. His latest update ("Plenty of success for White") urged me to have a closer look[...]

I think there is a slight natural bias, as he is a lifelong proponent of the Anti-Sicilians for White. Mikhalevski when he ran the KID section and and not being a KID player himself was biased towards White, the reverse was true when Joe Gallagher did his stint for the site.

Now we have John Watson, a lifelong Frenchie, and surprise surprise, Black has answers to all his woes.

As a subscriber you have to filter out these biases, figure out the subtext, ignore the spin and add your own critical analysis to help inform your own judgements.

Never take an annotator's word/analysis or conclusions as gospel. There is nearly always some bias:

A couple of personal revelations I'd like to share further illustrate the point:

Smerdon in a number of updates has expressed the opinion that the ending arising after the following moves is slightly better for White: 1.e4 c5 2.c3 Nf6 3.e5 Nd5 4.Nf3 Nc6 5.d4 cxd4 6.cxd4 d6 7.Bc4 dxe5 8.dxe5 Ndb4 9.Qxd8+ Nxd8 10.Na3. I on the other hand think its just dead equal after both the common 10...Bg4 and the less common 10...Be6.

Watson in particular and McDonald to a lesser extent really like the Guimard French for Black against the Tarrasch and have promoted it in many updates over the years. Nevertheless I have never believed the hype, and have always concluded and still do that the Guimard is comfortably better for White and in more than one mainline. For instance: 1.e4 e6 2.d4 d5 3.Nd2 Nc6 4.Ngf3 Nf6 5.e5 Nd7 6.Bd3 f6 7.exf6 Qxf6 8.Nf1! e5 9.Ne3 Nb6 10.dxe5 Nxe5 11.Nxe5 Qxe5 12.0-0 Bd6 13.f4 Qf6 14.a4 a5 15.Ng4 being just one rather problematic variation these days, which though covered by Watson et al, they have still managed somehow to understate its potency.

I'm sure other subscribers can relate and or have similar experiences.

Also, the notion of an objective assessment of a position falls apart if you try to make it  too precise. Mathematically, there are three possible assessments--absolutely won, absolutely lost, or absolutely drawn. The rest, including the whole apparatus of "slightly better," etc., exclusively concerns the flaws of imperfect players. So "easier to play" and "objectively better" blur into each other. And, of course, different imperfect players have different flaws! Authors doing a good job will disagree a little.

So if Watson evaluates a position as equal and a White proponent as not quite equal, then, call it what we want, neither one need be making a gross error. I agree that this kind of  variance (also relative to one'si own assessments) is normal. Critical reading is as important in chess as anywhere else.
« Last Edit: 06/28/17 at 10:45:58 by ReneDescartes »  
Back to top
 
IP Logged
 
bragesjo
God Member
*****
Offline


CCE at ICCF 2021 and CCM
at ICCF 2023

Posts: 1828
Location: Eskilstuna
Joined: 06/30/06
Gender: Male
Re: Sam Collins' bias towards White
Reply #15 - 06/27/17 at 06:56:45
Post Tools
About statistics lets quate Mark Twain "There are lies, damned lies and statistics".
That the numbers are not even close to match each other is clearly a sign of only wanting to show only whites ideas and not new succefully black ideas or games where black eqalices without problems.

About 2 Nc3 could it be that  MegaBase statistics are filtered by ECO codes so games that became Open sicilians are not is those statistics?
There are a few lines in closed sicilian, particually e6 line, where whites most challenging move involves transposing to open sicilian or else black will equalice at the spot?
  
Back to top
WWW  
IP Logged
 
GabrielGale
Senior Member
****
Offline


Who was Thursday?

Posts: 471
Location: Sydney
Joined: 02/28/08
Gender: Male
Re: Sam Collins' bias towards White
Reply #14 - 06/27/17 at 04:55:34
Post Tools
@Topnotch, that is a good reply and very civilised. In the other thread on whether the forum is dying, perhaps this thread is an example of why ChessPubbers are not posting. They get shot down very quickly. there is no engagement, just a standard "biased" "prejudiced" response from a platform.
Now, I better run before I get shot down as well.
  

http://www.toutautre.blogspot.com/
A Year With Nessie ...... aka GM John Shaw's The King's Gambit (http://thekinggambit.blogspot.com.au/)
Back to top
 
IP Logged
 
TopNotch
God Member
*****
Offline


I only look 1 move ahead,
but its always the best

Posts: 2211
Joined: 01/04/03
Gender: Male
Re: Sam Collins' bias towards White
Reply #13 - 06/27/17 at 03:57:05
Post Tools
Pessoa wrote on 06/21/17 at 09:57:58:
For quite some time now I have been having the feeling that Sam Collins, in his updates for the Anti-Sicilian section, has a rather strong – and hence, in my view, unjustified – bias towards White. His latest update ("Plenty of success for White") urged me to have a closer look.

In the archives (Anti-Sicilian section) I found 231 games commented by Collins since 2010. He mainly discusses five systems / move orders. In the respective games, White scored:

1.e4 c5 2.Nf3 d6 3.Bb5+ : 57,5%
1.e4 c5 2.Nf3 Nc6 3.Bb5 : 77,3%
1.e4 c5 2.Nf3 e6 3.c3 : 76,7%
1.e4 c5 2.Nc3 : 83,3%
1.e4 c4 2.c3 : 54,5%

By contrast, in the "top games" of MegaBase 2016, played since 2010, White scored:

1.e4 c5 2.Nf3 d6 3.Bb5+ : 50,2%
1.e4 c5 2.Nf3 Nc6 3.Bb5 : 57,2%
1.e4 c5 2.Nf3 e6 3.c3 : 49,6%
1.e4 c5 2.Nc3 : 47,0%
1.e4 c4 2.c3 : 46,5%

Quite a different picture, isn't it? Of course I know that very often the result of a game has nothing to do with the result of the opening, and in his comments Collins regularly points out ways for Black to equalise. Still, as a Sicilian player, I find it depressing how often the games he chooses for his updates end with 1-0.

Come on, Mr. Collins, the Anti-Sicilians are not that good for White! Please, redress the balance! 


I think there is a slight natural bias, as he is a lifelong proponent of the Anti-Sicilians for White. Mikhalevski when he ran the KID section and and not being a KID player himself was biased towards White, the reverse was true when Joe Gallagher did his stint for the site.

Now we have John Watson, a lifelong Frenchie, and surprise surprise, Black has answers to all his woes.

As a subscriber you have to filter out these biases, figure out the subtext, ignore the spin and add your own critical analysis to help inform your own judgements.

Never take an annotator's word/analysis or conclusions as gospel. There is nearly always some bias:

A couple of personal revelations I'd like to share further illustrate the point:

Smerdon in a number of updates has expressed the opinion that the ending arising after the following moves is slightly better for White: 1.e4 c5 2.c3 Nf6 3.e5 Nd5 4.Nf3 Nc6 5.d4 cxd4 6.cxd4 d6 7.Bc4 dxe5 8.dxe5 Ndb4 9.Qxd8+ Nxd8 10.Na3. I on the other hand think its just dead equal after both the common 10...Bg4 and the less common 10...Be6.

Watson in particular and McDonald to a lesser extent really like the Guimard French for Black against the Tarrasch and have promoted it in many updates over the years. Nevertheless I have never believed the hype, and have always concluded and still do that the Guimard is comfortably better for White and in more than one mainline. For instance: 1.e4 e6 2.d4 d5 3.Nd2 Nc6 4.Ngf3 Nf6 5.e5 Nd7 6.Bd3 f6 7.exf6 Qxf6 8.Nf1! e5 9.Ne3 Nb6 10.dxe5 Nxe5 11.Nxe5 Qxe5 12.0-0 Bd6 13.f4 Qf6 14.a4 a5 15.Ng4 being just one rather problematic variation these days, which though covered by Watson et al, they have still managed somehow to understate its potency.

I'm sure other subscribers can relate and or have similar experiences.


  

The man who tries to do something and fails is infinitely better than he who tries to do nothing and succeeds - Lloyd Jones Smiley
Back to top
 
IP Logged
 
dali
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 45
Joined: 08/18/09
Re: Sam Collins' bias towards White
Reply #12 - 06/26/17 at 19:23:35
Post Tools
all of this time running statistics could have been used studying the open Sicilian!
  
Back to top
 
IP Logged
 
ReneDescartes
God Member
*****
Offline


Qu'est-ce donc que je
suis? Une chose qui pense.

Posts: 1236
Joined: 05/17/10
Gender: Male
Re: Sam Collins' bias towards White
Reply #11 - 06/26/17 at 17:57:24
Post Tools
gillbod wrote on 06/24/17 at 19:42:16:
Instead of judging bias by the outcome of the chosen games, I'd rather judge it by the quality of the analysis.

Also, asking authors to not be sympathetic towards the lines they play is a bit unfair, IMO. It is the fact that authors play and believe in these openings that makes their opinions worthwhile in the first place. E.g. Michalevski's Open Spanish coverage and Scherbakov's Triangle coverage.

Although I fully understand if you'd like more black ideas to be presented if you play the Sicilian as black. But this could be done by asking for it, rather than an accusation of bias.

With this said, I'm a subscriber, but don't play the Sicilian with either colour, so I'm not really qualified to comment on Sam's coverage. I generally skip through the Anti-Sicilian updates superficially.

I could not agree more strongly.

Today it is so easy to cast aspersions on people for thinking as human beings, taking any inequality in statistical results or deviation from some expected statistical norm as indicating "bias" (what a word to use here!).

For example, take the idea that there must be something fishy in the data if 1.e4 c5 2.Nc3 scores under 50% (as it does in my Megabase2017). Many, many factors come into play. One must ask not only who is using the line, but when they are using it. Perhaps in tournament situations where one player is in a mind to avoid a sharp game? Are there a lot of early draws that look like draws by agreement, as with the Queen's Gambit or Exchange Slav? If you follow the lines out until they transpose to an open Sicilian, White retains some advantage in every line-why? Is the difference between 49% and 50% more important than the difference between 53% and 54%, aside from the slight proportional difference? After all, it is not as if the opening loses some game by scoring under 50%! And statistical measures fluctuate randomly; estimating to what degree is a tricky business even for statisticians, so maybe it means literally nothing. In short, it is a misunderstanding to think that the statistics on an early move such as this, and in particular a result under 50%, is a measure of the quality of the line--exactly the kind of misunderstanding that we see in Collins' case, namely taking statistics to be prima facie evidence of a problem, or even to be the problem itself.

Is the result of the game any measure of the opening content? No. Would a book combating Magnus Carlsen's openings be flawed because it contained so many games Carlsen won from equal positions. No!

Do you really need your opening analysis text to be a pep talk?

The way to good content, like the way to good teaching, is to let good practitioners work--work as dedicated human beings, without interference based on superficial measurements or specious criteria. Editorial advice from an experienced human being using his or her whole brain is one thing; running writing through a mindless threshing machine is another.
« Last Edit: 06/26/17 at 22:45:23 by ReneDescartes »  
Back to top
 
IP Logged
 
gillbod
Full Member
***
Offline


I Love ChessPublishing!

Posts: 173
Joined: 03/26/13
Re: Sam Collins' bias towards White
Reply #10 - 06/24/17 at 19:42:16
Post Tools
Instead of judging bias by the outcome of the chosen games, I'd rather judge it by the quality of the analysis.

Also, asking authors to not be sympathetic towards the lines they play is a bit unfair, IMO. It is the fact that authors play and believe in these openings that makes their opinions worthwhile in the first place. E.g. Michalevski's Open Spanish coverage and Scherbakov's Triangle coverage.

Although I fully understand if you'd like more black ideas to be presented if you play the Sicilian as black. But this could be done by asking for it, rather than an accusation of bias.

With this said, I'm a subscriber, but don't play the Sicilian with either colour, so I'm not really qualified to comment on Sam's coverage. I generally skip through the Anti-Sicilian updates superficially.
  
Back to top
 
IP Logged
 
LeeRoth
God Member
*****
Offline


I love ChessPublishing.com!

Posts: 1520
Joined: 10/22/05
Re: Sam Collins' bias towards White
Reply #9 - 06/24/17 at 13:42:59
Post Tools
First of all, a huge thank you to Pessoa for raising this issue and doing all the work on this thread. I think it's fascinating. 

I had the same thought as Isa Vulpes about the selection process.  I do wonder whether the other sections might also have a White bias or whether this is something you expect to see only with the anti-sicilians and other more general lines.  I could see a Black bias occurring in some of the more specific sections.  If I was writing a section on a particular defense, say, the French, the Dragon or the Kings Indian, I would certainly be looking for Black wins to present to the fans of that defense.

As to what all this means to 2.Nc3, it might be worth noting that 2.Nc3 can lead to different systems.  Open, Closed, Grand Prix, 3.Bb5, etc.  So you probably have to be more granular to see what is really going on.  I've seen it said that White generally scores 54%, so if, just as an example, the Closed Sicilian scores 50%, you are giving something away from an opportunity cost perspective.

I should add that, for most of us, I don't think the statistics matter very much, at least within a reasonable range.  Better to play what you like and understand than to try to play a higher-scoring line that is totally beyond you.  The stats are interesting, but they are based on how the pros are doing, and, at the end of the day, it's your individual results with the opening that matter the most to you.
  
Back to top
 
IP Logged
 
Pessoa
Junior Member
**
Offline


As coisas não têm significação:
têm existência.

Posts: 58
Joined: 06/29/07
Gender: Male
Re: Sam Collins' bias towards White
Reply #8 - 06/23/17 at 10:19:40
Post Tools

an ordinary chessplayer wrote on 06/22/17 at 20:33:57:
Sorry if you took offense.

I didn't take offence. You made a valid point, and I was just wondering what "evenly matched" might mean in terms of ELO difference.

I have to agree that MegaBase's "top games" statistics after 1.e4 c5 2.Nc3 is a bit misleading. One point is that, after 2.Nc3, White quite often plays Nf3 and d4 anyway, which normally leads to an Open Sicilian after ...cxd4 Nxd4. But that alone shouldn't explain a white score below 50%. The main reason presumably has to do with the difference in playing strength – perhaps 2.Nc3 is essayed more often than not by weaker players against stronger ones …

To get a clue how well the Anti-Sicilians with 2.Nc3 do if the players are "evenly matched", I did the following analysis.

First, I looked at the games annotated by Sam Collins in the Anti-Sicilans section. Of those, 18 games began with 1.e4 c5 2.Nc3; they are classified under the ECO codes B23, B25 and B30. (As noted above, White scored 83.3% in these games ...)

Then I did a search in MegaBase 2017 with the following parameters:

Position: after 1.e4 c5 2.Nc3
ECO: B23
ELO: both White and Black 2400 – 2500 ("evenly matched", so to speak ...)

The result: 732 games; White scored 55.1%.

I repeated this procedure for three more ELO intervals. The combined results for B23:

B23      
ELO interval      number of games      White's score [%]
2400 – 2500      732                          55.1
2500 – 2600      308                          51.3
2600 – 2700      70                            57.9
2700 – 3000      20                            47.5
total                  1130                        54.1

Obviously with this method one does not find games where, say, White was rated 2590 and Black 2610. But my ChessBase 13 does not appear to provide any easy ways of searching for games where the ratings of the white and black players differ by not more than a specified amount. At least I haven't found any ...

However, I think the numbers above give us a first valid estimate.

The analogous results for B25 and B30:

B25            
ELO interval      number of games      White's score [%]
2400 – 2500      292                          52.1
2500 – 2600      95                            48.9
2600 – 2700      28                            51.8
2700 – 3000      8                              68.8
total                  423                          51.7

B30            
ELO interval      number of games      White's score [%]
2400 – 2500      128                          57.4
2500 – 2600      50                            57.0
2600 – 2700      21                            73.8
2700 – 3000      4                              75.0
total                  203                          59.3

If we combine the results for all three ECO codes (1756 games played by – more or less – evenly matched players), we get an average white score of 54.1%, which must be close to what White should expect to achieve in most "regular" openings.

By the way, Megabase 2017 tells me that, after 1.e4 c5 2.Nc3, in the "top games" White scores 47.3% after 2…Nc6. Go, figure … Huh
  
Back to top
 
IP Logged
 
an ordinary chessplayer
God Member
*****
Offline


I used to be not bad.

Posts: 1664
Location: Columbus, OH (USA)
Joined: 01/02/15
Re: Sam Collins' bias towards White
Reply #7 - 06/22/17 at 20:33:57
Post Tools
Sorry if you took offense. I think the statistics you quoted did support your original point. It's just that since I usually play 2.Nc3, the numbers there jumped out at me.
  
Back to top
 
IP Logged
 
Pessoa
Junior Member
**
Offline


As coisas não têm significação:
têm existência.

Posts: 58
Joined: 06/29/07
Gender: Male
Re: Sam Collins' bias towards White
Reply #6 - 06/22/17 at 14:56:18
Post Tools

an ordinary chessplayer wrote on 06/22/17 at 14:33:46:
If a healthy developing move like 2.Nc3 cannot score 50% for white, there is something wrong with the data. If the players are evenly matched, I would expect to see something like 50,5%.

What is, in your opinion, the biggest difference in playing strength (or ELO, for that matter) that would be acceptable for the players to be called "evenly matched"?

What is the lowest rating for such an analysis? Should games be included where both players are rated around, say, 1200? 1300? 1400? ... ?
  
Back to top
 
IP Logged
 
Page Index Toggle Pages: [1] 2 
Topic Tools
Bookmarks: del.icio.us Digg Facebook Google Google+ Linked in reddit StumbleUpon Twitter Yahoo